一种顾及纹理特征的自适应密集匹配方法

An adaptive dense matching method considering texture features

针对半全局立体匹配(SGM)算法对于弱纹理区域误匹配率较高的问题,该文提出了一种顾及纹理特征的自适应密集匹配方法,也是SGM算法与该文提出的一种跳过噪声点的区域增长自适应窗口匹配算法的结合。首先,以某一待匹配点为中心构建一个一定大小的初始窗口。其次,计算窗口内其余各点与待匹配点灰度差的绝对值之和,根据该值的大小判断纹理丰富程度。然后,根据纹理的丰富程度自适应地选择匹配算法,对于纹理丰富区域,采用SGM算法进行匹配;对于弱纹理区域,则采用该文提出的一种跳过噪声点的区域增长自适应窗口匹配算法。最后,利用Middleburry网站提供的标准测试图像对进行试验分析,结果表明:该算法能在有效提高SGM算法在影像弱纹理区域的匹配精度的同时,降低程序运行的时间,且在一定范围内,图像弱纹理区域占比越大,效果越明显。

Aiming at the problem of high mismatch rate of semi-global stereo matching(SGM) algorithm for weak texture region matching,an adaptive dense matching method Considering texture features was proposed,which was a combination of semi-global stereo matching algorithm and a region growth adaptive window matching algorithm that skips noise points proposed in this paper.Firstly,an initial window of a certain size was constructed with a certain matching point as the center.Secondly,the sum of the absolute values of the gray difference between the remaining points in the window and the points to be matched were calculated,and the texture richness was judged according to the size of the value.Thirdly,the matching algorithm was selected adaptively according to the texture richness.For texture rich areas,SGM algorithm was used for weak texture regions,an adaptive window matching algorithm for region growth skipping noise points was used.Finally,the experimental analysis was carried out by using the standard test image provided by the Middleburry website.The results showed that the algorithm could effectively improve the matching accuracy of the SGM algorithm in the weak texture region of the image and reduce the running time of the program.