SlideShare a Scribd company logo
1 of 107
本著作採用創用CC 「姓名標示」授權條款台灣3.0版



         Intelligent
   Video Surveillance and
       Sousveillance



        Wang, Yuan-Kai(王元凱)
    Electrical Engineering Department,
Fu Jen Univ., Taiwan (輔仁大學電機工程系)
      Email: ykwang@mail.fju.edu.tw
        URL: http://www.ykwang.tw
              2010/11/06
Wang, Yuan-Kai(王元凱)   Video Surveillance and Sousveillance   p.   2



                      Contents
     1.     Image, Vision and Intelligence
     2.     Intelligent Video Surveillance
     3.     Intelligent Video Sousveillance
     4.     Conclusions
Wang, Yuan-Kai(王元凱)   Video Surveillance and Sousveillance   p.   3



                 1. Image, Vision, and
                      Intelligence
         Human's vision system
Wang, Yuan-Kai(王元凱)        Video Surveillance and Sousveillance   p.   4



                       Image, Vision, and
                           Intelligence
                      in the Digital World
         Digital vision system




                                       +
         Robust Computer Vision Algorithms
Wang, Yuan-Kai(王元凱)     Video Surveillance and Sousveillance           p.   5



       Camera Based Environment
              Person : Camera                    =        1    :   N




    Outside-In : Video Surveillance, Human Computer Interface
    Inside-Out : Video Sousveillance, Egocentric Vision
Wang, Yuan-Kai(王元凱)    Video Surveillance and Sousveillance   p.   6



                 2. Video Surveillance

         Video surveillance
           Use video camera to monitor an
            area for crime investigation
         From video surveillance
            to visual surveillance
           Impose video analytics
              by computer vision algorithm
Wang, Yuan-Kai(王元凱)   Video Surveillance and Sousveillance          p.   7



           Video Surveillance Market
       • CCTV has been a mass-product
         market
       • Since the 911 event, the market
         is continuously increasing
(百萬美元)




                                                             Source: JP Freeman
Wang, Yuan-Kai(王元凱)   Video Surveillance and Sousveillance   p.   8



               Who’s Watching You?
         UK has the most CCTV cameras
          in Europe
           4.2 million cameras which is
             20% of the world's CCTV
           1 camera for every 14 people in UK
         On average a person can be
          caught on camera 200-300 times a
          day
Wang, Yuan-Kai(王元凱)              Video Surveillance and Sousveillance                   p.   9



          Crimes Breaking by CCTV
                          92年度             93年度前3季                          統計
         案類/年度
                      件數     人數           件數         人數            件數            人數
            總計        610        689       720        796          1330          1485
            竊盜        364        412       425        447           789          859
            搶奪        91         69        104        104           195          173
            強盜        44         53        39          65           83           118
            殺人        18         36        13          20           31           56
          擄人勒贖        5          15         1          2                6        17
           重傷害        4          8          3          3                7        11
          恐嚇取財        5          6          1          2                6         8
          強制性交        5          5          3          3                8         8
            其他        74         85        131        150           205          235
Wang, Yuan-Kai(王元凱)                  Video Surveillance and Sousveillance          p.   10



          CCTV v.s. Crime Breaking
         監視系統對破獲刑案的助益
               120000                                                       7000
                               監視器數量
               100000                                                       6000
                               因監視器破獲件數
                                                                            5000
                80000
                                                                            4000
                60000
                                                                            3000
                40000
                                                                            2000
                20000                                                       1000

                      0                                                     0
                          92    93         94        95     96       97
                                                年度


         監視器數量和監視器破獲件數兩者間呈現
          正向關係
Wang, Yuan-Kai(王元凱)   Video Surveillance and Sousveillance   p.   11




              Important Crime Cases
         近年來運用路口監視器
          偵破社會矚目重大案件
           白米炸彈客
           汐止市殺警奪槍案
           蠻牛千面人案
           台南國道襲警奪槍案
           新莊襲警奪槍案
           英國倫敦地鐵爆炸案
           台中角頭槍殺案
Wang, Yuan-Kai(王元凱)    Video Surveillance and Sousveillance   p.   12



                      Case Study
         94年5月17日台中市蠻牛千面人案,
          造成全省恐慌
         破案關鍵在於
          幾個放置毒蠻牛
           的超商監視器
           錄到千面人身影
          歹徒車號被
           提款機監視器
           清楚拍下
          動員500警員
           觀看6000小時的錄影資料
Wang, Yuan-Kai(王元凱)           Video Surveillance and Sousveillance               p.   13



           CCTV Video Surveillance
                       Video Display & Record

                                                                     VCR / DVR


                                                               Analog
                      Multiplexer                               components
                                                               Centralized
                                                                Monitoring
       Video Capture

         analogue        analogue          analogue              analogue
Wang, Yuan-Kai(王元凱)               Video Surveillance and Sousveillance   p.   14



           Digital Video Surveillance
                                             High scalibility
                                             IPCam + analog camera
                                             Network transmission
                                             Remote control
                                             Digital storgage

                                                               digital
                                 Network
                                                               digital


                                                               digital



      analogue        analogue                         analogue
                                    analogue
Wang, Yuan-Kai(王元凱)   Video Surveillance and Sousveillance          p.   15


                 Technology Trend of
                  Video Surveillance




                                                   資料來源:拓璞產業研究所,2008年5月
Wang, Yuan-Kai(王元凱)           Video Surveillance and Sousveillance            p.   16



                  Surveillance over IP
        Paradigm shift of video surveillance
          Role from security monitoring to the personalized video contents
          Advent of the intelligent surveillance


             Changes in technology & desire
            1. Network
            2. Video compression
            3. Live images                      Intelligent
                                               Surveillance
                                    IP Surveillance
                              CCTV (DVR)
                      CCTV (VCR)
                         1G
                                            2G                       3G
Wang, Yuan-Kai(王元凱)       Video Surveillance and Sousveillance   p.   17



        Why Intelligent Surveillance




         Too many cameras, too few human guards
         “After only 20 minutes, human attention to
          video monitors degenerates to an
          unacceptable level.” (Sandia National Laboratories)
Wang, Yuan-Kai(王元凱)        Video Surveillance and Sousveillance   p.   18


                        Applications of
                      Visual Surveillance
Wang, Yuan-Kai(王元凱)        Video Surveillance and Sousveillance   p.   19




                      Visual Surveillance
Visual Surveillance = Digital CCTV + Video Analytics
            Smart/Intelligent Surveillance
Wang, Yuan-Kai(王元凱)           Video Surveillance and Sousveillance                   p.     20




                       Video Analytics
影像擷取                  相機異常偵測                    人臉辨識                  查詢、過濾、聯防




  Video       Image     Object         Object          Object        Behavior
 Capture     Enhance    /Event        Tracking
                                                       /Event
                                                                     Analysis   Retrieval
                       Detection                     Recognition




           強光抑制              警戒線、
                                     跌倒、人潮行為分析
                          路徑追蹤、流浪漢監控
Wang, Yuan-Kai(王元凱)          Video Surveillance and Sousveillance            p.   21



                                IBM S3




     Exploratory Computer Vision Group in IBM T.J. Watson Research Center.
                     http://www.research.ibm.com/ecvg/
Wang, Yuan-Kai(王元凱)    Video Surveillance and Sousveillance   p.   22



                      ObjectVideo
Wang, Yuan-Kai(王元凱)   Video Surveillance and Sousveillance   p.   23



                              ITRI
王元凱                 以視覺為基礎之智慧型環境(Vision Based Intelligent Environment)           p. 24



           經濟部學界科專計畫 (1/3)
         MOEA’s Technology Development
          Program for Academia
           Name: Construction of Vision-
            Based Intelligent Environment
           Phase I : 2004/5 ~ 2008/4
           Phase II: 2008/11 ~ 2012/10
           Involved people
             29 professors from 18 universities
             110 staffs

Fu Jen University       Department of Electronic Engineering             Yuan-Kai Wang
Wang, Yuan-Kai(王元凱)       Video Surveillance and Sousveillance   p.   25



           經濟部學界科專計畫 (2/3)
     智慧型建築
         目標:開發智慧型建築內部空間不可或缺的全
          方位、主動式、機動性的智慧性視訊監控系統
                                             A1 日夜活動式廣域安全監視
                                             系統


                                             A2 視訊監控中央管理系統


                                             A3 室內突發事件分析系統
           攝影機網路      感測網路
Wang, Yuan-Kai(王元凱)   Video Surveillance and Sousveillance   p.   26



            經濟部學界科專計畫 (3/3)
     智慧型社區與城市
             目標:開發戶外社區及城市大範圍區域之穩定、成熟而
              具產品面向的智慧性視覺監控系統

                                         B1 人車偵測與辨識系統


                                         B2 都會區人物追蹤系統


                                         B3 室外事件分析與搜尋系統
Wang, Yuan-Kai(王元凱)   Video Surveillance and Sousveillance   p.   27



      Activities in the VBIE Project
         參加國際展覽
         研究技術需
             展示化:技術需能常駐展示
             系統化:大型整合展示
             指標化:技術有量化指標
             市場化:建立產業鏈地圖、政策規劃
             專利化:專利佈局、專利地圖分析
             商品化:網路行銷百餘項技術
         參與國際標準制訂(ONVIF)
         引導業界投資
         與警政機關合作
Wang, Yuan-Kai(王元凱)     Video Surveillance and Sousveillance   p.   28



       A Proposed Architecture for
            Police Office (1/2)




                  智慧型視訊監控技術在警政治安上之可行性研究,
                  詹毓青,中央警察大學資訊管理所碩士論文,2009
Wang, Yuan-Kai(王元凱)     Video Surveillance and Sousveillance   p.   29



       A Proposed Architecture for
            Police Office (2/2)




                  智慧型視訊監控技術在警政治安上之可行性研究,
                  詹毓青,中央警察大學資訊管理所碩士論文,2009
Wang, Yuan-Kai(王元凱)   Video Surveillance and Sousveillance   p.   30



