CAJ | 학술논문

提出一种基于稀疏空域分割与梯度方向直方图的异源图像匹配方法。将图像分为多个子区，计算每个子区的梯度方向直方图；计算直方图的汇聚度，并根据汇聚度计算直方图相似性；采用两步搜索法，寻找与基准图中与实时图直方图相似度最高的窗口即为正确匹配结果。采用仿真图和真实异源图像对算法进行了测试，结果表明，这种方法能够可靠快速实现异源图像匹配。
제출일충기우희소공역분할여제도방향직방도적이원도상필배방법。장도상분위다개자구，계산매개자구적제도방향직방도；계산직방도적회취도，병근거회취도계산직방도상사성；채용량보수색법，심조여기준도중여실시도직방도상사도최고적창구즉위정학필배결과。채용방진도화진실이원도상대산법진행료측시，결과표명，저충방법능구가고쾌속실현이원도상필배。
Multi-sensor images matching method based on multi-area histogram of oriented gradient is proposed.Firstly,the images were segmented into some sub-areas.Secondly,histogram of oriented gradient was calculated for every subregion.Then,the histograms similarity was computed with convergence degree as power.In the end,through two-step searching,the histogram with the highest similarity was found and the corresponding image was the matching result.The algorithm was tested using emulational images and real multi-sensor images.Experimental results showed that the new matching algorithm can match multi-sensor images effectively and efficiently.