改进的立体像对稠密匹配算法葛亮.pdfVIP

DOI :10.13229/ki.jdxbgxb2010.01.043 40 　 1 吉林 大学 学 报 (工学 版 ) 　Vol.40　No.1 2010 1 Journal of Jilin University (Engineering and Technology Edition) 　Jan.2010 葛　亮, 朱庆生, 傅思思, 罗大江, 刘金凤 (重庆大学 计算机学院, 重庆400030) 　:针对目前主流方法对图像中纹理单一区域匹配效果不佳 问题, 提出了 一种改进 立 体像对稠密匹配算法。该算法首先利用区域增长技术找到图像中 纹理单一区域, 然后将整 个区域作为匹配基元以得到纹理单 一区域 稠密视差图。 相对于点基元, 区域基元包含 信 息更多, 且在图像中不易重复出现, 因此可以减少误匹配发生 几率。在国际标准测试图像上 进行了实验, 结果证明该算法 可行性与准确性。 :计算机应用;计算机视觉;立体匹配;基于区域 匹配;纹理单 一;视差计算 :TP391　　:A 　　:1671-5497(2010)01-0212-06 Improved image dense stereo matching algorithm GE Liang, HU Qing-sheng, FU Si-si, LUO Da-jiang, LIU Jin-feng (College of Comp uter Science, Chongqing University, Chongqing 400030, China) Abstract:To obtain better disparity result of the less textured area of image, an improved dense stereo matching algorithm is proposed.First, the image is divided into several less textured or well textured areas based on region-growing strategy .Then, the dense disparity map of the less textured area is identified through the matching based on the whole region.Compared to pixel, a region contains more information and repeats seldom occur, thus reducing the probability of matching error.The proposed algorithm was tested with the international standard image data and compared with several existing algorithms.Experiment results show that the performance of the improved algorithm is better in handling less textured area and can be use in practice. Key words:computer application;computer vision;stereo matching;matching based on region;less textured;disparity computation [1] 　　, 2 : [3] 。