基于像素分类校正优化的立体匹配算法

Stereo Matching Algorithm Based on Pixel Classification Rectification and Optimization

为了提高双目立体匹配的效率、消除匹配歧义,本文采用粗细精度调节的方法,提出了一种基于匹配像素分类校正优化的立体匹配算法.使用局部算法生成初始可靠的视差图,进行相关置信度检测和弱纹理区域检测将匹配像素分类,并使用自适应权重算法和Hirschmüeller校正方法分别对分类像素进行校正,再通过基于分割的算法对视差值进行优化,得到最终的视差图.实验结果表明,该算法可以有效地消除匹配歧义,能得到分段平滑、精度高的稠密视差图.

In order to increase the efficiency and remove the ambiguity of binocular stereo, a stereo matich- ing algorithm employing the method of hierarchical adjustment based on pixel classification rectification and optimization is proposed. Firstly, the initial disparity was generated by using the local matching stereo model. And the pixels were classified by adopting correlation confidence measurement and the detection of texture-less regions. Then their disparity values were rectified with adaptive support-weight approach and the Hirschmiieller rectification. The final disparity map was obtained form segment-based optimization the recti- fied disparity. The experimental results indicate that this technique can eliminate matching ambiguity and ob- tain piecewise smooth, accurate and dense disparity map effectively.