摘要
用于近景摄影测量的普通数码影像与航空影像相比,存在更为复杂的影像变形和影像遮挡等问题,这使得其影像匹配的难度大大增加。针对近景数码影像的特殊性,提出一种改进搜索策略的概率松弛匹配算法。该算法采用格网点和特征点相结合的方式来确立初始点匹配过程;并从核线、视差等方面进行多重约束,保证匹配的连续性和正确性。实验结果表明,该算法适用于近景数码影像的匹配,在复杂的高山峡谷地区的立体匹配正确率可达到98%。
If being compared with aerial imagery, the ordinary digital imagery used for close-range photogram- metry brings more complex problems such as image distortion and occlusion, which greatly increases the difficulty of image matching. Given the particularity of close-range imagery, a probabilistic relaxation matching algorithm with improved search strategies is provided, which uses the combination of grid points and feature points to establish the initial points matching process, and multiply constraints such as epipolar line and parallax, to ensure the continuity and correctness of the image matching. The experiment shows the algorithm is applicable for close-range digital im- age matching and the stereo matching precision can reach up to 98% in the complex alpine-gorge regions.
出处
《科学技术与工程》
北大核心
2013年第7期1713-1717,共5页
Science Technology and Engineering
基金
国家自然科学基金(51079053)
江苏省自然科学基金(SBK201221489)资助
关键词
影像匹配
概率松弛
普通数码影像
近景摄影测量
特征点
metryimage matchingfeature pointprobabilistic relaxationordinary digital imagesclose-range photogram-