The Study And Application Of Feature Extraction Method Base On Fractal And Wavelet Theory

With the coming of informative society,the way from which people get information is not to be limited in the number,the word and the symbol,but also from more and more kinds of images.Most of problems in image processing are feature extraction and recognition,such as image retrieval and classification,image compression and coding,image restoration or image reconstruction,target detection and recognition,edge extraction,image filtering,image segmentation and image signal separation and so on.Due to its diversity and complexity,image feature extraction is not only a hot issue but also the difficult one.Fractal theory is an active branch of non-linear scientific field.Its research objects are geometric bodies that are unsmooth and irregular in nature.In the field of image processing,practice shows that there is a strong relativity between fractal dimension(FD)and textural coarseness.The larger the FD,the more coarseness the image texture has.On the contrary,the smaller the FD,the more smoothness the image surface has.Therefore,it is significant to use FD to research the textural image feature extraction.In the period of current research,people have gotten some results but problems still exist.Multi-resolution analysis of wavelet theory reflects the process that people's recognizing object is from low resolution to high resolution.Fractal theory gives people one of the dialectic thinking ways to research the relation of part and entirety. For this reasons,wavelet theory is a tool that reveal fractal local scaling property. That means using enlargement and displacement of wavelet to analyze the fractal local singularity.Base on above all,this paper did the following works:1.Research fractal theory.Researching DBC FD,Variation FD,Peleg-Blanket FD,Fractional Brownian Motion FD and Multi-fractal FD.There is confrontation program with comparing different kinds of FDs for one image.2.Research wavelet theory.This paper proposed a mixed method that combined fractal and wavelet.This paper shows the necessity of the mixed method,how to choose wavelet base,and then introduce two ways to get the mixed FD——fractal-wavelet FD.3.Based on the wavelet theory,FD could be got.This paper used program, which contrast fractal-wavelet FD to other FDs with different images.And the result proved the superiority of fractal-wavelet FD.4.Because of the complexity of texture image,texture segmentation's researching is always hot but difficult.There is texture analysis program for researching kinds of texture segmentation methods.Then the paper put forward a new texture segmentation method based on fractal and wavelet.Though experiment, we compared many images and evaluated the results,which proved that fractal-wavelet texture segmentation method has higher accuracy to other methods.5.At last,this paper introduced the fractal-wavelet VC++ program.