Aiming at the registration of multi-sensor remote sensing images, a fast and effective image registration method was presented. Based on hierarchical candidate point set matching, the registration algorithm makes full use of the advantages of hierarchical matching and candidate point set matching approach, reduces feature extraction and searching space, and introduces distance calculation into the determination of homologous control point pairs, it can availably decide homologous control point pairs and decrease mismatching rate. The increased distance calculation has much lower time cost, and it does not change with the size of images. It is concerned with the number of final matched points in images. The effect of image registration algorithm was evaluated by subjective and objective method. Experiment results show the proposed algorithm can accurately register multi-spectral images and panchromatic images with some shift, rotation angles and background noises, and it can increase the speed and precision of registration.
Van Wyk B I, Van Wyk M A. A POCS-based graph matching algorithm[J].IEEE Transactions on Pattern Analysis and Machine Intelligence.2004,26(11):1526-1530
Lowe D. Distinctive image features from scale-invariant keypoints[J].International Journal on Computer Vision.2004,60(2):91-110
Rosenfeld A. Digital picture processing[M]. Orlando: Academic Press, 1982:67-90
Woods N A, Galatsanos N P, Katsaggelos K K. Stochastic methods for joint registration, restoration, and interpolation of multiple undersampled images[J].IEEE Transactions on Image Processing.2006,15(1):201-213
Wang Zhijun,Ziou D,Armenakis C,et al. A comparative analysis of image fusion methods[J].IEEE Transactions on Geoscience and Remote Sensing.2005,43(6):1391-1402