            Demos from ISLab@FJU
         2.1 Mobile Video Surveillance
         2.2 Night Vision
         2.3 Reconfigurable Hardware for
              Moving Object Detection
         2.4 Super-resolution
         2.5 Camera Tampering Detection
         2.6 System Integration
Wang, Yuan-Kai(王元凱)    Video Surveillance and Sousveillance   p.   31




       2.1 Mobil Video Surveillance




              Event-driven Instant Messaging
Wang, Yuan-Kai(王元凱)        Video Surveillance and Sousveillance                    p.   32




        Current Mobile Surveillance
         Mobile phone directly connects to
          the camera

                         Mobile networks
                         (circuit-switched
                        & packet-switched)                        Remote monitor
         Camera                                                       client


                  Drawbacks:
                    1. Not event-driven
                    2. High communication cost
Wang, Yuan-Kai(王元凱)   Video Surveillance and Sousveillance   p.   33


            Instant Alarm Messaging
                  Is Important
Wang, Yuan-Kai(王元凱)           Video Surveillance and Sousveillance             p.   34



            Our System Architecture

                      Keyframe                                   Web
                      Selection                                 Server

                                                                          PC
               Object
                              1. Object Info.                 Video
              Detection
                               2. Keyframe                  Streaming
              &Tracking
                               3. Video Clip                             Mobile
                                                                         Phone

                         Video                                SMS
                      Transcoding                           Messaging
Wang, Yuan-Kai(王元凱)   Video Surveillance and Sousveillance   p.   35


             Our System: Web-based
               Browsing Interface
Wang, Yuan-Kai(王元凱)   Video Surveillance and Sousveillance   p.   36


              Our System –
      Event, Key Frame, Video Clip
Wang, Yuan-Kai(王元凱)      Video Surveillance and Sousveillance     p.   37



                      Mobile Interface
            SMS Message for                         Mobile
           Event Notification                   video streaming
Wang, Yuan-Kai(王元凱)                   Video Surveillance and Sousveillance                                      p.   38



                                       Protocol

                      1.監控                                                    1.監控
     新竹                視訊                                                                         台北
                                                                               視訊
                       分析                                                      分析

                     3.
   Video Analytic   確認                                                                         Video Analytic
         2                                                                         3.                1
                                                                                  確認
                              2.5G
                     5.                                              2.5G           5.   3GP
             3GP    確認                                                             確認
                                            2. SMS    2. SMS

                      3G
                                     4.關鍵
                                      畫面                 4.關鍵                3G
                                                          畫面
                           6.監控
                            視訊                                  6.監控
                            串流                                   視訊
                                                                 串流
                                              輔仁大學
Wang, Yuan-Kai(王元凱)                                 Video Surveillance and Sousveillance           p.   39



                                         Performance
           Instant messaging
             PC: Within 1 seconds
             Mobile phone: ~ 5 seconds
      上午 12:52 - 上午 12:00
          Average time of an event
                  3.9 sec
                                         上午 12:00
                                                           Face detection
                                     Skin detection
                                                              3.4 sec
                                         1.7 sec

                                   下午 11:45 - 上午 12:00
                                       間隔描述



                 上午 11:54 - 下午上午 11:31 - 下午 2:24
                              2:02                                    上午 11:40 - 下午 2:15
                   Detection &      Key frame                                MMS
                     tracking       Selection                               3~5 sec
                     5.7 sec         5.1 sec


                                                                                Mobile Streaming
                                                                                    (RTSP)
                            上午 12:52 - 上午 12:00
                               Transcoding
                                 4.1 sec
Wang, Yuan-Kai(王元凱)      Video Surveillance and Sousveillance   p.   40



                      2.2 Night Vision
         Night vision means to detect
          moving objects from night images
           < 1 lux
           High-level noises
         Night vision should be important
          because crimes
          usually happens
          at night
Wang, Yuan-Kai(王元凱)   Video Surveillance and Sousveillance   p.   41



            Moving Object Detection
         Background subtraction is the
          most important method
         However, it
          can not work
          at night
           Bad image
            quality
           Bad detection
            quality
Wang, Yuan-Kai(王元凱)              Video Surveillance and Sousveillance                                p.   42



           An Example Environment
        Entry of elevators in the hall
        Two challenges
          Low-light-level:
             Low contrast, strong noise
          Drastic light change:
             When elevators open, strong light emits
              from the elevators                                                             緊急出口警示燈 (微弱)
                          現場唯一光源(門外屋簷)      IPCam架設位置




                      1                        2                            3                              4
                                                                        真實場景 by RICOH RX200 (Defaults Setting)
Wang, Yuan-Kai(王元凱)   Video Surveillance and Sousveillance   p.   43
Wang, Yuan-Kai(王元凱)   Video Surveillance and Sousveillance      p.   44


      2.3 Reconfigurable Hardware
      for Moving Object Detection




      Background subtraction, ...
        • 2.8 GHz Intel CPU
        • Software: C/C++                                    FPGA
        • Frame rate: 10 fps for 1 channel
Wang, Yuan-Kai(王元凱)                    Video Surveillance and Sousveillance                      p.   45



             Background Subtraction
                                                   Current
                                                   Frame                  B   k + 1




                                                 Background
 M k +1 ( x, y )           P k+1                Image Update
                                                                      Background Image
 = Pk +1 ( x, y ) − Bk ( x, y ) -
                                                         Bk
                               M
                               k + 1
                                                               Bk +1 ( x, y )
                                                               = αBk ( x, y ) + (1 − α ) Pk +1 ( x, y )
                            Post Processing



                                                    Motion Object Image




        Speed up by (1) Circuit design, (2) Parallelization
Wang, Yuan-Kai(王元凱)   Video Surveillance and Sousveillance   p.   46



         Parallelization by Hardware
         Parallelism: 7-level pipeline
         SIMD with stream processing
Wang, Yuan-Kai(王元凱)      Video Surveillance and Sousveillance            p.   47



                      Design & Results
         Hardware: Altera Cyclone II 2C35
         Design: Verilog HDL with Quartus II




         Background           New Frame                         Result
              Frame rate
                • Background module : 368 fps
                • Whole system : 51 fps
Wang, Yuan-Kai(王元凱)          Video Surveillance and Sousveillance          p.   48



          Experimental Comparison
       PC: 2.8GHz CPU, C implementation
       FPGA can speed up 500 times
                      2.8G
                                                    51



                                                                    CPU
                                                                    FPGA
                             25M          10


                      Clock(Hz)               FPS
Wang, Yuan-Kai(王元凱)   Video Surveillance and Sousveillance               p.    49



                 2.4 Super-resolution
         Object images are too small
         Super-resolution is to
          Construct a high-resolution image
           from low-resolution images
         Challenges: Ill-posed problem



                                       Low-resolution        High-resolution
                                          image                  image
Wang, Yuan-Kai(王元凱)          Video Surveillance and Sousveillance        p.   50



                The Proposed Method
         Pervious methods
           Interpolation, reconstruction
         Feature-based learning
           Dimension Expansion
                      Y                                  X
  Image Space
                                  Dimension            • Maximum A Posteriori
                                  Reduction
                                                         (MAP) Estimator
  Feature Space                              Dimension
                                                       • Gibbs Prior
                                             Expansion
                                x                      • 2D2 Locality
                          High-resolution                Preserving Projection
                             Features
Wang, Yuan-Kai(王元凱)       Video Surveillance and Sousveillance                p.   51



                      Experiments
         Four-time scale
           Low-resolution is 70×50
           High-resolution is 140×100




       Ground Truth   Low-resolution      Bicubic                Our Method
                      NN interpolation interpolation
Wang, Yuan-Kai(王元凱)   Video Surveillance and Sousveillance   p.   52



                 Experimental Results
Wang, Yuan-Kai(王元凱)   Video Surveillance and Sousveillance   p.   53



              2.5 Camera Tampering
                    Detection
                                     Possible tampering
                                      Spray-painting
                                      Replacement
                                      Hit/collision
                                      Defocus
                                      Blocking
                                      ...
Wang, Yuan-Kai(王元凱)   Video Surveillance and Sousveillance           p.   54



          Tampering Examples (1/2)
                      Before                                 After


      Spray
     painting




      Defocus
Wang, Yuan-Kai(王元凱)   Video Surveillance and Sousveillance           p.   55



          Tampering Examples (2/2)
                       Before                                After

Replacement
   (Day)




Replacement
  (Night)
Wang, Yuan-Kai(王元凱)    Video Surveillance and Sousveillance   p.   56



                      The System
         Sabotage
          detection
          before visual
          surveillance
          algorithms
         Server-based
          solution for
          large-scale
          surveillance
          systems
王元凱                 以視覺為基礎之智慧型環境(Vision Based Intelligent Environment)           p. 57



               2.6 System Integration
                 in the VBIE Project
       Smart Building
         Integrated monitoring within building
         Tracking target: person
       Smart Campus
         Long range tracking within campus
         Tracking target: car and person


Fu Jen University       Department of Electronic Engineering             Yuan-Kai Wang
Wang, Yuan-Kai(王元凱)     Video Surveillance and Sousveillance        p.   58



            Heterogeneous Cameras
         We use various kinds of cameras
                             環場攝影機

                                             PTZ攝影機

                                                               紅外線熱像攝影機
     固定式攝影機



                              活動攝影機
                      活動攝影機畫面
Wang, Yuan-Kai(王元凱)        Video Surveillance and Sousveillance   p.   59



                      Smart Building (1/2)
           Characteristics
             11 techniques are integrated by top-
              down design
             It follows the CMMI software
              engineering method
              Documents of spec requirement,
                testing, ...
             A long-term test site is built
Wang, Yuan-Kai(王元凱)        Video Surveillance and Sousveillance           p.   60



                      Smart Building (2/2)




