Matching and Co-Registration of Satellite Images Using Local Features

Satellite image matching and co-registration are two key stages in image registration, fusion and super-resolution imaging processes where images are taken from different sensors, viewpoints or at different times. This paper presents: (1) An evaluation for the co-registration process using local features, (2) A registration scheme for registering optical images taken from different viewpoints in addition to radar images taken at different times. The selected feature detectors have been tested during the key point extraction, descriptor construction and matching processes. The framework suggests a sub-sampling process which controls the number of extracted key points for a real time processing and for minimizing the hardware requirements. After getting the pairwise matches between the two images, a registered image is composited by applying bundle adjustment and image warping enhancements. The results showed a good performance level for SURF over both SIFT and ORB detectors in terms of higher number of inliers and repeatability ratios. The Experiments were done on different optical and radar images from Rapid-Eye, TerraSAR-X, and ASTER satellite data for some areas in Germany and Egypt.