Affine Template Matching Based on Multi-Scale Dense Structure Principal Direction

Abstract

Reliable and robust matching of multimodal images is a challenging task because of the nonlinear grayscale distortion caused by radiometric differences and the geometric deformations caused by viewpoint changes between multimodal images. To address these problems, we propose a dense descriptor that uses multiscale structure principal direction to capture the structural features of the image, as well as a novel dissimilarity measurement. The proposed descriptor adapts to nonlinear gray distortion effectively and is robust to image noise. The dissimilarity measurement addresses the reversal of direction caused by intensity inversion. In addition, we propose an improved pixel selection method to speed up the algorithm. Experimental results show that the proposed matching algorithm is offers a superior performance relative to the current state-of-the-art methods.