Interacting with computational devices was limited to using a keyboard and mouse for many years. More recently, touchscreens have become popular, especially in mobile computing. However, accurate three dimensional gestures are still difficult to recognize. One approach is the use of depth cameras, which are currently receiving much attention again in academic and in commercial fields due to the release of inexpensive consumer electronics like Microsoft’s Kinect in 2010.
This presentation reflects depth camera technologies, and how gestures can be used for new interface designs. The first part establishes a framework to evaluate the suitability of depth cameras for gesture recognition. The second part focuses on interactions in three dimensional virtual spaces. I developed a gesture interface for an existing application to navigate in virtual spaces and to construct LEGO models. A user study was performed to compare the gesture input with traditional mouse and keyboard input. Finally, I discussed the influences of the findings for future research in gesture interfaces.
9. Agenda
Range-imaging technologies
How can a depth map be computed?
Experiment: Gestures based interfaces
Which gestures are suitable to control a computer?
Conclusion for gesture interfaces
Do gestures in"uence future interactions?
23. Comparison Suitable depth cameras
Range
Amplitude
modulation
Accuracy
Time of
Structured
light
"ight
Image resolution
<
Time of
"ight
Amplitude
modulation
Structured
light
<
<
Time of
"ight
Amplitude
modulation
Structured
light
24. Comparison Suitable depth cameras
§
§
§
§
Depth resolution and range
Update rate and latency
Robustness
Spatial resolution of depth image
29. Results Times for traditional and gesture input
time [min] 35
30
25
20
15
10
5
0
Mouse
Gesture
30. Results Further results from questionnaires
§ Gesture input is more fun, but less e$cient and
accurate
§ Correlation between participants who never played
physical LEGO and long construction times for gesture
interface
§ Semiotic gestures are easy to use
§ Most participants prefer a combination of gesture input
together with traditional input
32. Navigation Accuracy and precision
§ Imprecise actions can be controlled by gestures
§ Gestures might support creativity
33. Manipulation Accuracy and precision
§ Gesture input cannot replace precise mouse and
keyboard input
34. Conclusion Gesture Interpretation
§ Most current gesture interfaces are built for controlling
the computer; however, gestures are a subconscious
expression of human feelings
§ Experience of haptic feedback in ergotic physical
gestures is helpful for simulating the haptic feedback in
related pseudo-ergotic gestures