Image Denoising by Soft Shrinkage in Adaptive Dual Tree Discrete Wavelet Packet Domain

Image Denoising has remained a fundamental problem in the field of image processing. It still remains a challenge for researchers because noise removal introduces artifacts and causes blurring of the images. In the existing system the signal denoising is performed using neighbouring wavelet coefficients. The standard discrete wavelet transform is not shift invariant due to decimation operation. To overcome the problem, signal is transformed to wavelet domain using dual-tree complex wavelet transform which exhibits approximate shift invariance and improved angular resolution. In the proposed system the image denoising is performed using adaptive dual-tree discrete wavelet packets (ADDWP), which is extended from the dual-tree discrete wavelet transform (DDWT) as wavelets give a superior performance in image denoising. With ADDWP, DDWT sub bands are further decomposed into wavelet packets with anisotropic decomposition, so that the resulting wavelets have elongated support regions and more orientations than DDWT wavelets. For denoising the ADDWP coefficients, a statistical model is used to exploit the dependency between real and imaginary parts of the coefficients. Using the statistical model the noise in the image is removed and finally the reconstructed denoised image is achieved.