Research On Matching Technology For Two-dimensional Polygonal Image

Technology for two-dimensional image reassembly is an important topic in pattern recognition and computer vision, and is wildly applied in archeology, Forensic science and Molecular Biology. Matching and reassembly of image fragments is the key technology for two-dimensional image reassembly. It has been some achievements based on contour matching of image fragments. After summarizing the techniques of traditional method of image fragments, this paper discussed polygon image fragments which are very unique, and a suitable matching algorithm is proposed for2D polygons fragments. The main content of this paper is the following aspects:1. This paper reviews and summarizes the common research methods and procedures of the image fragments matching. After that, this paper focuses on the corner detection for two-dimensional image reassembly based on contour curves. A corner detection method based on improved Hough transform is proposed. This method effectively reduces the peak diffusion phenomena, and improves the accuracy of corner detection.2. According to the characteristics of two-dimensional polygon fragments, a matching algorithm based on the angle of the corner points of the polygon is proposed. After completing the previous step of corner detection, the counterclockwise rotation angle of corner points on the contour are considered as matching feature.3. An appropriate strategy for polygon fragments reassembly is proposed. After the completion of the match, choose the most effective reassembly order by matching strategy, which is from the edge to the center of the original image. The reassembly method proposed by this paper was validated experimentally, and the reassembly method of several special polygon fragments was discussed in the last part.