Image Dense Matching Method Based on Wavelet Edge Detection and Region Growing

Abstract

Aiming at the computer vision image, firstly, image edge feature is obtained using the edge detection algorithm based on wavelet analysis, selecting some seed points, comparing the gray similarity of seeds, the result of the initial match is based on the SIFT matching algorithm. The weighted SSD (Sum of Squared Difference) series, as the objective function, spread the seeds to the rest of the image matching area with the original growing strategy. Experimental results show that the algorithm was effective to large parallax pictures, images without calibration, and images which textures are sparse.