Fast Stereo Matching Method Based on Two-step AD-Census Fusion
Tong Wang, Yangfan Sun · 2021
With the development of industrial information technology, an online industrial quality inspection technology which can make the computer quickly and efficiently obtain the 3D information of the industrial production line environment is particularly important. With the development of computer technology and the extensive application of stereo matching technology, more and more industrial inspection systems have adopted online monitoring programs such as binocular stereo vision. Due to the shortage of the traditional stereo matching algorithm, such as complicated process and demanding hardware conditions, a fast AD-Census stereo matching algorithm was proposed in this paper, combining the idea of image pyramid, aiming at the characteristics of large resolution and high demand on algorithm efficiency of industrial online inspection. In this paper, AD algorithm is simple and fast, Census algorithm is insensitive to illumination difference and has high robustness to edge information and makes full use of the idea that multi-scale pyramid pays attention to the main trend, which can achieve a higher efficiency of stereo matching algorithm without too much loss of accuracy.