自适应扩散张量的各向异性图像去噪  

Diffusion tensors for anisotropic image denoising

作　　者：郑满满[1] 胡小兵[1] 郑申海[1] 

机构地区：[1]重庆大学数学与统计学院,重庆401331

出　　处：《计算机应用》2013年第A02期168-170,208,共4页journal of Computer Applications

基　　金：重庆市自然科学基金资助项目(CSPC,2005BB2197);重庆大学“211工程”三期创新人才培养计划建设基金资助项目(S-09110)

摘　　要：图像去噪过程中,为了在有效平滑噪声的同时较好地保护图像的边缘和细节,提出了基于扩散张量的自适应去噪模型。该模型设计了随梯度大小变化的边缘增强扩散张量,改进了结构张量。在图像质量评判标准中,提出了基于相关系数函数的最佳停止时间评判准则。实验结果表明,改进的模型优于传统各向异性扩散模型,且能很好地吻合评判准则。In the process of image denoising, in order to remove noise effecitively and preserve edges and key details, the adapative diffusion tensor denoising model was proposed. A new diffusion tensor of edge enhancement according to gradient value was designed and the structure tensor was improved in the model. The best stop time evaluation criteria based on correlation coefficient was proposed as well. The experimental results show that the improved model is superior to traditional anisotropie diffusion model, and can better coincide with the judge standard.

关 键 词：图像去噪 扩散张量 自适应 相关系数 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]