         NTSC一般攝影機                PTZ網路攝影機                        魚眼攝影機
Smart Campus (1/2)
                       13.機器人導引訪客導引
  6.智慧型車牌偵測
                             12.快速人臉辨識
  7.行人交錯偵測與追蹤
                             11.人員連續追蹤
  8.特定人員追蹤與特寫
                   10.軌跡式人員追蹤與檢索

3.循園多攝影機轉場監控           9.多頻道視訊平滑轉場

  2.多攝影機整合式視訊監控

                   4.多重解析度顯示及轉換
                             5.手勢指定物件
 1.串聯式車牌辨識

        14.整合平台及信號傳輸
                        61
Wang, Yuan-Kai(王元凱)    Video Surveillance and Sousveillance   p.   62


                  Smart Campus (2/2)
Wang, Yuan-Kai(王元凱)   Video Surveillance and Sousveillance   p.   63



               3. Video Sousveillance
       Environment cameras
         Outside-in images
       Body-worn cameras
         Inside-out images
Wang, Yuan-Kai(王元凱)     Video Surveillance and Sousveillance   p.   64



                      Paradigm Shift
         Evolution of computing paradigm
           Desktop computing
           Mobile computing
           Wearable computing
         Evolution of camera technology
           Desktop vision
           Mobile vision
           Wearable vision⇒ Egocentric vision
Wang, Yuan-Kai(王元凱)        Video Surveillance and Sousveillance       p.   65



                      Desktop Computing
                                                No interaction with
                                                the environment




         User is focused
         on the system



              stationary
Wang, Yuan-Kai(王元凱)   Video Surveillance and Sousveillance   p.   66



        Jurassic Mobile Computing




         Goal: Stay in touch for mobile usage
Wang, Yuan-Kai(王元凱)     Video Surveillance and Sousveillance   p.   67



         Current Mobile Computing




          Still has drawbacks
               Hands are not free
               Screen is too small
               Need keyboard
               Not proactive
Wang, Yuan-Kai(王元凱)   Video Surveillance and Sousveillance              p.   68



                 Wearable Computing
         Wear the computer on the body




                                                             Smart glasses
  Smart clothing
Wang, Yuan-Kai(王元凱)     Video Surveillance and Sousveillance                         p.   69



                      The Hardware
         Computer: Embedded system
         Input
           No keyboard
            & mouse                                                  Gestures


           But sensors
         Output                             Text          Audio   Special Purpose
Wang, Yuan-Kai(王元凱)    Video Surveillance and Sousveillance   p.   70



                  Input Sensor - Glove
         Hand Gesture Interaction
Wang, Yuan-Kai(王元凱)     Video Surveillance and Sousveillance   p.   71



                      Output Display
         Optical see-through HMD (Head
          Mounted Display)
Wang, Yuan-Kai(王元凱)       Video Surveillance and Sousveillance       p.   72



                      Add Vision Sensor
         Video see-through HMD
                                                       Wearable Vision
Wang, Yuan-Kai(王元凱)   Video Surveillance and Sousveillance   p.   73



          Why Wearable Vision (1/2)
         Capture of casual events
Wang, Yuan-Kai(王元凱)   Video Surveillance and Sousveillance   p.   74



          Why Wearable Vision (2/2)
         Game assistant
Wang, Yuan-Kai(王元凱)             Video Surveillance and Sousveillance           p.   75



        Where to Wear an Extra Eye




              Mayol-Cuevas, Tordoff, Murray, "On the Choice and Placement of
                Wearable Vision Sensors," IEEE Trans. SMC A, March 2009.
Wang, Yuan-Kai(王元凱)   Video Surveillance and Sousveillance   p.   76



        Where to Wear an Extra Eye
Wang, Yuan-Kai(王元凱)            Video Surveillance and Sousveillance              p.   77



           Steve Mann’s WearComp




       S. Mann, "Humanistic Computing: WearComp as a New Framework
       and Application for Intelligent Signal Processing", Proc. of IEEE, vol.
       86, no. 11, 1998.
Wang, Yuan-Kai(王元凱)             Video Surveillance and Sousveillance       p.   78



             US Army's Land Warrior
                                                       Helmet with OLED
                                                        display to show
                                                        map & troop
                                                        locations




 http://en.wikipedia.org/wiki/Land_Warrior
Wang, Yuan-Kai(王元凱)   Video Surveillance and Sousveillance   p.   79




                 Demo of ISLab@FJU

                      The X-Eye
Wang, Yuan-Kai(王元凱)   Video Surveillance and Sousveillance     p.   80



                      The X-Eye
         Live photo management
           with more comfortable display                    Demo
Wang, Yuan-Kai(王元凱)                Video Surveillance and Sousveillance        p.   81



                      X-Eye Components



                                           觸控面板



               顯示器                                    移動電源
                                                     Camera
                                     自製
                       USB
                                     外殼                          USB      筆電
                      連接線
                                                                   Hub
                                                BeagleBoard
                                     微投影機        SD卡 USB-WIFI卡
                                                                          鍵盤

                 USB-RS232
                             讀卡機
                   控制線              滑鼠
Wang, Yuan-Kai(王元凱)   Video Surveillance and Sousveillance   p.   82



           1st Generation Prototype
Wang, Yuan-Kai(王元凱)         Video Surveillance and Sousveillance                       p.   83



                   Embedded Computer
                   USER   RESET


                                                                   OMAP3530 Processor
Peripheral I/O                                                     •600MHz Cortex-A8
                                                                    •NEON+VFPv3
•USB Host                                                           •16KB/16KB L1
                                                                    •256KB L2
•JTAG                                                               •430MHz C64x+ DSP
                                                                    •32K/32K L1
•DVI-D video out                                                    •48K L1D
                                                                    •32K L2
•S-Video out                                                          •Power VR SGX GPU
                                                                   •64K on-chip RAM
•SD/MMC+                                                           POP Memory
                                                                   •256MB LPDDR RAM
•Stereo in/out                                                     •256MB NAND flash

•RS-232 serial1

•Alternate power

•USB 2.0 HS OTG
                                   7.6 cm
  83
Wang, Yuan-Kai(王元凱)     Video Surveillance and Sousveillance                                 p.   84



              Output - Pico Projector
         Very small projector
         Large screen for mobile usage
              Screen: 15”~30”
              Resolution: 480 x 320
              Brightness: 7 lumens
              Contrast ratio: 1000:1




                                          2009~2018年微型投影機市場預估 (單位:百萬台)
                            (來源:DisplaySearch Pocket Projector Technology and Market Forecast Report)
Wang, Yuan-Kai(王元凱)     Video Surveillance and Sousveillance   p.   85



                      Input - Camera
Wang, Yuan-Kai(王元凱)         Video Surveillance and Sousveillance   p.   86



                        User Interface
         Gesture Recognition with
          Bare Hand
          No keyboard & mouse




                      Capture egocentric images
Wang, Yuan-Kai(王元凱)    Video Surveillance and Sousveillance             p.   87



                      Photo Mode




                                      Capture Command: capture images




                                        Switch Command: Mode switch
Wang, Yuan-Kai(王元凱)    Video Surveillance and Sousveillance   p.   88



                      Manage Mode
                            Original Photos




                            Next Command




                           Previous Command




                               Switch Command 2(to photo)
                                                              88
Wang, Yuan-Kai(王元凱)   Video Surveillance and Sousveillance                p.   89



       X-Eye Hardware Architecture




                                                             2010.04.25
  89
Wang, Yuan-Kai(王元凱)   Video Surveillance and Sousveillance   p.   90



       X-Eye Software Architecture
Wang, Yuan-Kai(王元凱)         Video Surveillance and Sousveillance                  p.   91



                X-Eye Software Stack
               Module              Version                          Function
       Gesture Command
                                        1.0             Gesture Recognition
            Module
              OpenCV                    1.0                 Image Processing
              FFMpeg                  0.5.1                Audio/Video Codec
                  QT                  4.6.2                 Windows Interface
              VMWare                  6.5.3                         Virtual OS
           Ubuntu (host)               9.04                  Development OS
          Ubuntu (client)              9.10                        Filesystem
          Kernel (client)            2.6.29                        Linux Kernel
  91
Wang, Yuan-Kai(王元凱)                        Video Surveillance and Sousveillance                                         p.        92



                               Algorithm (1/2)
          Gaussian Mixture Model (GMM)
            Model four colors
                                                                                     1
                                     N
                                                1                                 ( − ( x − mi )T Covi−1 ( x − mi ))
                       p ( x | c) = ∑ ωi                  e                          2

                                    i =1 2π || Covi ||1/2


          Expectation Maximization (EM)
            Parameter estimation of GMM
                     E Step                                                       M Step
                                                                    N
                                                          1
                                                                 ∑ E(z
                                                                                                                  N
                                                                                                        1
                     ω p( x j | m , C )          ωit +1 =                                   t +1
                                                                                        ) m =                   ∑ E(z
                       t         t     t
                                                                                                                                  )x j
  E ( zij ) =                                                                                        Nω     t +1
                       i         i    i                                            ij       i                                ij
                 i                                        N         j =1                                    i    j =1

                ∑ ω tp p( x j | mtp , C tp )
                                           =C         t +1      1
                                                                    t +1 ∑
                                                                           N
                                                                           E ( zij )[( x j − mit +1 )( x j − mit +1 )T ]
                p =1
                                                             N ωi
                                                     i
                                                                           j =1

  92
Wang, Yuan-Kai(王元凱)             Video Surveillance and Sousveillance   p.   93



                        Algorithm (2/2)
          Color Identification
            c
             ˆ         =
                      arg max P( x | c j ), j 1 ~ k
                           cj


            Performance optimization by Look
             Up Table (LUT) for real-time
          Gesture Recognition
            Four gestures: capture, switch,
             next, previous,


  93
Wang, Yuan-Kai(王元凱)      Video Surveillance and Sousveillance        p.   94



                      Smart Camera
         Egocentric vision needs
          small and smart cameras

                                                                FJUCam1
                                                                FJUCam2

          Image          Image                      Image
         Capturing     Processing                 Recognition
Wang, Yuan-Kai(王元凱)        Video Surveillance and Sousveillance              p.   95



