Remote Sensing Imagedenoising And Fuzzy Enhancement Basedon Nonsubsampled Contourlet Transform

Remote sensing images can be collected as we provide a lot of useful informationis to a large extent dependent on the collection of remote sensing image quality,imaging and remote sensing image transfer process, always inevitable under theinfluence of various factors, leading to remote sensing images usually have a numberof disadvantages, effects such as noise, poor vision, low resolution and brightness.Cause the grayscale of remote sensing image can not to cover entire range of remotesensing sensors can achieve, leading to a decline in the quality of remote sensingimage, therefore, it is necessary first of all on the remote sensing image denoising andenhancement for further analysis.Denoising and enhancement of Remote sensing image is a very classic andimportant issues of image signal processing, since denoising and enhancement ofremote sensing images can improve the readability of remote sensing images,although after improvement of remote sensing image and not necessarily similar tothe original image, but it highlights the contours, edges of the image information, andattenuation of noise.In recent years, the algorithm of fuzzy and Multiresolution analysis in the field ofimage preprocessing has been widely applied and was a huge success, which underCunha and M N Do, who proposed a nonsubsampled Contourlet transform (NSCT),image denoising and enhancement using this mothod can access a very good results.This research under the base of nonsubsampled Contourlet transform (NSCT), andcombined with the fuzzy algorithm, proposed a method of Remote sensing imagedenoising and fuzzy enhancement based on nonsubsampled Contourlet. Main contentsare as follows:1. Proposed image adaptive threshold denoising method based on wavelet andnonsubsampled Contourlet transform (NSCT). First this method uses wavelet estimate the noise strength of Noisy Remote sensing images, then according to thestrength of noise determines the shrinkage threshold according to the neighbouringnonsubsampled Contourlet transform coefficients, the scale of the coefficients and thenoise level. Compared with the wavelet hard-thresholding, the contourlethard-thresholding and the nonsubsampled Contourlet transform hard-thresholdingdenoising method,the proposed method is obviously reduces the Gibbs phenomenonand superiors both in PSNR and in vision.2. A new remote-sensing image fuzzy enhancement transform based onnonsubsampled Contourlet (NSCT) is proposed. Firstly, we using nonsubsampledContourlet transform make the Remote sensing images to be transformed into thehigh-pass sub-band and low-pass sub-band, secondly we set the threshold in thehigh-pass sub-band, with the high-pass subband coefficients which usually is greaterthan the threshold was enhanced by linear enhancement, the high-pass subbandcoefficients which less than the threshold was set zero, the low-pass subbandcoefficients was enhanced by the fuzzy contrast enhancement method. Experimentalresults in the improvement of entropy and the average show the effectiveness of thismethod.