A stereo matching algorithm based on census transformation and dynamic programming

Abstract

Stereo matching is a research hotspot in binocular vision. Speed and accuracy are key problem. "Stripes" phenomenon often appears in disparity map generated by using traditional stereo matching algorithm based on dynamic programming. In order to solve this problem, the absolute difference combining Census transformation similarity measure strategy is constructed, the reliability of the initial matching cost is improved. At the same time, the new smooth model is established to simplify the parameter selection and the uncertainty of matching is effectively eliminated. Dynamic method of variable search radius is designed, which is able to keep the disparity smooth in the area where view image gradient is small and allow the disparity jump in the area where view image with strong gradient. The matching accuracy and speed are improved by using this method. The experimental results show that the effectiveness of the proposed algorithm in this paper.

Real-Time Depth Map Estimation From Infrared Stereo Images of RGB-D Cameras

“…For a pixel in the match image, a Census transformation window with a size 7 × 7 is selected. Then a 0/1bit string for the Census transformation can be obtained (Lu and et al, 2014). In the same way, the bit string for the search point of the target image is obtained.…”

A Real-Time Infrared Stereo Matching Algorithm for RGB-D Cameras’ Indoor 3D Perception

“…If it is larger, it is set to 1, and if it is smaller, it is set to 0. Finally, a 0/1-bit string can be obtained [40].…”