Research and development of stereo matching algorithm

Abstract:
Binocular vision is an important method for obtaining stereo perception of the real world, and is widely used in many fields. Stereo matching is the key part of binocular stereo vision, which finds the corresponding points in the image pair to establish the matching relationship, and then calculates the depth value of the target in the scene based on the internal and external parameters obtained during the camera calibration. This paper analyzes the current research status of binocular stereo matching algorithms at home and abroad, introduces the traditional stereo matching algorithms and the stereo matching algorithms based on convolutional neural network, and compares the characteristics of various algorithms, finally, the challenges faced by stereo matching algorithms and the outlook on future development are summarized.