摘要
针对同一目标的红外与可见光所形成的异源图像对,提出了一种基于改进SIFT与互信息的算法,寻找两幅图像中的相关点,从而实现两幅图像的匹配。为了改善SIFT算子的匹配效果,本文提出的改进方法,成功地增加了正确匹配点数;同时使用局部互信息量作为判断标准,可以在正确匹配点数即使占劣势的情况下,依然能排除错误的匹配点,扩展了匹配算法的适用性。结果和实际测试表明,两种算法结合能显著改善匹配效果。
In order to match IR and visible images from the same scene, this paper proposes an algo-rithm based on mutual information and improved SIFT to find the relevant points in the two images.First, this paper proposes an improved SIFT to improve the matching effect of SIFT for IR image, and this meth-od can successfully increase the correct matching points.Then, local mutual information is used as a cri-terion, as which could eliminate error matching points, even if the number of error matching points is more than correct matching points.Finally, the test results show that the combination of the two algo-rithms can significantly improve the matching effect.
出处
《航空兵器》
2014年第5期15-18,59,共5页
Aero Weaponry
基金
航空科学基金项目资助项目(20120177004)
国家自然科学基金资助项目(51275120)
关键词
异源图像匹配
改进SIFT
互信息
multi-sensor images matching
improved SIFT
mutual information