Multi modal registration of structural features and mutual information of medical image

摘要

Based on automatic matching of medical images synthesized by complementary characteristic, this paper makes full use of SURF algorithm's good scale rotation invariance and HOG algorithm to describe local shape information better, showing good robustness when affected by geometric and optical deformation of graphics. Each performance of algorithm is comprehensively evaluated through the experiment, with accuracy of reaching 98%, meeting the actual application requirements. The algorithm shows good performance and application potential in the field of medical image matching. At the same time, since medical images are affected by lighting, equipment, scene and other factors in the acquisition process, there are large differences, which affect the robustness of the algorithm to a certain extent. Experiments show that this method has high accuracy in digital medical image matching and can be used to construct automatic matching system of medical images. (C) 2018 Elsevier B.V. All rights reserved.