| 摘要 |
A processing method for a pair of stereo images is provided. The method includes: extracting a pair of edge images from the stereo images, each edge image having edge pixels, each edge pixel of one of the edge images being associated with an overlap record and a disparity record; providing a plurality of image lateral shifts to sequentially determine a plurality of overlap levels between the pair of edge images; and updating the overlap record and the disparity record associated with a selected edge pixel to a maximum overlap level and a most-likely disparity, respectively. The maximum overlap level is a maximum among a plurality of associated overlap levels to which the selected edge pixel contributes. The most-likely disparity corresponds to the maximum overlap level. |
