Image Denoising Using Adaptive Dual-Tree Discrete Wavelet Packets Based on Signal-to-Noise Ratio Estimation

Abstract

In order to remove the noise in the image while preserving the the edges and texture details of the image, the image denoising algorithm using adaptive dual-tree discrete wavelet packets(ADDWP) is proposed in this paper. It combines the dual-tree discrete wavelet transform(DDWT) and the wavelet packets. We estimate the signal-to-noise ratio of the sub bands by median estimator. In ADDWP, DDWT sub bands are further decomposed into wavelet packets with an isotropic decomposition, so that the resulting wavelets have elongated support regions and more orientations than DDWT wavelets. To determine the decompoisition structure, we using the signal-to-noise ratio to estimate the distributing of the denoising in order to search the more denoising sub bands to decomposition it again. So we can get adaptive decompoisition structure of wavelet packets. The new algorithm has significantly lower computational complexity than a previously developed optimal basis selection algorithm. For denoising the ADDWP coefficients, the new threshold function is proposed based on the signal-to-noise ratio of the sub bands. The experiment result improves our algorithm is efficient.