摘要
计算机视觉领域的PMVS算法在多视角影像匹配中具有很好的匹配效果,但将其直接引入航空倾斜影像匹配还存在一些问题:随着影像数据量的增加,PMVS算法对计算机内存占用呈几何增长;PMVS算法针对大倾角的影像匹配存在一些问题;在候选面片生成时,PMVS算法采用固定大小的窗口采集待匹配影像像素,若窗口选择过小,则参与匹配的像素信息少,匹配效果不佳,若窗口选择过大,则将导致匹配耗时过长。针对上述问题,采用分块PMVS策略,提出了改进的自适应窗口匹配算法。实验结果表明,该算法能有效解决将PMVS算法直接引入航空倾斜影像密集匹配中存在的问题,具有一定的实用性。
In the field of computer vision,PMVS algorithm has very good matching effect.However,there are still some problems in introducing PMVS algorithm directly into aerial oblique image matching.With the increase of the amount of image data,PMVS algorithm also has a sharp increase in computer memory.PMVS algorithm has some problems for image matching of large dip angle.PMVS algorithm adopts fixed size window to collect pixels of image to be matched,if the window selection is too small,the pixel information involved in the matching is less,and the matching effect is not good.While,if the window selection is too large,the matching time will be too long.According to the above problems,we used the block PMVS strategy to put forward a matching algorithm based on self-adaption window.The experimental result shows that the proposed algorithm can effectively solve the problems of introducing PMVS into the dense matching of aerial oblique images,which has certain practicability.
作者
倪标
王铮尧
NI Biao;WANG Zhengyao
出处
《地理空间信息》
2020年第7期66-69,I0006,共5页
Geospatial Information
关键词
密集匹配
倾斜摄影
PMVS算法
分块
自适应窗口
dense matching
oblique photography
PMVS algorithm
block
selfadaption window
