Image De-noising Algorithm Combined With Fourier Transform And NSCT

The images have a huge impact on all aspects of human’s life and production.However, due to the influence of the imaging equipments and the externalenvironment, the images are often contaminated by noise. Noise may affect human’svision and image post-processing. In order to effectively extract the information fromthe noisy image, it needs image processing technology to reduce noise. the research ofFourier transform de-noising algorithm and NSCT de-noising algorithm has studiedinclude:(1) Expounding the significance of image de-noising and research backgroundand researching several classic airspace de-noising algorithms and transform domainde-noising algorithms, which make adequate theoretical preparation for the improvedalgorithm of subject follow-up research.(2) The improved Fourier de-noising algorithm. Firstly determine the bestWiener filtering window’s parameter values and the adjustment factor value byexperiment, and then analyze the effectiveness of the algorithm.(3) The improved NSCT de-noising algorithm which based on the hierarchicalBayes threshold to estimate the noise standard deviation at different scales. In addition,an adaptively energy correction threshold according is proposed to the relationship ofenergy in all directions between image information and the NSCT coefficients.(4) Combined Adaptive threshold de-noising algorithm with Fourier transformand NSCT. Firstly the noisy image is de-noised in Fourier domain. Secondly theremaining noise is removed in NSCT domain by the algorithm that based on stratifiednoise estimation and adaptive Bayes threshold,which is combined with a flexiblemulti-scale and multi-directional energy correction threshold scheme; Finally, thede-noising image is reconstructed by low-frequency and high-frequency sub-bandscoefficients. The experimental results show that the proposed algorithm is moreeffective in image de-noising,(5) An image de-noising algorithm combined with Fourier transform and NSCT.First, extract the details of the template and the edge of the template using originalimage; and then take NSCT transform and Fourier transform for de-noising imagerespectively to get the NSCT edge image and Fourier detail image; Finally fuse theNSCT edge image and Fourier detail image using Local Energy Fusion.In order to verify the effectiveness of the de-noising algorithm, performing thede-noising experiments to the noisy image of Lena, Barbara and Library, subjectiveevaluation and PSNR value evaluation method of combining the effect of imagede-noising evaluation, and using subjective evaluation and PSNR value evaluationmethod of combining to evaluate the effect of image de-noising.