Ultrasound Image Registration Based on Polar Transform and Block Matching

This paper describes a novel image registration algorithm based on polar transform and block matching method to cover the image geometric transformations. Polar transform makes the algorithm easily and more accurately to find rotation angle under the fact that rotation in Cartesian domain corresponds to pure translation in polar domain. And the widely used sum of absolute difference (SAD) matching criteria and modified gradient descent search algorithm are introduced to improve the search speed on the condition that does not decrease the search accuracy. The algorithm can automatically find the registration parameters between input images in the presence of translation, rotation, and scale simultaneously. Experiment results show the accuracy and robustness of the novel image registration algorithm.