High-Resolution Depth Map Generation by Applying Stereo Matching Based on Initial Depth Informaton

In this paper, we propose a new algorithm to generate a high definition (HD) depth map using a standard definition (SD) depth camera and HD stereoscopic cameras. In order to obtain the initial depth information for the left camera, we perform the three-dimensional (3-D) warping technique using depth information acquired by a depth camera. After depth values from a depth camera are converted into initial disparity values for stereo matching, we search only a small neighboring region of the initial disparity to find a more accurate disparity value. Then, we generate 3-D edge segments to refine the depth map in boundary areas of objects. Experimental results show that the proposed scheme generates a more accurate depth map than traditional stereo matching algorithms.