Medical Image Fusion Brain SURF Algorithm

Image fusion is the multiple images of the same scene to obtain a plurality of image sensor of different, or multiple images of the same sensor acquisition in different period of time into a contain more image information process. Over the years, due to the rapid progress of image sensor, image fusion becomes a research hotspot in many fields. This paper focuses on image fusion algorithm based on surf algorithm. The main work can be summarized as follows:First of all, this paper expounds in detail the principle of image imaging and image fusion process and a variety of methods are introduced, and some image registration algorithm is introduced in the paper. So this paper studies the feature point matching source images and get the surf algorithm of image fusion.Secondly, the SIFT algorithm and SURF algorithm of the commonly used image feature detection algorithm was carefully introduced, at the same time, this paper proposed an improved algorithm based on SURF algorithm. The improved algorithm proposed in this paper has two aspects:one is to control the number of feature point detection, on the other hand, is to limit the feature point detection area. The image fusion is looking for pixels with the same feature in some may overlap region, so it can be the first feature point possible region, then can be gathered in the feature points in the region to choose one point, at this point as the center of a circle with a radius, number to form a circular interval to control features point.Finally, through the matching images using the RANSAC algorithm to eliminate the error feature matching, the matching of image transformation matrix using the method of least squares to solve the optimization, the transformation matrix parameters obtained by progressive gradually out of the weighted average method for image fusion. After the combination of subjective and objective evaluation, studies the performance of fusion algorithm in this paper.