Ali K. Nahar; Hadeel N. Abduallah
Abstract
In many of the digital image processing applications, observed image ismodeled to be corrupted by different types of noise that result in a noisy version.Hence image denoising is an ...
Read More ...
In many of the digital image processing applications, observed image ismodeled to be corrupted by different types of noise that result in a noisy version.Hence image denoising is an important problem that aims to find an estimateversion from noisy image that is as close to the original image as possible. In thispaper, introduces firstly was applied method of computing one and twodimensionalframelet transform .The applying method reduces heavily processingtime for decomposition of image keeping or overcoming the quality ofreconstructed images. In addition, it cuts heavily the memory demands .Also, theinverse procedures of all the above transform for multi- dimensional casesverified. Secondly, many techniques are proposed for denoising of gray scale andcolor image. A new threshold method is proposed and compared with the otherthresholding methods. For hard thresholding, PSNR gives (13.548) value whilethe PSNR was increased in the proposed soft thresholding, it gives (14.1734)PSNR value when the noise variance is (20). Some of the above denoisingschemes are tested on Peppers image to find its effect on denoising application.The noisy version with SNR is equal to (11.9373 dB), the denoising image usingWT with SNR is equal to (17.4661 dB), the denoising image using SWT withSNR is equal to (18.1459 dB), the denoising image using WPT with SNR is equalto (19.3640 dB), the denoising image using FT with SNR is equal to (21.9138dB). Finally the denoising image for color image using FT with SNR is equal to(27.3443 dB).
