An adaptive search range algorithm for multiview motion and disparity estimation

An adaptive search range algorithm is proposed for multiview motion and disparity estimation. By analyzing the motion and disparity vector distribution in neighboring views and anchor frames respectively, an adequate motion and disparity search range can be chosen. As the search range can be effectively reduced, the saving of computation time of motion and disparity estimation can be up to 73% with negligible PSNR degradation. Furthermore, this algorithm has been implemented in a hardwired motion/disparity estimation module for a H.264 MVC encoder and achieves the requirement of real-time coding for HD1080p 30fps stereoscopic video operating at the frequency of 300 MHz.