Sub-pixel matching for small baseline and height ratio stereo image pairs based on enhanced phase correlation algorithm
Li L · 2014
Phase correlation is very popular for optical flow surveillance,remote sensing image matching and computational vision.This paper proposed an enhanced phase correlation algorithm for small baseline and height ratio stereo image pairs based on traditional algorithm.Integer of disparity from image pairs was computed by traditional algorithm,and each pixel both on reference image and matching image was made windows for Hanning transformation.When PC matrix was decomposed by SVD and unwrapped,both vector of line and column were identified and then estimated the disparity on sub-pixel level by least square fit.At last,a median shift propagation algorithm was used to refine all the matching results.The result of the experiment was proved that the algorithm would have a high ratio of success with high precision.