X-ray Medical Images Stitching Based On Corner Matching Using Two-way Asynchronous Voting Strategy

Medical image stitching is very important in medical diagnosis and treatment such as the measurement of scoliosis, lower limb deformity and extremity fractures correction, and so on.Traditional film image stitching process, which is mainly achieved manually, is not only time-consumption but also low precision; With the development of digital imaging technology, it is possible to stitch medical images automatically and seamlessly. Research status of digital image mosaic technology and the characteristics of medical images were analysed, and then, an image stitching method, which employs feature matching based on two-way asynchronous voting strategy, is proposed in this paper. Specific methods are as follows:Firstly, B-spline function and wavelet transform function were introduced to improve the harris corner detection operator, and then an improved multi-scale harris corner detection algorithm is proposed to extract the feature space of images to be registered. For poor quality images, the improved MSR algorithm based on human visual was used to filter and enhance them before corner detecting; Then, a feature matching based on two-way asynchronous voting strategy was proposed to complete corner matching coarsely, so as to make the features, which belong to images to be stitched, correspond to each other roughly; In order to make the images to be stitched be registered precisely, the RANSAC algorithm was used to remove mismatching points, and to optimize the image transformation model;Finally, many methods of image reconstruction were analysed, and weighted average fusion strategy was used to reconstruct the image, which has been registrated, to achieve image stitching seamlessly.