Research On Image Denoising Based On Wavelet Transform

An image is often corrupted by much noise in its acquisition or transmission, noise significantly impairs the quality of the received image, this will result in a big problem for further image processing (such as feature extraction, image compress and signal detection etc). So, in order to raise the quality of the image, we must remove the noise while retaining as much as possible the important signal features.In recent years, the wavelet theory praised as "mathematical microscope" has been rapidly developed. For the characteristics of its low entropy, multi-analysis, relativity removal and flexible bases, the wavelet transform has been widely applied in the field of denoising. The algorithm of image denoising will be presented in this paper based on wavelet theory, in the setting of additive white Gaussian noise. The proposed algorithms are as follows:With the wavelet threshold denoising theory proposed by D.L.Donoho, a new threshold function is presented in this paper, which based on the setting of additive white Gaussian noise. This new function extends the existing algorithm such as hard threshold and soft threshold functions. Through adjusting the parameter, it overcomes the shortcomings of the hard threshold function which is discontinuous and soft threshold function which has warp in processing, at the same time, it is adaptive for energy.Least Squares Wavelet Support Vector Machines (LS-WSVM) is presented in this paper based on wavelet kernel and Least Squares Support Vector Machines, and then a new image filter operator is obtained. At the same time, taking into account that wavelet coefficients are correlated in a small neighborhood, a new diamond window is proposed. Finally, a new image denoising algorithm is presented based on LS-WSVM.