Denoising in Wavelet Packet Domain via Approximation Coefficients

In this paper we propose a new approach in the wavelet domain for image denoising. In recent researches wavelet transform has introduced a time-Frequency transform for computing wavelet coefficient and eliminating noise. Some coefficients have effected smaller than the other''s from noise، so they can be use reconstruct images with other subbands. We have developed Approximation image to estimate better denoised image. Naturally noiseless subimage introduced image with lower noise. Beside denoising we obtain a bigger compression rate. Increasing image contrast is another advantage of this method. Experimental results demonstrate that our approach compares favorably to more typical methods of denoising and compression in wavelet domain. 100 images of LIVE Dataset were tested، comparing signal to noise ratios (SNR)، soft thresholding was %1. 12 better than hard thresholding، POAC was %1. 94 better than soft thresholding and POAC with wavelet packet was %1. 48 better than POAC.