The Research Of Image Enhancement Algorithm Based On Nonsampled Contourlet Transform And Fuzzy Logic

With the development and in-depth research of electronic technology and computer technology, digital image processing technology has been gotten giant strides development. Image enhancement, a low-level image processing operations and pre-processing stage of image processing, plays an important role in high-level operation of the image and has been widely used in many fields, such as medical treatment,industries, national defense, agriculture,aeronautics and aerospace.In this dissertation, in-depth and comprehensive research work has been done on transform and fuzzy enhancement algorithm, and two image enhancement algorithms were proposed based on non-sampling Contourlet transform (NSCT) and fuzzy logic for gray and color image.NSCT is sparser to represent image than Contourlet transform and wavelet transform, and is more flexible multi-scale and multi-directional, and its shift invariance can avoid the Gibbs phenomenon effectively, and consequently, can capture geometric information of the image fully to represent the image accurately. The proposed algorithm estimate the variance using the adaptive bayesian threshold, and classify NSCT coefficients into three categories: strong edges, weak edges, and noise using local orientation statistics, and then introduce the frequency domain fuzzy contrast definition of weak edges and enhance the fuzzy contrast using the segmented fuzzy enhancement function. The experimental results demonstrate that the proposed algorithm achieves better robustness, for it can not only enhance the details and texture and avoid enhancement of background information, but also can suppress noise effectively.For RGB color images, in the HSV color space according with human vision, the luminance component was enhanced using proposed enhancement algorithms for gray image, and the saturation component was stretched applying the adaptive histogram equalization (AHEQ), and color component was unchanged. Comparing with classic color restoration multi-scale Retinex (MSRCR) enhancement algorithm, the experiment results show that the proposed algorithm can enhance the image brightness, detail and color information more effectively, which can not only eliminate the color distortion and halo artifact phenomenon, but also has better ability of anti-noise and more perfect visual effect according with human visual perception characteristics.