FPGA Implementation for Binocular Stereo Matching Algorithm Based on Sobel Operator

Aiming at the low accuracy of stereo matching algorithm caused by the larger gray change, an improved algorithm is proposed. Sobel operator is used to compute the gradient of pixels. Based on the gradient histogram, the adaptive thresholds are derived and the support window is determined automatically. The neighbor pixels are compared with the average of all pixels in the window instead of the center pixel to complete the census transform. The disparity image is abtained through searching the best match point in the left and right images. The hardware implementation with parallel procssing is elaborately designed to improve the capability of the large data processing and computational efficiency based on FPGA. The experimental results show that the improved scheme and hardware structure can obtain disparity map with higher accuracy and stronger robustness.