Application of wavelet threshold to image de-noising

Wavelet based techniques have been used for a number of years to de-noise images. By means of the wavelet packet analysis, the low frequency part and high frequency part of the super stratum can be concurrently further broken down, and therefore, more exact analysis of localities can be conducted. By thresholding the wavelet packet transform coefficients of the noisy image, the original image can be reconstructed correctly. Different threshold selections and thresholding methods are discussed. A novel cubic thresholding function is presented based on wavelet packet shrinkage. Quantifying the performance of image de-noising schemes by using the peak signal-to-noise ratio (PSNR) and mean square error (MSE), the performance of the cubic threshold scheme is compared with the hard and soft threshold schemes. The experiment shows that image de-noising using the cubic threshold performs better rthan that using the hard and soft threshold.