一种改进的区域双目立体匹配方法

摘要

Binocular stereo vision matching is currently a hot and difficult problem in computer vision. Advantages and disadvantages of several common stereo matching algorithm are analyzed. An improved method that based on image segmentation of binocular stereo matching algorithm is proposed. Firstly, images are collected by binocular vision camera,then the images are corrected and denoising processing. Secondly,images are segmented using color features algorithm,and then a fast and efficient block matching algorithm are used for stereo image matching. In matching process,small window of sum of absolute differences (SAD) is used to look for matching points between two images. Finally, the final disparity map is got through filtering. Experimental results show that the algorithm can solve mis-matching error caused by the repeat regions, the low texture, texture similar,and block regions, clearly dense and accurate disparity map can be got.%双目立体匹配是机器视觉中的热点、难点问题.分析了区域立体匹配方法的优缺点,提出了改进的区域立体匹配方法.首先,采集双目视觉图像对对图像对进行校正、去噪等处理,利用颜色特征进行图像分割,再用一种快速有效的块立体匹配算法对图像进行立体匹配.然后,在匹配过程中使用绝对误差累积(SAD)的小窗口来寻找左右两幅图像之间的匹配点.最后,通过滤波得到最终的视差图.实验表明:该方法能够有效地解决重复区域、低纹理区域、纹理相似区域、遮挡区域等带来的误匹配问题,能得到准确清晰的稠密视差图.