A Study Of Denoising In Shaped Fiber Image

Image denoising algorithm is of great significance as a prerequisite for digital image processing.Any good image processing system makes image denoising as a matter of priority.This image denoising research is a part of the research that is sponsored by the Scientific Research Foundation for the Returned Overseas Chinese Scholars,State Education Ministry(The Project-sponsored by SRF for ROCS,SEM) and the Foundation of National Excellent Doctoral Dissertation of China.Also, this research is with the need of automatic textile fiber inspection from Shanghai Entry-Exit Inspection and Quarantine Bureau.In the course of capturing,shaped fiber images are often mixed with a lot of noise.Image noise not only seriously affects the image of visual effects,but also brings certain difficulties to the subsequent image analysis and understanding in the image processing.So images denoising in pre-processing of shaped fiber images is a very important step for the follow-up treatment. Traditional image denoising algorithms such as mean filter,median filter and gauss filter belong to the isotropic diffusion.They removed the noise of the image while undermine the character of the image.The image denoising technology based on partial differential equation,which belongs to the anisotropic diffusion,can solve this problem.During the course of denoising,image features and their directions are extracted. Because there are less smoothes in the locations with strong image features,more smoothes in the locations with weak image feature,and there are minimal smoothes in the directions across the image features, and maximal smoothes in the directions along the image features.The denoising can make noises removed while preserve edge well.In this paper,the partial differential equation's theoretical foundation and application in the digital image processing is researched, and more attention is paid to the partial differential equation in image denoising,including its theory,denoising model,and development. What's more,this paper analyses the Person-Malik model,Cattle model, ALM model,TV model and the model you & kaveh in two ways of filter and energy minimization,and sums up the characteristics,the strengths and weaknesses of each model. After analyzing the noise source in the shaped fiber image and make certain the noise model,a denoising scheme is presented to smooth the Gaussian noise and impulsive noise in the image.This scheme is based on Total Variation algorithm and median filtering,can keep the information of fiber edges while denoising.Meanwhile,based on the threshold of the parameters,the efficiency of the algorithm and the effects of denoising is well balanced.The experimental result shows the good performance of the proposed algorithm.