基于SIFT算法的异源遥感影像自动匹配研究  

Research on automatic registration of multi-source remote sensing images based on SIFT algorithm

作　　者：张建花[1] 白仲斐[1] 惠广裕[1] 

机构地区：[1]中国飞行试验研究院测试所,陕西西安710089

出　　处：《电子设计工程》2011年第13期176-179,共4页Electronic Design Engineering

摘　　要：由于不同传感器、多时相、多分辨率、多波段的遥感图像的光谱特征、空间特征、纹理特征等存在较大差异,为遥感图像的匹配带来了困难。主要利用图像特征点提取方法,使用具有尺度不变特性的SIFT(Scale Invariance Feature Transform)方法,对异源遥感图像进行配准和图像进行拼接操作,并进一步对SIFT算法进行优化,采用双向匹配策略。实验证明该算法具有稳定、可靠、快速等特点,适用于存在光谱特征、空间特征、纹理特征差异的异源遥感图像的精确配准,同时实验验证了双向匹配算法用于SIFT特征点匹配中的优越性,证明其为一种好的匹配测度。Since multi-source,multi-temporal,multi-resolution and multi-band remote sensing images are too different in spectral characteristics,spatial characteristics,as well as texture features.So it is full of difficulty to match these remote sensing images.We have focused our studies on the point feature extraction method to match the multi-source images in this paper.Then match the images.Based on the above theory,we have done some experiments.The results show that SIFT is one of the most effective local feature of rotation,scale,and illumination invariant.Changes of perspective,affine transformation and noise also maintain a certain degree of stability.SIFT based image matching algorithm is analyzed,and a bidirectional matching algorithm is proposed to improve the accuracy of image matching.The experimental results show that the proposed algorithm is effective.

关 键 词：遥感图像 SIFT 配准 双向匹配 RANSAC 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]