Denoising Images using Wiener Filter in Directionalet Domain

Abstract

An algorithm for denoising image based on directionalet is proposed, and in each high-frequency subbands, adaptive Wiener filtering (window size 25 1× ) is performed. Then, the denoised image is obtained by reconstruction from the filtered coefficients. At last, all directional sub-images are averaged.. The experiments results showed that this method outperforms Wiener filtering in standard 2-D wavelet domain. The SNR is improved 1~3dB.

De-noising algorithm using sparse 3D transform-domain collaborative filtering and adaptive soft thresholding

“…Among them, spatial domain de-noising methods using total variation, bilateral filter, and non-local mean filter [2][3][4] are useful for noisy still images, but they have limitations in preserving details such as edges. So, a lot of transform-domain de-noising algorithms [5][6][7] were proposed to overcome the drawback of the abovementioned spatial domain de-noising algorithms. Especially, wavelet-based techniques dominate the current de-noising field [5][6].…”

“…So, a lot of transform-domain de-noising algorithms [5][6][7] were proposed to overcome the drawback of the abovementioned spatial domain de-noising algorithms. Especially, wavelet-based techniques dominate the current de-noising field [5][6]. They typically utilize the sparsity and the statistical properties of multi-resolution representation as well as the inherent correlation between frames in temporal dimension.…”

“…For the last decades, many de-noising algorithms for still images and video sequences have been developed [1][2][3][4][5][6][7][8][9][10]. Among them, spatial domain de-noising methods using total variation, bilateral filter, and non-local mean filter [2][3][4] are useful for noisy still images, but they have limitations in preserving details such as edges.…”

Multi-scale wavelet algorithm based noise reduction in ultrasonic images at elevated temperatures

“…Though this is achieved conventionally by Linear processing [5] in recent times non-linear techniques are being used. Mallat et al [3] have proposed that the wavelet transform has good local performance in both spatial and frequency domains.…”

“…The SNR is calculated using the equation (1) [5]. The denoising capabilities of different wavelet transforms are evaluated based on the PSNR value.…”