A Study On Image Matching Based On Fractal Method

Fractal theory is a very active branch in modern mathematics and non-lineal science. In the last decade,it has achieved a great success in image processing. Especially fractal coding, which uses compact Iterative Function System (IFS ) to denote the image, has a good effect on image compression as well as image matching. After analyzing and comparing several traditional Gray-Matching algorithms, this thesis proposes several image matching methods based on fractal coding.The first method: Image matching method based on a pair of parameters ( D , LMSE ) of a range block.Fractal coding theory is based on self-similarity or mutual similarity of image. As a range block, it can be described by a group of parameters (a domain block andLMSE ). By using the pair of parameters correspondent to a range block as the image matching characteristic, a new image matching is improved. The matching image is divided into range blocks with the same size, then the range blocks are coded in terms of the preset domain blocks, which results in a LMSEi and the LMSE of the matching image. An optimal matching image is obtained by the Euclid distance between the above LMSE and the LMSE orgthat the domain blocks found by itself.The second method: Image matching method based on Fractal Neighbor Distance (FND).FND gives a quantitative measure of the input-output characteristics of the fractal code of the encoded image. The fundamental mechanism behind this matching lies in the uniqueness of the attractor of a fractal code, the attractor has the invariance to image translation, rotation, scaling and illumination. It's proved to be a standard to measure image similarity in image matching. With the object image as the input image while decoding, we get the output image after the first iteration of the IFS .then compare this output image with the input image by FNDi , Thus find the most optimal matching block.