Highperformance local-texture-information weighted SAR template image matching

A high performance matching method for Synthetic Aperture Radar (SAR) image matching based on sub-block local texture information weighted Normalized Cross Correlation (NCC) is proposed in the paper. Though The NCC measure is robust under uniform illumination changes in image matching, there are two main drawbacks: false matching result due to the High ratio occlusions and noise of high level, and the high computational cost if Full Search Strategy (FSS) is used. To tackle the two problems, Block partitioning strategy, combined with texture information analysis and the Fast Fourier Transformation (FFT) algorithm are designed to improve the performance of the conventional NCC algorithm. According to block partitioning strategy, template image is firstly divided into sub-blocks of certain rows and cols; then texture information related to each sub-block is extracted and designed as weight of NCC. FFT algorithm and Integral Images are adopted to make algorithm much Faster. Experimental results show that the proposed algorithm is more robust and much faster than the conventional NCC algorithm. (5 pages)