A mixture of Noise Image Denoising using Sevenlets Wavelet Techniques

Abstract

The noise has always been present in digital images when coding, image acquisition, transmission and processing steps has often corrupted by noise. Noise is challenging to confiscate from the digital images without the noise model’s prior information preserving edges. That is why the assessment of noise models is essential in the revise of image denoising techniques. A novel approach to improve the performance of an image’s quality and visual perception must be noise-free. The essential features like edge details should be retained as much as possible due to the increased traffic caused by multimedia information and digitized form of representation of images. This research articulates a brief general fundamental proposal of the noise model. The input image has debased with different noise probability density of Gaussian (G), Speckle (S), Salt and Pepper (SP) noise and a mixture of noise (G + S + SP). The Wavelet technique’s methodology using Sevenlets wavelet are Haar, Daubechies, Coiflets, Symlets, Discrete Meyer, Biorthogonal and Reverse Biorthogonal input Lena standard image has decomposed using Discrete Wavelet Transform (DWT). The decomposition process, as accomplished by discriminating the input image with lower and higher image coefficients. Filtering techniques are employed to deplete the noise present in an image. Hence the quantitative investigation of noise model at hard and soft thresholding is analyzed, improving image quality by increasing PSNR and decreasing MSE to have better performance. HIGHLIGHTS The images may be corrupted randomly in the transmission line; it is significant to retrieve exactly without loss of data; it is a challenging task. The images are examined with a different mixture of noises with corrupting at various percentage levels. In transmission channels the data will be corrupted, the primary aim is to retrieve actual data. For the testing purpose Gaussian (G), Speckle (S), and Salt & Pepper (SP) noise added to the Lena image have been corrupted and denoised. Denoising using methodology Discrete Wavelet Transform (DWT), to obtain better clarity image quality furthermore noise is filtered using Weiner and Median Filters. The mixture of noises of 0.1, 1, and 10 % density noise (G+S) (Gaussian (G) pulse Speckle (S)), G+S+SP (Gaussian (G) plus Speckle (S) pulse Salt & Pepper (SP)) added to the Lena image and denoised, have obtained improved denoised images comparing to previous research results.