Study On Denoising Method Of Digital Image Based On Correlative Characteristics

Image denoising is the one of very important and basic image processing technical.The accurate and reliable result is the important guarantee of image restoration,division,feature extraction,image recognition etc. The purpose of Image denoising is to process degradation image and make it back to the ideal state. An important standard of image denoising is to remove the noise and protecting image detail.Because Image noise and details have similar features,this makes the image denoising and details keeping become a hard to solve the contradiction.This paper based commonly used image denoising methods, through the relevance theory focused on the research image noise detection and value filling. Its main job content as follows:1) The significance and dynamic of image denoising were introduced, and the digital image and its noise were analyzed.The denoising method of wider salt and pepper noise was selected to study.2) The basic characteristics of digital image were described,and the characteristics of gray image and color image were analyzed.The general method of subjective assessment and the objective evaluation were given.3) The theory of correlation was studied, the description and derivation of one-step correlation were given, and the methods of one-step singular correlation detection and one-step correlation fill value were introduced. Through the preliminary experiment of gray image, the flowchart of this paper was given.4) According to the characteristics of gray image and its noise, one-step singular correlation method was modified to improve its accuracy of noise detection.By studying the neighborhood, the weighted value suitable for gray image was given.5) According to the characteristics of color images and its noise,and the complexity of the color images,one-step singular correlation method was further to improvement.The pretreatment of passivation image detai outstanding noise characteristics was added before noise detection,to furthei increase the accuracy of noise detection. For the problem of false color, the double correlation of image pixels and pixels weights was studied,then the most relevant vector fill value method were given.Finally,through the simulation experiments and the corresponding image quality evaluation analysis, and the corresponding image denoising method were compared, the validity and practical of this method were verified.The detection and fill value of this paper based on one-step correlation and the weighted correlation method,made the image edge details and signal to noise ratio index improve obviously, meanwhile, the related features of images were into parameters of algorithm, the whole process of denoising is more universality.