Application of self-similarity of image wavelet coefficient in image denoising

wavelet transform;  Mallat algorithm;  Hidden Markov tree model;  Image denoising

The structure and statistical law of the wavelet coefficient of the image are discussed, and it is pointed out that the wavelet coefficient at each layer of the quadtree has self-similarity, and a hybrid Gaussian model is established to describe the wavelet coefficient by using this self-similarity. In addition, the four-tree layer has correlation between the nodes between the layers, and an implicit Markov tree model (HMT) is established by using the self-similarity information and correlation between the wavelet domain coefficients to remove Gaussian white noise in the image signal, and the Lenna image is used for experimental simulation. Compared with the traditional low-pass filtering method, this method can better preserve the details and edge information of the image while filtering out the noise. The results show that the implicit Markov tree(HMT) model reflects the characteristics of the image more accurately and has a better denoising effect.