                 FJUCam1 - Hardware
 • Weight: 35gm
 • Power sources:                                                 •Size:
   • 5V DC current                                                 6 x 4.5 x 5 (cm)
   • 5V Cell Battery                                               (W x H x D)
 • Power
   consumption:
   1W



                              Three Modules
                      1. Main board, 2. Lens module
                            3. Storage module
Wang, Yuan-Kai(王元凱)    Video Surveillance and Sousveillance   p.   96



                  FJUCam1 - Software
         Development environment
           C Language
           PC Windows + Cygwin + GCC
           cc3 library (open source developed
            by CMU)
Wang, Yuan-Kai(王元凱)       Video Surveillance and Sousveillance   p.   97



         FJUCam1 - Face Detection
         The Adaboost algorithm
           Proposed by Viola and Jones in 2001
           Cascaded weak classifiers(21 cascades)
         Algorithm refinement
           Reduced to 5 cascades
           Fixed-point arithmetic
           Stream processing for only 64KB
            memory utilization



                                                     Image
                        FJUCam                       Display
                      Face Detection
Wang, Yuan-Kai(王元凱)      Video Surveillance and Sousveillance   p.   98



                       FJUCam2 (1/5)
         Cons of FJUCam1
           Low speed, small memory
           Low image resolution
         Next-generation FJUCam
           Adopts multicore technology
            CPU + DSP (MPSoC chip)
                           99/12            Image
    RAM
   128MB                                    Resolution
                        FJUCam2              VGA


                98/2
     64KB FJUCam1                            CIF
                                               Processor
               60MHz       600MHz
                                               Speed
Wang, Yuan-Kai(王元凱)               Video Surveillance and Sousveillance                   p.   99



                          FJUCam2 (2/5)
      TI OMAP3530@ Cortex-A8 600MHz
                    @ DSP C64x+ 412MHz
                    @ PowerVR SGX 530
      Integrated L1 memory for ARM (16kB I-
       Cache, 16kB D-Cache, 256kB L2)
      256MByte low power mobile DDR
      512MByte NAND Flash
      Mini HDMI interface
      Serial port(UART/RS232) x 2
                                                                    PCB dimensions
      USB port (Host)                                                    55mm × 55mm
      USB port (Client)                                            Outside dimensions
      Micro SD card slot                                            (L×W×H)
      USB power           Back side – Expansion                          63mm 59mm×13 mm
      GPIO                socket for extra peripherals
                                                                    Weight : 20 grams
Wang, Yuan-Kai(王元凱)     Video Surveillance and Sousveillance                      p.   100



                      FJUCam2 (3/5)
   Software (on board)
      Angstrom with Linux
       kernel 2.6.32                              Camera Input using            ARM
      VCam Image/Video                           OpenCV+FFMpeg
                                                                                DSP
       processing demo                                Image/Video
         Using DSP and the                            Processing
          integrated vision
          libraries                              GUI Output using QT
      Other supported
       embedded OS:                                                    easy interface &
                                                                    powerful computing
       Android

   We  are compiling “VCam Laboratory Manual”
     which will be made publicly at the end of 2010
Wang, Yuan-Kai(王元凱)      Video Surveillance and Sousveillance   p.   101



                      FJUCam2 (4/5)
           Algorithms going to be
            developed for FJUCam2
                Color tracking
                Gesture recognition
                Face tracking and recognition
                Human/hand/finger detection
                Event detection
                Video summarization
                Distributed vision processing
Wang, Yuan-Kai(王元凱)     Video Surveillance and Sousveillance   p.   102



                      FJUCam2 (5/5)
         Possible applications
              Wearable computing
              Augmented reality
              Robotic vision
              Visual surveillance
              Medical applications for
               patient monitoring
Wang, Yuan-Kai(王元凱)     Video Surveillance and Sousveillance   p.   103



                      4. Conclusions
         Image processing
          and computer vision
          is funny, is cool
          Has been realistic,
           comes into real life
         It is time to study it
          Hardware is small, cheap, and
           wearable
          More robust algorithms
Wang, Yuan-Kai(王元凱)    Video Surveillance and Sousveillance   p.   104



                      How to Study
         Basic courses
              Programming skills
              Engineering Mathematics
              Signal and Systems
              Digital Signal Processing
         Advanced courses
              Digital Image Processing
              Computer Vision
              Pattern Recognition
              Artificial Intelligence
Wang, Yuan-Kai(王元凱)   Video Surveillance and Sousveillance   p.   105



                      Activities
         中華民國影像處理與圖形識別學會(IPPR)
         電腦視覺、圖學暨影像處理研討會(CVGIP)
         IEEE (國際電機電子學會)
           Transactions on
             Image Processing
             Pattern Analysis and Machine
              Intelligence
             Medical Imaging
           Conferences
             ICIP, ICME, ICPR, ...
Wang, Yuan-Kai(王元凱)        Video Surveillance and Sousveillance   p.   106




                        The End



                      Free for Any Questions
Wang, Yuan-Kai(王元凱)    Video Surveillance and Sousveillance   p.




                      本簡報授權聲明
      此簡報內容採用   Creative Commons 「姓名標示 - 非商
       業性台灣 3.0 版」授權條款
      歡迎非商業目的的重製、散布或修改本簡報的內容,但
       請標明: (1)原作者姓名:王元凱; (2)圖標示:
      簡報中所取用的部份圖形創作乃截取自網際網路,僅供
       演講者於自由軟體推廣演講時主張合理使用,請讀者不
       得對其再行取用,除非您本身自忖亦符合主張合理使用
       之情狀,且自負相關法律責任。

More Related Content

Similar to 以視覺為基礎之智慧型環境研究

Beyond the Lens: Ensuring Success with CCTV Camera Installation
Beyond the Lens: Ensuring Success with CCTV Camera InstallationBeyond the Lens: Ensuring Success with CCTV Camera Installation
Beyond the Lens: Ensuring Success with CCTV Camera InstallationDaisy Kaur
 
Panpisco Technologies Intelligent CCTV Solutions
Panpisco Technologies Intelligent CCTV SolutionsPanpisco Technologies Intelligent CCTV Solutions
Panpisco Technologies Intelligent CCTV SolutionsPanpisco Technologies
 
Surveillance camera control system
Surveillance camera control systemSurveillance camera control system
Surveillance camera control systemkvinitha
 
03 outdoor pcam specification sheet
03 outdoor pcam specification sheet03 outdoor pcam specification sheet
03 outdoor pcam specification sheetIlias Varsamis
 
GTC 2016 Taiwan Startups
GTC 2016 Taiwan StartupsGTC 2016 Taiwan Startups
GTC 2016 Taiwan StartupsMindos Cheng
 
Detailed Study of CCTV Cameras
Detailed Study of CCTV CamerasDetailed Study of CCTV Cameras
Detailed Study of CCTV CamerasHans Khanna
 
Detailed study of_cctv_cameras
Detailed study of_cctv_camerasDetailed study of_cctv_cameras
Detailed study of_cctv_camerasYaser Al-Abdali
 
System detects copper thieves
System detects copper thievesSystem detects copper thieves
System detects copper thievesIlias Varsamis
 
Avoiding privacy in cinema using ir camera (1)
Avoiding privacy in cinema using ir camera (1)Avoiding privacy in cinema using ir camera (1)
Avoiding privacy in cinema using ir camera (1)Shanker Rajendiran
 
Dr. rustom kanga latest trends in video analytics for railways
Dr. rustom kanga latest trends in video analytics for railwaysDr. rustom kanga latest trends in video analytics for railways
Dr. rustom kanga latest trends in video analytics for railwaysimadhammoud
 
Do you know ai is making video surveillance systems smarter
Do you know ai is making video surveillance systems smarterDo you know ai is making video surveillance systems smarter
Do you know ai is making video surveillance systems smarterrajpatel787077
 
Mr. Tamer el - Bahey - Leveraging open source intelligence v1.1
Mr. Tamer el - Bahey - Leveraging open source intelligence v1.1Mr. Tamer el - Bahey - Leveraging open source intelligence v1.1
Mr. Tamer el - Bahey - Leveraging open source intelligence v1.1promediakw
 
CIO Survival Guide -Sony Handbook to IP Video Surveillance Jan08
CIO Survival Guide -Sony Handbook to IP Video Surveillance Jan08CIO Survival Guide -Sony Handbook to IP Video Surveillance Jan08
CIO Survival Guide -Sony Handbook to IP Video Surveillance Jan08Magesh Srinivasan
 
Techfinder Electronics Private Limited, Surat, Industrial Security Solutions
Techfinder Electronics Private Limited, Surat, Industrial Security SolutionsTechfinder Electronics Private Limited, Surat, Industrial Security Solutions
Techfinder Electronics Private Limited, Surat, Industrial Security SolutionsIndiaMART InterMESH Limited
 
Implementation of-a-motion-detection-system
Implementation of-a-motion-detection-systemImplementation of-a-motion-detection-system
Implementation of-a-motion-detection-systemCemal Ardil
 
Body Worn-cameras vs CCTV cameras.pdf
Body Worn-cameras vs CCTV cameras.pdfBody Worn-cameras vs CCTV cameras.pdf
Body Worn-cameras vs CCTV cameras.pdfmagePoint
 
02 a pcam dcv specification sheet
02 a pcam dcv specification sheet02 a pcam dcv specification sheet
02 a pcam dcv specification sheetIlias Varsamis
 

Similar to 以視覺為基礎之智慧型環境研究 (18)

Beyond the Lens: Ensuring Success with CCTV Camera Installation
Beyond the Lens: Ensuring Success with CCTV Camera InstallationBeyond the Lens: Ensuring Success with CCTV Camera Installation
Beyond the Lens: Ensuring Success with CCTV Camera Installation
 
