Shadow Detection and Disparity Processing Based on Binocular Vision.

In order to solve the problem of the large disparity errors in shadows, this paper proposes a sound framework for shadow detection and disparity processing. Firstly, the shadow boundary is obtained by segmenting the road image; secondly, the disparity is recalculated by template matching with a larger sliding window for the shadow boundary; then, the disparity with lower confidence in the original disparity is removed; finally, the blank value is filled by interpolating the disparity of the surrounding non-shadowed part. It is verified that the method proposed in this paper can effectively remove the disparities in the shaded part that cause the wrong height, and the height at the shaded part is significantly reduced after the treatment by this paper.