融合绝对误差和与Census变换的双目立体图像匹配算法

针对传统立体匹配算法受光照和纹理结构等变化的影响，匹配成功率低或容易出现匹配失败的问题，提出一种融合SAD和Census变换的双目立体图像匹配算法。该算法基于图像预处理，首先集成SAD和Census算法的优势，以取代单一匹配代价计算方法；然后，使用一种十字交叉代价聚合策略来提升匹配效果；最后，为验证本文算法的有效性，选择不同的实验场景，对不同分辨率的图像进行立体匹配的综合对比实验，实验结果表明本文算法在匹配效率和匹配精度等方面具有显著优势，适用于复杂场景中高质量视差图的生成。

Binocular Stereo Image Matching Algorithm Combining SAD and Census Transformation

Aiming at the problem that the traditional stereo matching algorithm was affected by changes in lighting and texture structure. An optimization strategy which combines sum of absolute differences(SAD) and Census transform was proposed. Firstly, on the basis of image preprocessing, the algorithm integrated the advantages of SAD and Census algorithms to replace a single matching cost. Then it used a simple cross method to aggregate cost to improve matching accuracy. Finally, in order to verify the effectiveness of the algorithm, we chose different experimental scenes to carry out the comprehensive comparative experiment of stereo matching for different resolution images. The experimental results show that the algorithm has significant advantages in matching efficiency and can generate disparity maps with good results. It is suitable for generating high-quality disparity maps in complex scenes.