Hardware design of disparity computation for stereo vision using guided image filtering
Shen‐Fu Hsiao, Chih-Hsuan Chang · 2018
Disparity estimation is the key operation in stereo matching. In general, there are two categories of stereo matching methods: global and local. Local stereo matching methods are fast due to less computation while global methods can generate more accurate depth information at the cost of more computation complexity. This paper presents a low-cost and high-quality local stereo matching design which utilizes mean filtering and guided image filtering to improve the quality of depth computation. Mean filtering is implemented using a moving-sum method, instead of the conventional integral image method, in order to reduce bit-width of internal memory buffers. Vertical-striped-based guided image filtering is applied to the weighted median filtering for disparity refinement with low memory cost. Experimental results show that the proposed design can achieve satisfactory disparity estimation with much smaller hardware cost compared with prior similar designs.