Panpisco Technologies Intelligent CCTV Solutions
Panpisco Technologies Intelligent CCTV SolutionsPanpisco Technologies Intelligent CCTV Solutions
Panpisco Technologies Intelligent CCTV Solutions
 
Surveillance camera control system
Surveillance camera control systemSurveillance camera control system
Surveillance camera control system
 
03 outdoor pcam specification sheet
03 outdoor pcam specification sheet03 outdoor pcam specification sheet
03 outdoor pcam specification sheet
 
GTC 2016 Taiwan Startups
GTC 2016 Taiwan StartupsGTC 2016 Taiwan Startups
GTC 2016 Taiwan Startups
 
Detailed Study of CCTV Cameras
Detailed Study of CCTV CamerasDetailed Study of CCTV Cameras
Detailed Study of CCTV Cameras
 
Detailed study of_cctv_cameras
Detailed study of_cctv_camerasDetailed study of_cctv_cameras
Detailed study of_cctv_cameras
 
System detects copper thieves
System detects copper thievesSystem detects copper thieves
System detects copper thieves
 
VIRDI AC 6000 VOIP
VIRDI AC 6000 VOIPVIRDI AC 6000 VOIP
VIRDI AC 6000 VOIP
 
Avoiding privacy in cinema using ir camera (1)
Avoiding privacy in cinema using ir camera (1)Avoiding privacy in cinema using ir camera (1)
Avoiding privacy in cinema using ir camera (1)
 
Dr. rustom kanga latest trends in video analytics for railways
Dr. rustom kanga latest trends in video analytics for railwaysDr. rustom kanga latest trends in video analytics for railways
Dr. rustom kanga latest trends in video analytics for railways
 
Do you know ai is making video surveillance systems smarter
Do you know ai is making video surveillance systems smarterDo you know ai is making video surveillance systems smarter
Do you know ai is making video surveillance systems smarter
 
Mr. Tamer el - Bahey - Leveraging open source intelligence v1.1
Mr. Tamer el - Bahey - Leveraging open source intelligence v1.1Mr. Tamer el - Bahey - Leveraging open source intelligence v1.1
Mr. Tamer el - Bahey - Leveraging open source intelligence v1.1
 
CIO Survival Guide -Sony Handbook to IP Video Surveillance Jan08
CIO Survival Guide -Sony Handbook to IP Video Surveillance Jan08CIO Survival Guide -Sony Handbook to IP Video Surveillance Jan08
CIO Survival Guide -Sony Handbook to IP Video Surveillance Jan08
 
Techfinder Electronics Private Limited, Surat, Industrial Security Solutions
Techfinder Electronics Private Limited, Surat, Industrial Security SolutionsTechfinder Electronics Private Limited, Surat, Industrial Security Solutions
Techfinder Electronics Private Limited, Surat, Industrial Security Solutions
 
Implementation of-a-motion-detection-system
Implementation of-a-motion-detection-systemImplementation of-a-motion-detection-system
Implementation of-a-motion-detection-system
 
Body Worn-cameras vs CCTV cameras.pdf
Body Worn-cameras vs CCTV cameras.pdfBody Worn-cameras vs CCTV cameras.pdf
Body Worn-cameras vs CCTV cameras.pdf
 
02 a pcam dcv specification sheet
02 a pcam dcv specification sheet02 a pcam dcv specification sheet
02 a pcam dcv specification sheet
 

More from IEEE International Conference on Intelligent Information Hiding and Multimedia Signal Processing

More from IEEE International Conference on Intelligent Information Hiding and Multimedia Signal Processing (16)

Computer Vision in the Age of IoT
Computer Vision in the Age of IoTComputer Vision in the Age of IoT
Computer Vision in the Age of IoT
 
2014/07/17 Parallelize computer vision by GPGPU computing
2014/07/17 Parallelize computer vision by GPGPU computing2014/07/17 Parallelize computer vision by GPGPU computing
2014/07/17 Parallelize computer vision by GPGPU computing
 
Towards Embedded Computer Vision - New @ 2013
Towards Embedded Computer Vision - New @ 2013Towards Embedded Computer Vision - New @ 2013
Towards Embedded Computer Vision - New @ 2013
 
老師與教學助理的互動經驗分享 1010217
老師與教學助理的互動經驗分享 1010217老師與教學助理的互動經驗分享 1010217
老師與教學助理的互動經驗分享 1010217
 
Parallel Vision by GPGPU/CUDA
Parallel Vision by GPGPU/CUDAParallel Vision by GPGPU/CUDA
Parallel Vision by GPGPU/CUDA
 
Markov Random Field (MRF)
Markov Random Field (MRF)Markov Random Field (MRF)
Markov Random Field (MRF)
 
07 approximate inference in bn
07 approximate inference in bn07 approximate inference in bn
07 approximate inference in bn
 
06 exact inference in bn
06 exact inference in bn06 exact inference in bn
06 exact inference in bn
 
08 probabilistic inference over time
08 probabilistic inference over time08 probabilistic inference over time
08 probabilistic inference over time
 
05 probabilistic graphical models
05 probabilistic graphical models05 probabilistic graphical models
05 probabilistic graphical models
 
04 Uncertainty inference(continuous)
04 Uncertainty inference(continuous)04 Uncertainty inference(continuous)
04 Uncertainty inference(continuous)
 
03 Uncertainty inference(discrete)
03 Uncertainty inference(discrete)03 Uncertainty inference(discrete)
03 Uncertainty inference(discrete)
 
01 Probability review
01 Probability review01 Probability review
01 Probability review
 
02 Statistics review
02 Statistics review02 Statistics review
02 Statistics review
 
Monocular Human Pose Estimation with Bayesian Networks
Monocular Human Pose Estimation with Bayesian NetworksMonocular Human Pose Estimation with Bayesian Networks
Monocular Human Pose Estimation with Bayesian Networks
 
Towards Embedded Computer Vision邁向嵌入式電腦視覺
Towards Embedded Computer Vision邁向嵌入式電腦視覺Towards Embedded Computer Vision邁向嵌入式電腦視覺
Towards Embedded Computer Vision邁向嵌入式電腦視覺
 

Recently uploaded

Advanced Views - Calendar View in Odoo 17
Advanced Views - Calendar View in Odoo 17Advanced Views - Calendar View in Odoo 17
Advanced Views - Calendar View in Odoo 17Celine George
 
fourth grading exam for kindergarten in writing
fourth grading exam for kindergarten in writingfourth grading exam for kindergarten in writing
fourth grading exam for kindergarten in writingTeacherCyreneCayanan
 
IGNOU MSCCFT and PGDCFT Exam Question Pattern: MCFT003 Counselling and Family...
IGNOU MSCCFT and PGDCFT Exam Question Pattern: MCFT003 Counselling and Family...IGNOU MSCCFT and PGDCFT Exam Question Pattern: MCFT003 Counselling and Family...
IGNOU MSCCFT and PGDCFT Exam Question Pattern: MCFT003 Counselling and Family...PsychoTech Services
 
social pharmacy d-pharm 1st year by Pragati K. Mahajan
social pharmacy d-pharm 1st year by Pragati K. Mahajansocial pharmacy d-pharm 1st year by Pragati K. Mahajan
social pharmacy d-pharm 1st year by Pragati K. Mahajanpragatimahajan3
 
Web & Social Media Analytics Previous Year Question Paper.pdf
Web & Social Media Analytics Previous Year Question Paper.pdfWeb & Social Media Analytics Previous Year Question Paper.pdf
Web & Social Media Analytics Previous Year Question Paper.pdfJayanti Pande
 
SOCIAL AND HISTORICAL CONTEXT - LFTVD.pptx
SOCIAL AND HISTORICAL CONTEXT - LFTVD.pptxSOCIAL AND HISTORICAL CONTEXT - LFTVD.pptx
SOCIAL AND HISTORICAL CONTEXT - LFTVD.pptxiammrhaywood
 
Introduction to Nonprofit Accounting: The Basics
Introduction to Nonprofit Accounting: The BasicsIntroduction to Nonprofit Accounting: The Basics
Introduction to Nonprofit Accounting: The BasicsTechSoup
 
Russian Escort Service in Delhi 11k Hotel Foreigner Russian Call Girls in Delhi
Russian Escort Service in Delhi 11k Hotel Foreigner Russian Call Girls in DelhiRussian Escort Service in Delhi 11k Hotel Foreigner Russian Call Girls in Delhi
Russian Escort Service in Delhi 11k Hotel Foreigner Russian Call Girls in Delhikauryashika82
 
Sanyam Choudhary Chemistry practical.pdf
Sanyam Choudhary Chemistry practical.pdfSanyam Choudhary Chemistry practical.pdf
Sanyam Choudhary Chemistry practical.pdfsanyamsingh5019
 
Activity 01 - Artificial Culture (1).pdf
Activity 01 - Artificial Culture (1).pdfActivity 01 - Artificial Culture (1).pdf
Activity 01 - Artificial Culture (1).pdfciinovamais
 
Arihant handbook biology for class 11 .pdf
Arihant handbook biology for class 11 .pdfArihant handbook biology for class 11 .pdf
Arihant handbook biology for class 11 .pdfchloefrazer622
 
9548086042 for call girls in Indira Nagar with room service
9548086042  for call girls in Indira Nagar  with room service9548086042  for call girls in Indira Nagar  with room service
9548086042 for call girls in Indira Nagar with room servicediscovermytutordmt
 
Call Girls in Dwarka Mor Delhi Contact Us 9654467111
Call Girls in Dwarka Mor Delhi Contact Us 9654467111Call Girls in Dwarka Mor Delhi Contact Us 9654467111
Call Girls in Dwarka Mor Delhi Contact Us 9654467111Sapana Sha
 
BASLIQ CURRENT LOOKBOOK LOOKBOOK(1) (1).pdf
BASLIQ CURRENT LOOKBOOK  LOOKBOOK(1) (1).pdfBASLIQ CURRENT LOOKBOOK  LOOKBOOK(1) (1).pdf
BASLIQ CURRENT LOOKBOOK LOOKBOOK(1) (1).pdfSoniaTolstoy
 
