Fast morphological pyramid matching algorithm based on the Hausdorff distance

摘要

Because the two images used in realistic image matching are generally from the different kinds of sensors, and the physical peculiarity of these images are different, it is very difficult to match the different kinds of images. Aiming at the maching problem between the synthesis aperture radar (SAR) images and the optic images, a fast morphological pyramid matching algorithm based on the Hausdorff distance was proposed. Firstly, in this algorithm edge images were taken as the characteristic space which were extracted respectively from the optic images and the SAR images, and then a distance measurement based on partial mean-Hausdorff distance was adopted as similarity measurement, after that a multiscale flat-structuring elements close-open morphological hybrid pyramid based on the contour bougie (CB) morphology was adopted as the search strategy. This matching algorithm between the real SAR image and optical image was simulated and the results show that it has high matching accuracy and fast computation speed.