Research on LIF Image Denoising Based on Wavelet-Domain Multiscale HMT Model

In the process of computing laser induced fluorescence (LIF) image, a lot of noises will be brought into the system. Hidden Markov models have proven to be useful tools for statistical signal and image processing. In this paper, wavelet-domain multiscale hidden Markov tree HMT model is employed to denoise LIF image in order to acquire the process intermediate values of chemical mechanical polishing (CMP). This method utilizes the relativity of wavelet coefficients well. Firstly, the statistical characteristics of wavelet coefficients and transform are summarized. Then the dependencies between the wavelet coefficients are modeled as dependencies between the hidden states with the mixture Gaussian model and HMT model. Because the degree of coefficients shrinkage is determined based not only on the size of the coefficients but also on its relationship with its neighbors across scale, HMT-based denoising typically outperforms the standard threshold techniques. Finally, the experiments show that the denoising results of LIF image using this method are typically better than other standard method. The method can not only get rid of the noise preferably, but also increase the power signal-to-noise ratio (PSNR)