Research And Implementation Of Image Matching Based On Fan Grid

Image matching techniques is widely used in image analysis, image registration,face recognition, image reconstruction and cultural studies. Common grid-basedimage matching is mapping the boundary of the image information to a rectangulargrid on the law, through the rectangular grid matching to determine the degree ofsimilarity of the images, this method can’t handle the rotation of the imagedeformation, so we proposed a fan-shaped grid-based image matching method.This method is scaling the image proportional to the standard disc, standard discis the smallest circum circle of the image. Then putting standard disc cut intofan-shaped grid, and image maps to the fan-shaped grid, by the rotation of thefan-shaped grid to find the best matching, compared with preset threshold todetermine whether the image is matching or not. For simple image, because theimage is only a small part of the boundary extraction boundaries are mapped to thefan-shaped grid, while, most of the information is meaningless backgroundinformation, we took advantage of the uniqueness of prime factors, which after theprime product is decomposed. We present the query image boundaries with thecorresponding prime product, the candidate image borders are also presented withcorresponding prime number. By charging whether the remainder is exist, we canjudge the fan grid of query image and the fan grid of candidate image is matching ornot, for improving the matching efficiency.In order to remove some images which the difference degree is over thethreshold, and increase the efficiency of the matching time, we define the image of themutual information entropy. The experiments confirmed that the use of the combineof the fan-shaped grid method and the image of the mutual information entropy candeal effectively with the deformation of image rotation. By the experiments analysiswe achieve the best value of the threshold relationships, we also compared with otheralgorithms, the algorithm which proposed in our paper is better than the others.