基于Criminisi的图像修复算法的设计与实现  

Design and Implementation of Image Inpainting Algorithms Based on Criminisi

作　　者：徐浩栋 XU Haodong(Nanjing Xiaozhuang University,Nanjing 211100,China)

机构地区：[1]南京晓庄学院,江苏南京211100

出　　处：《中国有线电视》2023年第5期35-38,共4页China Digital Cable TV

摘　　要：图像修复是数字图像处理领域的重要研究方向之一。提出一种基于Criminisi的电视图像修复算法,该算法利用邻域信息和纹理连续性的特点,通过多次迭代和纹理合成实现了图像的高效修复。通过实验和分析,该算法在处理较复杂的缺陷类型时表现出更加优异的修复效果,具有广泛的应用前景和研究价值。Image inpainting is one of the important research directions in the field of digital image processing.A TV image inpainting algorithm based on Criminisi is proposed,which utilizes the characteristics of neighborhood information and texture continuity,and achieves efficient image inpainting through multiple iterations and texture synthesis.Through experiments and analysis,the algorithm shows a better repair effect when dealing with more complex defect types,and has broad application prospects and research value.

关 键 词：图像修复 Criminisi 邻域信息 纹理连续性 多次迭代 纹理合成 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TN948.6[自动化与计算机技术—计算机科学与技术]