New multi-baseline stereo by counting interest points

This paper proposes a novel method for estimating depth from a long image sequence captured by a moving camera. Our idea for estimating a depth map is very simple; only counting interest points in images is integrated with the framework of multi-baseline stereo. Even by a simple algorithm, depth can be determined without computing similarity measures such as SSD and NCC that have been used for traditional stereo matching. Our method realizes robust depth estimation against image distortions and occlusions that are caused by camera motion. Note that, in this paper, intrinsic and extrinsic parameters of video camera are assumed to be calibrated in advance.
conference paper