Measures of Central Tendency: Mean, Median and Mode
Measures of Central Tendency: Mean, Median and ModeMeasures of Central Tendency: Mean, Median and Mode
Measures of Central Tendency: Mean, Median and ModeThiyagu K
 
A Critique of the Proposed National Education Policy Reform
A Critique of the Proposed National Education Policy ReformA Critique of the Proposed National Education Policy Reform
A Critique of the Proposed National Education Policy ReformChameera Dedduwage
 
Ecosystem Interactions Class Discussion Presentation in Blue Green Lined Styl...
Ecosystem Interactions Class Discussion Presentation in Blue Green Lined Styl...Ecosystem Interactions Class Discussion Presentation in Blue Green Lined Styl...
Ecosystem Interactions Class Discussion Presentation in Blue Green Lined Styl...fonyou31
 
The Most Excellent Way | 1 Corinthians 13
The Most Excellent Way | 1 Corinthians 13The Most Excellent Way | 1 Corinthians 13
The Most Excellent Way | 1 Corinthians 13Steve Thomason
 
Disha NEET Physics Guide for classes 11 and 12.pdf
Disha NEET Physics Guide for classes 11 and 12.pdfDisha NEET Physics Guide for classes 11 and 12.pdf
Disha NEET Physics Guide for classes 11 and 12.pdfchloefrazer622
 

Recently uploaded (20)

Advanced Views - Calendar View in Odoo 17
Advanced Views - Calendar View in Odoo 17Advanced Views - Calendar View in Odoo 17
Advanced Views - Calendar View in Odoo 17
 
fourth grading exam for kindergarten in writing
fourth grading exam for kindergarten in writingfourth grading exam for kindergarten in writing
fourth grading exam for kindergarten in writing
 
IGNOU MSCCFT and PGDCFT Exam Question Pattern: MCFT003 Counselling and Family...
IGNOU MSCCFT and PGDCFT Exam Question Pattern: MCFT003 Counselling and Family...IGNOU MSCCFT and PGDCFT Exam Question Pattern: MCFT003 Counselling and Family...
IGNOU MSCCFT and PGDCFT Exam Question Pattern: MCFT003 Counselling and Family...
 
social pharmacy d-pharm 1st year by Pragati K. Mahajan
social pharmacy d-pharm 1st year by Pragati K. Mahajansocial pharmacy d-pharm 1st year by Pragati K. Mahajan
social pharmacy d-pharm 1st year by Pragati K. Mahajan
 
Web & Social Media Analytics Previous Year Question Paper.pdf
Web & Social Media Analytics Previous Year Question Paper.pdfWeb & Social Media Analytics Previous Year Question Paper.pdf
Web & Social Media Analytics Previous Year Question Paper.pdf
 
SOCIAL AND HISTORICAL CONTEXT - LFTVD.pptx
SOCIAL AND HISTORICAL CONTEXT - LFTVD.pptxSOCIAL AND HISTORICAL CONTEXT - LFTVD.pptx
SOCIAL AND HISTORICAL CONTEXT - LFTVD.pptx
 
Introduction to Nonprofit Accounting: The Basics
Introduction to Nonprofit Accounting: The BasicsIntroduction to Nonprofit Accounting: The Basics
Introduction to Nonprofit Accounting: The Basics
 
Russian Escort Service in Delhi 11k Hotel Foreigner Russian Call Girls in Delhi
Russian Escort Service in Delhi 11k Hotel Foreigner Russian Call Girls in DelhiRussian Escort Service in Delhi 11k Hotel Foreigner Russian Call Girls in Delhi
Russian Escort Service in Delhi 11k Hotel Foreigner Russian Call Girls in Delhi
 
INDIA QUIZ 2024 RLAC DELHI UNIVERSITY.pptx
INDIA QUIZ 2024 RLAC DELHI UNIVERSITY.pptxINDIA QUIZ 2024 RLAC DELHI UNIVERSITY.pptx
INDIA QUIZ 2024 RLAC DELHI UNIVERSITY.pptx
 
Sanyam Choudhary Chemistry practical.pdf
Sanyam Choudhary Chemistry practical.pdfSanyam Choudhary Chemistry practical.pdf
Sanyam Choudhary Chemistry practical.pdf
 
Activity 01 - Artificial Culture (1).pdf
Activity 01 - Artificial Culture (1).pdfActivity 01 - Artificial Culture (1).pdf
Activity 01 - Artificial Culture (1).pdf
 
Arihant handbook biology for class 11 .pdf
Arihant handbook biology for class 11 .pdfArihant handbook biology for class 11 .pdf
Arihant handbook biology for class 11 .pdf
 
9548086042 for call girls in Indira Nagar with room service
9548086042  for call girls in Indira Nagar  with room service9548086042  for call girls in Indira Nagar  with room service
9548086042 for call girls in Indira Nagar with room service
 
Call Girls in Dwarka Mor Delhi Contact Us 9654467111
Call Girls in Dwarka Mor Delhi Contact Us 9654467111Call Girls in Dwarka Mor Delhi Contact Us 9654467111
Call Girls in Dwarka Mor Delhi Contact Us 9654467111
 
BASLIQ CURRENT LOOKBOOK LOOKBOOK(1) (1).pdf
BASLIQ CURRENT LOOKBOOK  LOOKBOOK(1) (1).pdfBASLIQ CURRENT LOOKBOOK  LOOKBOOK(1) (1).pdf
BASLIQ CURRENT LOOKBOOK LOOKBOOK(1) (1).pdf
 
Measures of Central Tendency: Mean, Median and Mode
Measures of Central Tendency: Mean, Median and ModeMeasures of Central Tendency: Mean, Median and Mode
Measures of Central Tendency: Mean, Median and Mode
 
A Critique of the Proposed National Education Policy Reform
A Critique of the Proposed National Education Policy ReformA Critique of the Proposed National Education Policy Reform
A Critique of the Proposed National Education Policy Reform
 
Ecosystem Interactions Class Discussion Presentation in Blue Green Lined Styl...
Ecosystem Interactions Class Discussion Presentation in Blue Green Lined Styl...Ecosystem Interactions Class Discussion Presentation in Blue Green Lined Styl...
Ecosystem Interactions Class Discussion Presentation in Blue Green Lined Styl...
 
The Most Excellent Way | 1 Corinthians 13
The Most Excellent Way | 1 Corinthians 13The Most Excellent Way | 1 Corinthians 13
The Most Excellent Way | 1 Corinthians 13
 
Disha NEET Physics Guide for classes 11 and 12.pdf
Disha NEET Physics Guide for classes 11 and 12.pdfDisha NEET Physics Guide for classes 11 and 12.pdf
Disha NEET Physics Guide for classes 11 and 12.pdf
 

