Optimization Of Image Stitching Algorithm And Research Of Roaming System

Image stitching technology is one of the most popular topics in recent years. It mainly is applied in panoramic roaming system, panorama video stitching, medicine and remote sensing images processing and so on. Image registration is the most important phase of image stitching technology. Scale Invariant Feature Transform algorithm (SIFT) and Speeded Up Robust Feature (SURF) are the most classic algorithms in image registration phase. Aiming at the existing problem of the two, this paper verifies two kinds of optimization algorithm and develop a panoramic roaming system to verify the image stitching algorithm.In order to reduce the time complexity of the traditional SIFT algorithm related the feature point detection and matching, an optimization image registration algorithm is verified. The feature points are extracted by Trajkovic algorithm. The SIFT algorithm distribution descriptor method is adopted to allocate feature points descriptor parameters. Then sparse dimension reduction principle is utilized to reduce feature points descriptor dimensions. Finally, the similarity measure algorithm based on bidirectional matching is used to match them. Simulation experiments show that optimization algorithm in comparison with the traditional SIFT algorithm and other optimization SIFT algorithms is nearly the same in the accuracy of feature point matching. However, it has obvious improvement in terms of feature points matching speed.In view of the fact that feature point extraction speed is slow and its matching accuracy is not high in the current image stitching process, an optimization image stitching algorithm is verified. Namely, firstly, the stitching images are maked by the down-sampling process; then detect the feature points on the half image area and use the Sequential Similarity Detection Algorithm (SSDA) to purify the feature points; finally, fuse the images. Simulation experiments show that compared with the image stitching method of the traditional SIFT algorithm and the image stitching method of the traditional SURF algorithm, the optimization algorithm greatly reduces calculated data and has obvious advantages in image stitching time.Finally, on the basis of image stitching technology, this paper develop a panoramic roaming system of Anhui Polytechnic University. The system can realize topographic map navigation, the scene music and the background music play, Anhui Polytechnic University video playback, thumbnail navigation and so on.