An accurate stereo matching method based on color segments and edges

Highlights：

• We propose a new matching algorithm that divides the matching problem into segment and edge matching and combines them for post-processing to obtain the disparity.

• This new matching method transforms pixel matching to pixel segment matching and edge matching, which reduces the time complexity. The algorithm can be used on lightweight devices.

• Stereo matching algorithms of binocular vision still suffer from low accuracy when dealing with natural scenes different from the training set scenes, making it tough to guide industrial robots with high precision. Our algorithm can be implemented in an industrial robot environment for high-precision needle threading guidance, something that neither traditional binocular matching nor deep learning matching algorithms can do at present.