engleski

Identification of Human Motion by Nonparametric Background Modeling and Star-skeletonization

Using video in machine understanding has recently become a significant research topic. One of the most explored areas is activity understanding from video sequences. The automatic detection and identification in video has many useful applications like biometrics, pedestrian safety, robot-human interaction, etc. There are several commercially available programs for human motion detection and analysis like [1] and Pfinder[2]. Some of this programs use stereo imagery, which is not always available, and also cardboard human models, which has many drawbacks.Used algorithm eliminates some of these drawbacks.

background substraction; nonparametric modeling; human motion identification; skeletonization

nije evidentirano