基于米字型各向异性扩散的破损藏式古唐卡修复研究  

An Improved BSCB Image-Inpainting Model Based on UK-flag Shaped Anisotropic Diffusion

作　　者：翟东海[1,2] 李帆[2] 鱼江[1] 

机构地区：[1]西南交通大学信息科学与技术学院,四川成都610031 [2]西藏大学工学院,西藏拉萨850000

出　　处：《西南科技大学学报》2014年第2期77-81,91,共6页Journal of Southwest University of Science and Technology

基　　金：国家社会科学基金项目(12EF119);国家级大学生创新创业训练计划项目(201210694019)

摘　　要：在利用原始的BSCB模型修复图像时,其扩散过程会造成图像的等照度线有一定程度上的交叉,使得修复后的图像视觉效果不太理想。针对以上问题,通过对原始的PM各向异性扩散模型进行改进,提出了米字型各向异性扩散模型,然后将其应用到BSCB模型的扩散过程中,提出了改进的BSCB图像修复模型。最后,利用文中提出的修复模型和其他的改进的BSCB模型对破损古唐卡图像进行修复。结果表明,和原始的BSCB模型以及其他改进的BSCB模型相比较,提出的模型在没有增加算法的时间复杂度的情况下,在视觉效果上和均方误差(MeanSquareError,MSE)的评价指标上表现都很突出。When the original BSCB model was used to inpaint images,the isophotes of the images may overlap to some extent in the diffusion process,and its visual effect is of poor visual satisfaction. To solve the above problem,the original PM anisotropic diffusion model was improved with a UK-flag shaped anisotropic diffusion model proposed. Then,UK-flag shaped anisotropic diffusion model was applied in the diffusion process of the BSCB model to propose an improved BSCB model. Finally,the proposed inpainting model and other improved BSCB models were applied to inpaint the damaged ancient Thangka. The experimental results show that the visual effect and the Mean Square Error（ MSE） of the inpainted image are outstanding without increasing the time complexity by using the proposed model in comparison with the original BSCB model and other improved BSCB models.

关 键 词：图像修复 BSCB模型 PM各向异性扩散模型 米字型各向异性扩散 视觉效果 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]