Enhanced Feature Matching Algorithm for Medical Image Registration

Abstract

AbstractIn medical imaging technology, image registration plays a vital role. Image registration is a process of transforming the test image to reference image based on different parameters and features. In traditional feature extraction methods the feature points are unevenly distributed for which the redundant features are more. This results to time consuming process and low registration accuracy. Hence it is a challenging task to match the images with higher accuracy. In the proposed method enhanced and robust image registration algorithm is based on SURF and BRISK feature extraction methods with upgraded GMS and RANSAC algorithm. In this method the keypoints of the medical images are detected through SURF algorithm then features are extracted by BRISK algorithm and false match keypoints are removed through GMS algorithm and improved RANSAC algorithm is used. Experimental results show the efficiency and effectiveness in terms of runtime, accuracy compared to traditional feature extraction matching algorithms which can be implemented in real time for occasion with the requirements of high accuracy.KeywordsImage registrationSURFFASTBRISKGMSRANSAC