Multi-feature fusion stereo matching algorithm based on improved Census transform

Abstract:
Aiming at the problem of low matching accuracy and poor anti-interference performance of the local stereo matching algorithm in depth discontinuous and weak texture areas, we propose a multi-feature fusion stereo matching algorithm based on improved Census transform. Firstly, the improved Census transform, the normalized sum of the four-direction gradient information and color information are used as the matching cost to enhance the reliability. Then, dynamic cross-domain and 4-path scan line optimization are used for cost aggregation to improve the matching accuracy. Finally, the disparity is calculated according to the Winner-take-all (WTA) rule, and the multi-step disparity optimization strategy is introduced to obtain the final disparity map. Experimental results show that the proposed algorithm can effectively solve the problem of mismatching weak texture and edge regions, and the average mismatch rate of our algorithm is 4.56%. Compared with other algorithms, the matching accuracy is significantly improved, and it has good practicability.