杨晟, 李学军, 朱诗兵. 遥感影像匹配中的反定位误匹配剔除算法[J]. 武汉大学学报 ( 信息科学版), 2013, 38(9): 1083-1087.
引用本文: 杨晟, 李学军, 朱诗兵. 遥感影像匹配中的反定位误匹配剔除算法[J]. 武汉大学学报 ( 信息科学版), 2013, 38(9): 1083-1087.
YANG Sheng, LI Xuejun, ZHU Shibing. The Reverse Positioning Refining Algorithm for Auto-matching of the Remote Sensing Images[J]. Geomatics and Information Science of Wuhan University, 2013, 38(9): 1083-1087.
Citation: YANG Sheng, LI Xuejun, ZHU Shibing. The Reverse Positioning Refining Algorithm for Auto-matching of the Remote Sensing Images[J]. Geomatics and Information Science of Wuhan University, 2013, 38(9): 1083-1087.

遥感影像匹配中的反定位误匹配剔除算法

The Reverse Positioning Refining Algorithm for Auto-matching of the Remote Sensing Images

  • 摘要: 图像匹配的准确率和精度一直是瓶颈问题,直接制约遥感影像的自动化处理。在简要分析现有典型误匹配剔除方法的基础上,提出了能够适应遥感影像自动化匹配与处理的REPRAM(Reverse Positioning Re-fining Algorithm in Matching)算法,给出了REPRAM排除法和优选法的原理和性能分析,在不需要相机的外参数、内参数以及控制点的情形下,相对分离了正确同名点与错误同名点之间的相互影响,尽量保留正确同名点,最大限度剔除错误同名点。大量遥感影像和航拍影像测试表明,算法的稳健性强,在初始匹配正确率为30%的情况下,在设定误差容限内,算法的可靠率仍能在99.5%以上。

     

    Abstract: Image matching is one of the most important and challenging areas in computer vision and remote sensing.Especially for the wide base line images with large distortions,the accuracy and reliability of the matching is a bottleneck in the auto-disposition given the large volume of images.Based on an analysis of the latest refining methods,the Reverse Positioning Refining Algorithm in Matching(REPRAM) composed of REPRAM Excluding and the REPRAM Including parts is proposed.The theory and properties are discussed in relation to the refining ability for excluding more bads and including more rights,where the mutual impact is separated.Thus,there is no need for extrinsic and intrinsic parameters of the camera or control points.Although the original reliability of the rough couples is about 30 percent,the algorithm works robustly with a high degree of reliability above 99.5 percents under the ranging error threshold for several kinds of remote sensing images as well as for close shot pictures.

     

/

返回文章
返回