关键词: 小波阈值, 贝叶斯统计模型, 上下文法模型, 非高斯二元分布

Abstract: A new spatial adaptive wavelet threshold denoising method was presented, which was based on a non-Gaussian bivariate distribution and context model for image denoising inspired by image coding. The dependency between coefficients and their parents was carefully studied and a new distribution model composed of two variables and a free parameter was proposed. Context model is the core method in image coding and is applied in this project to choose the spatial adaptive threshold derived in a Bayesian framework. Experiment results show that this new method outperforms the best of the recently published methods, such as SureShrink, Wiener2, and BayesShrink.

Key words: wavelet threshold, Bayes statistics, context model, non-Gaussian bivariate distribution