Denoising of an image using embedded zerotree wavelet (EZW) coding algorithm

The recent research on image processing, compression and denoising techniques, had given wavelet image coding the most attention due to its high performance with reasonable quality of image. Major developments have been achieved by wavelet transform along with subband classification and zerotree quantization. Among the various image compression techniques, embedded zerotree wavelet (EZW) is the pioneer as it has introduced the concept of zerotree quantization and embedded coding algorithm. Till now, the EZW has been used for image compression. The thesis studied image compression and denoising using EZW. The code was written in MATLAB. The biorthogonal filter which has both linear phase and symmetry was used to decorrelate the noised image data (i.e. to determine the wavelet coefficients). After finding the wavelet coefficients, soft thresholding was applied on them to denoise the image. Then the EZW algorithm was applied on these coefficients to get the compression. The algorithm, code and the results of denoising an image were presented. For compression part, this method formed only a part of the compression process and hence the compression is complete only when followed by a symbol encoder like arithmetic coding.