제목

Bounding box based on probability density for the pedestrian tracking

저자명 Jaehyun So, Hernsoo Hahn, Youngjun Han

초 록

This paper proposes bounding box using probability density for the pedestrian tracking. If we track
some pixel like KLT algorithm[1] in the pedestrian tracking, tracking will be difficult. Because tracking environment is very complex, some tracking points are converged in background. For supplement tracker faults, this paper implements bounding box using probability density of tracking points. We use that tracker is KLT(Kanade-Lucas Tomasi feature tracker). Because this tracking algorithmcan track some pixels in image, it has robust performance and fast processing speed.

Keyword The pedestrian tracking; Probability density; Bounding box; Kanade-Lucas Tomasi feature tracker
원문 수록처The Fourth International Conference on Smart IT Applications(SITA 2012), Technical Session 812
자료유형 학술대회자료