Direct Depth Value Extraction Method for Various Stereo Camera Arrangements

In this paper, we propose a direct depth map acquisition method for a convergent camera array as well as a parallel camera array. Since image rectification is necessary for conventional stereo matching methods, disparity values are found in the same horizontal line of stereo images. Consecutively, the acquired disparity values are transformed to depth values. However, image rectification may lead to unwanted consequences for a convergent camera array. Thus, the proposed method excludes image rectification and directly extracts depth values via the epipolar constraint. The MRF energy for depth estimation is optimized by modified constantspace belief propagation (CSBP) which has low complexity. In order to acquire a more accurate depth map, we apply postprocessing finally. Experimental results show that the proposed method has fewer limitations than conventional methods and generates depth map stably. Keywords—disparity map, epipolar constraint, stereo matching