Abstract: Image stitching is a technique that combines two or more images that are taken from different view of the same scene to obtain a panoramic image. Image stitching is used in medical applications for stitching of X-ray images. As the traditional system of X-ray machine cannot capture the whole body structure in a single image. So, images stitching solves this problem by combining two or more x-ray images into a large view one. This paper proposes an algorithm which automatically stitches the x-ray images with overlapped region. The stitching method is based on ORB features (Oriented FAST and Rotated BRIEF features). The proposed system is designed with five stages, preprocessing, features extraction, features matching, Homography estimation and images stitching. In feature detection stage, Oriented FAST approach is used. In feature description stage, Rotated BRIEF approach is applied. The two primary parameters for measuring the stitching performance are the quality of the resultant image and the processing time. Therefore, the main objective of this paper is to produce a high quality image stitching system with low processing time. First, we compare many different features detectors. We test SIFT method, SURF method, Harris corner detector and ORB approach to measure the correct detection rate of the feature points and computation time. Second, we measure the quality of result images that produced by stitching system of different feature detection methods. From experimental results, we conclude that ORB approach is the fastest, more accurate, and higher performance. Keywords: Biomedical images; Feature based approach ; Image stitching; ORB features ; Panorama

Applied Science and Computer Science Publications