区域生长的半全局密集匹配算法
发布时间：2018-10-05 08:37
【摘要】：针对传统的半全局匹配算法在处理视差变化大、遮挡严重的城市航空影像时,存在匹配精度下降、匹配效率低下的问题,提出了一种基于区域生长的半全局密集匹配方法。采用区域生长算法获取影像的初始视差,并从初始匹配点中挑选可靠的点作为视差控制点;利用区域生长获取的视差图,限制各个方向动态规划的过程以加速最优路径的搜索;通过视差控制点对动态规划的路径进行修正,避免错误匹配代价的传播。基于城区无人机影像的实验结果表明,所提算法不仅可以提高匹配结果正确率,还能使耗费的内存和时间都不到原算法的50%。
[Abstract]:A semi-global dense matching method based on region growth is proposed to solve the problem of poor matching accuracy and low matching efficiency in the traditional semi-global matching algorithm when dealing with urban aerial images with large parallax variation and severe occlusion. The region growth algorithm is used to obtain the initial parallax of the image, and the reliable point is selected as the parallax control point from the initial matching point, and the parallax map obtained by the region growth is used to limit the process of dynamic programming in each direction in order to accelerate the search of the optimal path. The path of dynamic programming is modified by parallax control points to avoid the propagation of error matching cost. Experimental results based on urban UAV images show that the proposed algorithm can not only improve the accuracy of the matching results, but also make the memory and time consumed less than 50% of the original algorithm.
【作者单位】： 河海大学;
【基金】：江苏省自然科学基金项目(BK2012812)
【分类号】：P231;TP391.41
本文编号：2252720
[Abstract]:A semi-global dense matching method based on region growth is proposed to solve the problem of poor matching accuracy and low matching efficiency in the traditional semi-global matching algorithm when dealing with urban aerial images with large parallax variation and severe occlusion. The region growth algorithm is used to obtain the initial parallax of the image, and the reliable point is selected as the parallax control point from the initial matching point, and the parallax map obtained by the region growth is used to limit the process of dynamic programming in each direction in order to accelerate the search of the optimal path. The path of dynamic programming is modified by parallax control points to avoid the propagation of error matching cost. Experimental results based on urban UAV images show that the proposed algorithm can not only improve the accuracy of the matching results, but also make the memory and time consumed less than 50% of the original algorithm.
【作者单位】： 河海大学;
【基金】：江苏省自然科学基金项目(BK2012812)
【分类号】：P231;TP391.41
【相似文献】
相关硕士学位论文 前2条
1 张宏伟;高分辨率影像中农村道路提取方法研究[D];重庆大学;2016年
2 魏冬梅;基于遥感影像滑坡边界自动提取方法的研究[D];西南交通大学;2013年
，本文编号：2252720
本文链接：https://www.wllwen.com/kejilunwen/dizhicehuilunwen/2252720.html
