CAJ | 학술논문

图像小波系数存在很大的相关性.该文考虑到小波系数层间相关性,引入双变量概率分布模型.基于贝叶斯估计理论,得到了相应的非线性阈值函数(双变量收缩函数).基于小波系数层内相关性,利用双变量收缩作者提出了局域自适应收缩去噪算法.在实验中,将该文去噪结果与Donoho's hard thresholding、BayesShrink和HMT作了比较.实验结果显示该文算法能获得更好的去噪效果.
도상소파계수존재흔대적상관성.해문고필도소파계수층간상관성,인입쌍변량개솔분포모형.기우패협사고계이론,득도료상응적비선성역치함수(쌍변량수축함수).기우소파계수층내상관성,이용쌍변량수축작자제출료국역자괄응수축거조산법.재실험중,장해문거조결과여Donoho's hard thresholding、BayesShrink화HMT작료비교.실험결과현시해문산법능획득경호적거조효과.
There are strong dependencies between wavelet coefficients of images. In this paper, considering inter- scale dependency, we introduced a bivariate probability distribution model. The corresponding nonlinear threshold functions (bivariate shrinkage function) are derived from the model using the Bayesian estimation theory. And then, based on the intra - scale dependency we present a locally adaptive denoising algorithm using the bivariate shrinkage function. We compare this algorithm with the Donoho's hard thresholding, BayesShrink and HMT. Experimental results show this algorithm can receive better denoising results.