计算机与数字工程2019，Vol.47Issue(3)：643-648,6.DOI:10.3969/j.issn.1672-9722.2019.03.032
基于改进Census变换与最小生成树的立体匹配算法
An Algorithm Based on Improved Census Transform and Minimum Spanning Tree for Stereo Matching
摘要
Abstract
Aiming at the problems of noise-sensitive and with low matching ratio in the disparity discontinuity region and weak texture region of the existing local matching algorithm,a multi-scale stereo matching algorithm for improved Census transform is proposed. The weighted average gray value of all the pixels in the support window is used as the reference value of the Census transform and the weight is determined by spatial information and pixel different information together. Then the noise tolerance α is set for the reference value,and the bit string is obtained through four states. At multiple scales,the minimum spanning tree is merged to complete the cost aggregation. The experimental results demonstrate that the average false matching ratio of standard ste?reo image pairs obtained by the proposed algorithm is 3.65%,and it is 4.53% with the noise. The average false matching ratio of the 27 extended stereo image pairs is 10.8%. In the parallax discontinuity region and the weak texture region,the false matching ratio is further reduced by the proposed algorithm,and it shows better robustness for noise.关键词
机器视觉/立体匹配/Census变换/最小生成树Key words
machine vision/stereo matching/Census transform/minimum spanning tree分类
信息技术与安全科学引用本文复制引用
于修成,宋燕,李航..基于改进Census变换与最小生成树的立体匹配算法[J].计算机与数字工程,2019,47(3):643-648,6.基金项目
国家自然科学基金(编号:61673276) （编号:61673276）
上海市自然科学基金(编号:18ZR1427100) （编号:18ZR1427100）
中国空气动力研究与发展中心开放课题(编号:20184101)资助. （编号:20184101）
