摘要
根据尺度间小波系数的相关性和方差是双变量分布模型参数的理论,提出了应用基于上下文模型的空间自适应方法估计方差,并用双变量收缩法进行图像去噪的新方法.将新方法与仅使用待估计点与它的方形邻域窗来估计方差的双变量阈值去噪方法进行了比较.结果表明,用新方法去噪时图像的P SN R值与视觉效果都有提高和改善,新去噪方法具有理论上的一致性.
Based on the theory of inter-scale correlation of wavelet coefficients and the variance being a parameter of the bivariate distribution model,a bivariate shrinkage method is proposed for image denoising which estimates the variance by context modeling for spatial adaptivity(CMSA).Compared new method with the bivariate threshold denoising method which estimates of variance only by neighbor coefficients,experiment results show that the new method has good performance in both PSNR value and visual quality for image denoising.And the new denoising method is consistent in theory.
出处
《中北大学学报(自然科学版)》
CAS
北大核心
2010年第6期651-654,共4页
Journal of North University of China(Natural Science Edition)
关键词
双变量模型
方差
空间自适应上下文模型
双树复小波变换
bivariate model
variance
context modeling for spatial adaptivity(CMSA)
dual tree complex wavelet transform(DTCWT)