以視覺為基礎之智慧型環境研究

  • 1. 本著作採用創用CC 「姓名標示」授權條款台灣3.0版 Intelligent Video Surveillance and Sousveillance Wang, Yuan-Kai(王元凱) Electrical Engineering Department, Fu Jen Univ., Taiwan (輔仁大學電機工程系) Email: ykwang@mail.fju.edu.tw URL: http://www.ykwang.tw 2010/11/06
  • 2. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 2 Contents 1. Image, Vision and Intelligence 2. Intelligent Video Surveillance 3. Intelligent Video Sousveillance 4. Conclusions
  • 3. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 3 1. Image, Vision, and Intelligence  Human's vision system
  • 4. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 4 Image, Vision, and Intelligence in the Digital World  Digital vision system + Robust Computer Vision Algorithms
  • 5. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 5 Camera Based Environment Person : Camera = 1 : N Outside-In : Video Surveillance, Human Computer Interface Inside-Out : Video Sousveillance, Egocentric Vision
  • 6. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 6 2. Video Surveillance  Video surveillance  Use video camera to monitor an area for crime investigation  From video surveillance to visual surveillance  Impose video analytics by computer vision algorithm
  • 7. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 7 Video Surveillance Market • CCTV has been a mass-product market • Since the 911 event, the market is continuously increasing (百萬美元) Source: JP Freeman
  • 8. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 8 Who’s Watching You?  UK has the most CCTV cameras in Europe  4.2 million cameras which is  20% of the world's CCTV  1 camera for every 14 people in UK  On average a person can be caught on camera 200-300 times a day
  • 9. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 9 Crimes Breaking by CCTV 92年度 93年度前3季 統計 案類/年度 件數 人數 件數 人數 件數 人數 總計 610 689 720 796 1330 1485 竊盜 364 412 425 447 789 859 搶奪 91 69 104 104 195 173 強盜 44 53 39 65 83 118 殺人 18 36 13 20 31 56 擄人勒贖 5 15 1 2 6 17 重傷害 4 8 3 3 7 11 恐嚇取財 5 6 1 2 6 8 強制性交 5 5 3 3 8 8 其他 74 85 131 150 205 235
  • 10. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 10 CCTV v.s. Crime Breaking  監視系統對破獲刑案的助益 120000 7000 監視器數量 100000 6000 因監視器破獲件數 5000 80000 4000 60000 3000 40000 2000 20000 1000 0 0 92 93 94 95 96 97 年度  監視器數量和監視器破獲件數兩者間呈現 正向關係
  • 11. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 11 Important Crime Cases  近年來運用路口監視器 偵破社會矚目重大案件  白米炸彈客  汐止市殺警奪槍案  蠻牛千面人案  台南國道襲警奪槍案  新莊襲警奪槍案  英國倫敦地鐵爆炸案  台中角頭槍殺案
  • 12. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 12 Case Study  94年5月17日台中市蠻牛千面人案, 造成全省恐慌  破案關鍵在於  幾個放置毒蠻牛 的超商監視器 錄到千面人身影  歹徒車號被 提款機監視器 清楚拍下  動員500警員 觀看6000小時的錄影資料
  • 13. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 13 CCTV Video Surveillance Video Display & Record VCR / DVR Analog Multiplexer components Centralized Monitoring Video Capture analogue analogue analogue analogue
  • 14. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 14 Digital Video Surveillance High scalibility IPCam + analog camera Network transmission Remote control Digital storgage digital Network digital digital analogue analogue analogue analogue
  • 15. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 15 Technology Trend of Video Surveillance 資料來源:拓璞產業研究所,2008年5月
  • 16. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 16 Surveillance over IP Paradigm shift of video surveillance  Role from security monitoring to the personalized video contents  Advent of the intelligent surveillance Changes in technology & desire 1. Network 2. Video compression 3. Live images Intelligent Surveillance IP Surveillance CCTV (DVR) CCTV (VCR) 1G 2G 3G
  • 17. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 17 Why Intelligent Surveillance  Too many cameras, too few human guards  “After only 20 minutes, human attention to video monitors degenerates to an unacceptable level.” (Sandia National Laboratories)
  • 18. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 18 Applications of Visual Surveillance
  • 19. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 19 Visual Surveillance Visual Surveillance = Digital CCTV + Video Analytics Smart/Intelligent Surveillance
  • 20. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 20 Video Analytics 影像擷取 相機異常偵測 人臉辨識 查詢、過濾、聯防 Video Image Object Object Object Behavior Capture Enhance /Event Tracking /Event Analysis Retrieval Detection Recognition 強光抑制 警戒線、 跌倒、人潮行為分析 路徑追蹤、流浪漢監控
  • 21. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 21 IBM S3 Exploratory Computer Vision Group in IBM T.J. Watson Research Center. http://www.research.ibm.com/ecvg/
  • 22. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 22 ObjectVideo
  • 23. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 23 ITRI
  • 24. 王元凱 以視覺為基礎之智慧型環境(Vision Based Intelligent Environment) p. 24 經濟部學界科專計畫 (1/3)  MOEA’s Technology Development Program for Academia  Name: Construction of Vision- Based Intelligent Environment  Phase I : 2004/5 ~ 2008/4  Phase II: 2008/11 ~ 2012/10  Involved people  29 professors from 18 universities  110 staffs Fu Jen University Department of Electronic Engineering Yuan-Kai Wang
  • 25. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 25 經濟部學界科專計畫 (2/3)  智慧型建築  目標:開發智慧型建築內部空間不可或缺的全 方位、主動式、機動性的智慧性視訊監控系統 A1 日夜活動式廣域安全監視 系統 A2 視訊監控中央管理系統 A3 室內突發事件分析系統  攝影機網路  感測網路
  • 26. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 26 經濟部學界科專計畫 (3/3)  智慧型社區與城市  目標:開發戶外社區及城市大範圍區域之穩定、成熟而 具產品面向的智慧性視覺監控系統 B1 人車偵測與辨識系統 B2 都會區人物追蹤系統 B3 室外事件分析與搜尋系統
  • 27. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 27 Activities in the VBIE Project  參加國際展覽  研究技術需  展示化:技術需能常駐展示  系統化:大型整合展示  指標化:技術有量化指標  市場化:建立產業鏈地圖、政策規劃  專利化:專利佈局、專利地圖分析  商品化:網路行銷百餘項技術  參與國際標準制訂(ONVIF)  引導業界投資  與警政機關合作
  • 28. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 28 A Proposed Architecture for Police Office (1/2) 智慧型視訊監控技術在警政治安上之可行性研究, 詹毓青,中央警察大學資訊管理所碩士論文,2009
  • 29. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 29 A Proposed Architecture for Police Office (2/2) 智慧型視訊監控技術在警政治安上之可行性研究, 詹毓青,中央警察大學資訊管理所碩士論文,2009
  • 30. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 30 Demos from ISLab@FJU  2.1 Mobile Video Surveillance  2.2 Night Vision  2.3 Reconfigurable Hardware for Moving Object Detection  2.4 Super-resolution  2.5 Camera Tampering Detection  2.6 System Integration
  • 31. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 31 2.1 Mobil Video Surveillance Event-driven Instant Messaging
  • 32. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 32 Current Mobile Surveillance  Mobile phone directly connects to the camera Mobile networks (circuit-switched & packet-switched) Remote monitor Camera client Drawbacks: 1. Not event-driven 2. High communication cost
  • 33. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 33 Instant Alarm Messaging Is Important
  • 34. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 34 Our System Architecture Keyframe Web Selection Server PC Object 1. Object Info. Video Detection 2. Keyframe Streaming &Tracking 3. Video Clip Mobile Phone Video SMS Transcoding Messaging
  • 35. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 35 Our System: Web-based Browsing Interface
  • 36. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 36 Our System – Event, Key Frame, Video Clip
  • 37. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 37 Mobile Interface SMS Message for Mobile Event Notification video streaming
  • 38. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 38 Protocol 1.監控 1.監控 新竹 視訊 台北 視訊 分析 分析 3. Video Analytic 確認 Video Analytic 2 3. 1 確認 2.5G 5. 2.5G 5. 3GP 3GP 確認 確認 2. SMS 2. SMS 3G 4.關鍵 畫面 4.關鍵 3G 畫面 6.監控 視訊 6.監控 串流 視訊 串流 輔仁大學
  • 39. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 39 Performance  Instant messaging  PC: Within 1 seconds  Mobile phone: ~ 5 seconds 上午 12:52 - 上午 12:00 Average time of an event 3.9 sec 上午 12:00 Face detection Skin detection 3.4 sec 1.7 sec 下午 11:45 - 上午 12:00 間隔描述 上午 11:54 - 下午上午 11:31 - 下午 2:24 2:02 上午 11:40 - 下午 2:15 Detection & Key frame MMS tracking Selection 3~5 sec 5.7 sec 5.1 sec Mobile Streaming (RTSP) 上午 12:52 - 上午 12:00 Transcoding 4.1 sec
  • 40. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 40 2.2 Night Vision  Night vision means to detect moving objects from night images  < 1 lux  High-level noises  Night vision should be important because crimes usually happens at night
  • 41. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 41 Moving Object Detection  Background subtraction is the most important method  However, it can not work at night  Bad image quality  Bad detection quality
  • 42. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 42 An Example Environment  Entry of elevators in the hall  Two challenges  Low-light-level: Low contrast, strong noise  Drastic light change: When elevators open, strong light emits from the elevators 緊急出口警示燈 (微弱) 現場唯一光源(門外屋簷) IPCam架設位置 1 2 3 4 真實場景 by RICOH RX200 (Defaults Setting)
  • 43. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 43
  • 44. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 44 2.3 Reconfigurable Hardware for Moving Object Detection Background subtraction, ... • 2.8 GHz Intel CPU • Software: C/C++ FPGA • Frame rate: 10 fps for 1 channel
  • 45. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 45 Background Subtraction Current Frame B k + 1 Background M k +1 ( x, y ) P k+1 Image Update Background Image = Pk +1 ( x, y ) − Bk ( x, y ) - Bk M k + 1 Bk +1 ( x, y ) = αBk ( x, y ) + (1 − α ) Pk +1 ( x, y ) Post Processing Motion Object Image Speed up by (1) Circuit design, (2) Parallelization
  • 46. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 46 Parallelization by Hardware  Parallelism: 7-level pipeline  SIMD with stream processing
  • 47. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 47 Design & Results  Hardware: Altera Cyclone II 2C35  Design: Verilog HDL with Quartus II Background New Frame Result Frame rate • Background module : 368 fps • Whole system : 51 fps
  • 48. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 48 Experimental Comparison  PC: 2.8GHz CPU, C implementation  FPGA can speed up 500 times 2.8G 51 CPU FPGA 25M 10 Clock(Hz) FPS
  • 49. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 49 2.4 Super-resolution  Object images are too small  Super-resolution is to  Construct a high-resolution image from low-resolution images  Challenges: Ill-posed problem Low-resolution High-resolution image image
  • 50. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 50 The Proposed Method  Pervious methods  Interpolation, reconstruction  Feature-based learning  Dimension Expansion Y X Image Space Dimension • Maximum A Posteriori Reduction (MAP) Estimator Feature Space Dimension • Gibbs Prior Expansion x • 2D2 Locality High-resolution Preserving Projection Features
  • 51. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 51 Experiments  Four-time scale  Low-resolution is 70×50  High-resolution is 140×100 Ground Truth Low-resolution Bicubic Our Method NN interpolation interpolation
  • 52. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 52 Experimental Results
  • 53. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 53 2.5 Camera Tampering Detection  Possible tampering  Spray-painting  Replacement  Hit/collision  Defocus  Blocking  ...
