Study Of Image Enhancement Based On Fuzzy Theory

With the development of computer technology, modern mathematicalapplications in image processing have become increasingly broad. In recentyears, fuzzy theory in image processing application is promising. Thispaper studies several image enhancement algorithms based on fuzzy theory.Firstly the paper discusses the filtering process. Because part of theimage contain noises, these image must be filtered first. The paper studiesa de-noising filter of multi-value image fuzzy entropy. This method picksnoise points first, and then de-noise based on the minimum fuzzy entropyfilter. Then membership function is added to improve noise detection,which is used for weighting calculations at the time of de-noisingrecovery.In gray-scale correction, the paper studies the Lee imageenhancement algorithm and its improved algorithm. Nonlinear mappingstructure is improved, so that we can tensile the gray value of image that iscentralized. In order to adaptively select a nonlinear configuration toadjust the gray scale for different brightness image, probability of histogram of the image is divided. On this basis, the paper improves theprobability to fuzzy division to determine the brightness of images,avoiding artificially selecting nonlinear structure.In the domain-transformation algorithms, the paper studies thesingle-level fuzzy enhancement with its improve algorithm and improvemulti-level fuzzy enhancement algorithm. For value selection ofmulti-level fuzzy edge enhancement algorithm, the paper improves edgethreshold selection method. The paper extracts the edge set by partialfuzzy entropy, and calculates the histogram of the edge set, and thenselects the local maxima as the edge threshold.