Gait recognition using a view transformation model in the frequency domain
ECCV'06 Proceedings of the 9th European conference on Computer Vision - Volume Part III
| Hi-index | 0.00 |
Does self-viewpoint "life-log" movie taken by a CCD camera at a person's forehead contain any cue information by which one can identify his/her own movie from others? In particular, do dynamic and non-episodic aspects of the self-viewpoint movie (scene shake by walking) make it possible to distinguish "my" movie from others? To examine this question, subjects were asked to distinguish a movie taken by a CCD camera placed at their forehead and body from a movie taken at the head and body of others. Before experiment, all subjects walked through a gymnasium whose visual condition was kept constant between subjects to take a movie of subjects' viewpoint as experimental stimuli. Any episodic visual event was eliminated from the content of the movie. The rate of correct judgment of distinguishing the movie from "my" viewpoint from others was examined with 2IFC procedure and was significantly higher than that of control condition in which subjects' judgment were executed with still image, suggesting that self-viewpoint movie might contain some non-episodic ID information.