Foreground Segmentation from Occlusions Using Structure and Motion Recovery

The segmentation of foreground objects in camera images is a fundamental step in many computer vision applications. For visual effect creation, the foreground segmentation is required for the integration of virtual objects between scene elements. On the other hand, camera and scene estimation is needed to integrate the objects perspectively correct into the video. In this paper, discontinued feature tracks are used to detect occlusions. If these features reappear after their occlusion, they are connected to the correct previously discontinued trajectory during sequential camera and scene estimation. The combination of optical flow for features in consecutive frames and SIFT matching for the wide baseline feature connection provides accurate and stable feature tracking. The knowledge of occluded parts of a connected feature track is used to feed an efficient segmentation algorithm which crops the foreground image regions automatically. The presented graph cut based segmentation uses a graph contraction technique to minimize the computational expense. The presented application in the integration of virtual objects into video. For this application, the accurate estimation of camera and scene is crucial. The segmentation is used for the automatic occlusion of the integrated objects with foreground scene content. Demonstrations show very realistic results.