  • 54. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 54 Tampering Examples (1/2) Before After Spray painting Defocus
  • 55. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 55 Tampering Examples (2/2) Before After Replacement (Day) Replacement (Night)
  • 56. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 56 The System  Sabotage detection before visual surveillance algorithms  Server-based solution for large-scale surveillance systems
  • 57. 王元凱 以視覺為基礎之智慧型環境(Vision Based Intelligent Environment) p. 57 2.6 System Integration in the VBIE Project  Smart Building  Integrated monitoring within building  Tracking target: person  Smart Campus  Long range tracking within campus  Tracking target: car and person Fu Jen University Department of Electronic Engineering Yuan-Kai Wang
  • 58. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 58 Heterogeneous Cameras  We use various kinds of cameras 環場攝影機 PTZ攝影機 紅外線熱像攝影機 固定式攝影機 活動攝影機 活動攝影機畫面
  • 59. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 59 Smart Building (1/2)  Characteristics  11 techniques are integrated by top- down design  It follows the CMMI software engineering method Documents of spec requirement, testing, ...  A long-term test site is built
  • 60. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 60 Smart Building (2/2) NTSC一般攝影機 PTZ網路攝影機 魚眼攝影機
  • 61. Smart Campus (1/2) 13.機器人導引訪客導引 6.智慧型車牌偵測 12.快速人臉辨識 7.行人交錯偵測與追蹤 11.人員連續追蹤 8.特定人員追蹤與特寫 10.軌跡式人員追蹤與檢索 3.循園多攝影機轉場監控 9.多頻道視訊平滑轉場 2.多攝影機整合式視訊監控 4.多重解析度顯示及轉換 5.手勢指定物件 1.串聯式車牌辨識 14.整合平台及信號傳輸 61
  • 62. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 62 Smart Campus (2/2)
  • 63. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 63 3. Video Sousveillance  Environment cameras  Outside-in images  Body-worn cameras  Inside-out images
  • 64. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 64 Paradigm Shift  Evolution of computing paradigm  Desktop computing  Mobile computing  Wearable computing  Evolution of camera technology  Desktop vision  Mobile vision  Wearable vision⇒ Egocentric vision
  • 65. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 65 Desktop Computing No interaction with the environment User is focused on the system stationary
  • 66. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 66 Jurassic Mobile Computing Goal: Stay in touch for mobile usage
  • 67. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 67 Current Mobile Computing  Still has drawbacks  Hands are not free  Screen is too small  Need keyboard  Not proactive
  • 68. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 68 Wearable Computing  Wear the computer on the body Smart glasses Smart clothing
  • 69. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 69 The Hardware  Computer: Embedded system  Input  No keyboard & mouse Gestures  But sensors  Output Text Audio Special Purpose
  • 70. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 70 Input Sensor - Glove  Hand Gesture Interaction
  • 71. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 71 Output Display  Optical see-through HMD (Head Mounted Display)
  • 72. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 72 Add Vision Sensor  Video see-through HMD Wearable Vision
  • 73. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 73 Why Wearable Vision (1/2)  Capture of casual events
  • 74. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 74 Why Wearable Vision (2/2)  Game assistant
  • 75. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 75 Where to Wear an Extra Eye Mayol-Cuevas, Tordoff, Murray, "On the Choice and Placement of Wearable Vision Sensors," IEEE Trans. SMC A, March 2009.
  • 76. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 76 Where to Wear an Extra Eye
  • 77. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 77 Steve Mann’s WearComp S. Mann, "Humanistic Computing: WearComp as a New Framework and Application for Intelligent Signal Processing", Proc. of IEEE, vol. 86, no. 11, 1998.
  • 78. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 78 US Army's Land Warrior  Helmet with OLED display to show map & troop locations http://en.wikipedia.org/wiki/Land_Warrior
  • 79. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 79 Demo of ISLab@FJU The X-Eye
  • 80. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 80 The X-Eye  Live photo management  with more comfortable display Demo
  • 81. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 81 X-Eye Components 觸控面板 顯示器 移動電源 Camera 自製 USB 外殼 USB 筆電 連接線 Hub BeagleBoard 微投影機 SD卡 USB-WIFI卡 鍵盤 USB-RS232 讀卡機 控制線 滑鼠
  • 82. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 82 1st Generation Prototype
  • 83. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 83 Embedded Computer USER RESET OMAP3530 Processor Peripheral I/O •600MHz Cortex-A8 •NEON+VFPv3 •USB Host •16KB/16KB L1 •256KB L2 •JTAG •430MHz C64x+ DSP •32K/32K L1 •DVI-D video out •48K L1D •32K L2 •S-Video out •Power VR SGX GPU •64K on-chip RAM •SD/MMC+ POP Memory •256MB LPDDR RAM •Stereo in/out •256MB NAND flash •RS-232 serial1 •Alternate power •USB 2.0 HS OTG 7.6 cm 83
  • 84. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 84 Output - Pico Projector  Very small projector  Large screen for mobile usage  Screen: 15”~30”  Resolution: 480 x 320  Brightness: 7 lumens  Contrast ratio: 1000:1 2009~2018年微型投影機市場預估 (單位:百萬台) (來源:DisplaySearch Pocket Projector Technology and Market Forecast Report)
  • 85. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 85 Input - Camera
  • 86. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 86 User Interface  Gesture Recognition with Bare Hand  No keyboard & mouse Capture egocentric images
  • 87. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 87 Photo Mode Capture Command: capture images Switch Command: Mode switch
  • 88. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 88 Manage Mode Original Photos Next Command Previous Command Switch Command 2(to photo) 88
  • 89. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 89 X-Eye Hardware Architecture 2010.04.25 89
  • 90. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 90 X-Eye Software Architecture
  • 91. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 91 X-Eye Software Stack Module Version Function Gesture Command 1.0 Gesture Recognition Module OpenCV 1.0 Image Processing FFMpeg 0.5.1 Audio/Video Codec QT 4.6.2 Windows Interface VMWare 6.5.3 Virtual OS Ubuntu (host) 9.04 Development OS Ubuntu (client) 9.10 Filesystem Kernel (client) 2.6.29 Linux Kernel 91
  • 92. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 92 Algorithm (1/2)  Gaussian Mixture Model (GMM)  Model four colors 1 N 1 ( − ( x − mi )T Covi−1 ( x − mi )) p ( x | c) = ∑ ωi e 2 i =1 2π || Covi ||1/2  Expectation Maximization (EM)  Parameter estimation of GMM E Step M Step N 1 ∑ E(z N 1 ω p( x j | m , C ) ωit +1 = t +1 ) m = ∑ E(z t t t )x j E ( zij ) = Nω t +1 i i i ij i ij i N j =1 i j =1 ∑ ω tp p( x j | mtp , C tp ) =C t +1 1 t +1 ∑ N E ( zij )[( x j − mit +1 )( x j − mit +1 )T ] p =1 N ωi i j =1 92
  • 93. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 93 Algorithm (2/2)  Color Identification  c ˆ = arg max P( x | c j ), j 1 ~ k cj  Performance optimization by Look Up Table (LUT) for real-time  Gesture Recognition  Four gestures: capture, switch, next, previous, 93
  • 94. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 94 Smart Camera  Egocentric vision needs small and smart cameras FJUCam1 FJUCam2 Image Image Image Capturing Processing Recognition
  • 95. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 95 FJUCam1 - Hardware • Weight: 35gm • Power sources: •Size: • 5V DC current 6 x 4.5 x 5 (cm) • 5V Cell Battery (W x H x D) • Power consumption: 1W Three Modules 1. Main board, 2. Lens module 3. Storage module
  • 96. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 96 FJUCam1 - Software  Development environment  C Language  PC Windows + Cygwin + GCC  cc3 library (open source developed by CMU)
  • 97. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 97 FJUCam1 - Face Detection  The Adaboost algorithm  Proposed by Viola and Jones in 2001  Cascaded weak classifiers(21 cascades)  Algorithm refinement  Reduced to 5 cascades  Fixed-point arithmetic  Stream processing for only 64KB memory utilization Image FJUCam Display Face Detection
  • 98. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 98 FJUCam2 (1/5)  Cons of FJUCam1  Low speed, small memory  Low image resolution  Next-generation FJUCam  Adopts multicore technology CPU + DSP (MPSoC chip) 99/12 Image RAM 128MB Resolution FJUCam2 VGA 98/2 64KB FJUCam1 CIF Processor 60MHz 600MHz Speed
  • 99. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 99 FJUCam2 (2/5)  TI OMAP3530@ Cortex-A8 600MHz @ DSP C64x+ 412MHz @ PowerVR SGX 530  Integrated L1 memory for ARM (16kB I- Cache, 16kB D-Cache, 256kB L2)  256MByte low power mobile DDR  512MByte NAND Flash  Mini HDMI interface  Serial port(UART/RS232) x 2  PCB dimensions  USB port (Host)  55mm × 55mm  USB port (Client)  Outside dimensions  Micro SD card slot (L×W×H)  USB power Back side – Expansion  63mm 59mm×13 mm  GPIO socket for extra peripherals  Weight : 20 grams
  • 100. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 100 FJUCam2 (3/5) Software (on board)  Angstrom with Linux kernel 2.6.32 Camera Input using ARM  VCam Image/Video OpenCV+FFMpeg DSP processing demo Image/Video  Using DSP and the Processing integrated vision libraries GUI Output using QT  Other supported embedded OS: easy interface & powerful computing Android  We are compiling “VCam Laboratory Manual” which will be made publicly at the end of 2010
  • 101. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 101 FJUCam2 (4/5)  Algorithms going to be developed for FJUCam2  Color tracking  Gesture recognition  Face tracking and recognition  Human/hand/finger detection  Event detection  Video summarization  Distributed vision processing
  • 102. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 102 FJUCam2 (5/5)  Possible applications  Wearable computing  Augmented reality  Robotic vision  Visual surveillance  Medical applications for patient monitoring
  • 103. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 103 4. Conclusions  Image processing and computer vision  is funny, is cool  Has been realistic, comes into real life  It is time to study it  Hardware is small, cheap, and wearable  More robust algorithms
  • 104. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 104 How to Study  Basic courses  Programming skills  Engineering Mathematics  Signal and Systems  Digital Signal Processing  Advanced courses  Digital Image Processing  Computer Vision  Pattern Recognition  Artificial Intelligence
  • 105. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 105 Activities  中華民國影像處理與圖形識別學會(IPPR)  電腦視覺、圖學暨影像處理研討會(CVGIP)  IEEE (國際電機電子學會)  Transactions on  Image Processing  Pattern Analysis and Machine Intelligence  Medical Imaging  Conferences  ICIP, ICME, ICPR, ...
  • 106. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 106 The End Free for Any Questions
  • 107. Wang, Yuan-Kai(王元凱) Video Surveillance and Sousveillance p. 本簡報授權聲明  此簡報內容採用 Creative Commons 「姓名標示 - 非商 業性台灣 3.0 版」授權條款  歡迎非商業目的的重製、散布或修改本簡報的內容,但 請標明: (1)原作者姓名:王元凱; (2)圖標示:  簡報中所取用的部份圖形創作乃截取自網際網路,僅供 演講者於自由軟體推廣演講時主張合理使用,請讀者不 得對其再行取用,除非您本身自忖亦符合主張合理使用 之情狀,且自負相關法律責任。