结合Tsallis熵的各向异性扩散模型  

Local Tsallis entropy combined anisotropic diffusion model

作　　者：彭扬[1] 何传江[1] 任泽民[1,2] 邹黎敏[1] 

机构地区：[1]重庆大学数学与统计学院,重庆401331 [2]重庆科技学院数理学院,重庆401331

出　　处：《计算机工程与设计》2014年第1期218-222,292,共6页Computer Engineering and Design

基　　金：国家自然科学基金项目(11201510)

摘　　要：为了在有效去除图像噪声的同时,保留更多的图像细节、纹理和弱边缘特征,在Perona-Malik各向异性扩散模型(P-M模型)的基础上,考虑到图像Tsallis熵在平滑区域和边缘处熵值有差异的特点,提出了结合图像局部Tsallis熵的各向异性扩散模型。该模型的扩散系数同时依赖于图像梯度和图像局部Tsallis熵,较好的克服了P-M模型在图像部分边缘和细节失真的问题。实验结果表明,该模型不仅能很好的保持图像的弱边缘和重要细节,而且能有效的去除噪声。To effectively remove the noise while retaining more detail, texture and weak edge features of image, by using the characteristics that the image Tsallis entropy of smooth area and edge are different, an anisotropic diffusion model is proposed combined with local Tsallis entropy based on the Perona-Malik anisotropic diffusion model （P-M model）. The diffusion coeffi cient of this model candepends on the image gradient and the local Tsallis entropy. This model overcomes the image distortion problem of P-M model. Experimental results show that the proposed model can not only maintain weak edge and important details of the image very well, but also effectively remove noise.

关 键 词：Perona-Malik模型 各向异性扩散 图像去噪 TSALLIS熵 偏微分方程 

分 类 号：TN911.73[电子电信—通信与信息系统]