1-4 屋外大空間における多視点の画像情報を用いた頑健なオブジェクト位置推定方式の一検討(第1部門 映像表現)

抄録

In this paper, we propose an object extracting method in wide outdoor space, which is based on an object tracking method using particle filter in 3D-space. The likelihood of each particle is calculated by projecting it to every viewpoint and summing up all the likelihoods. Experimental results show that the integration of multi-viewpoint is useful for extracting object regions in wide outdoor space.