改进的CDD图像修复模型算法  

Improved CDD image inpainting model algorithm

作　　者：李建民[1] 汪琦 林振荣[1] 袁梁 LI Jian-min1 , WANG Qi1 , LIN Zhen-tong1 , YUAN Liang2(1. School of Information Engineering, Nanchang University, Nanchang 330031, China; 2. School of ZTE information,Xinyu University, Xinyu 338004, Chin)

机构地区：[1]南昌大学信息工程学院,江西南昌330031 [2]新余学院中兴通讯信息学院,江西新余338004

出　　处：《计算机工程与设计》2018年第8期2564-2569,共6页Computer Engineering and Design

基　　金：江西省工业支撑计划基金项目(20122BBE500049)

摘　　要：为解决CDD图像修补模型修复时间长、迭代复杂、边缘过度不自然的问题,提出一种改进的CDD图像修复模型算法。通过简化曲率项运算来减少修复算法的运行时间,引入像素点灰度差值比率和权重平滑函数,克服图像边缘处的扩散相互抵制的现象,有效解决修复后图像边缘模糊的问题。通过实验数据中峰值信噪比(PSNR)和修复时间的对比,表明改进的CDD模型算法比传统的CDD模型算法修复时间更短,边缘过渡更清晰,视觉效果更佳。To solve the problems of long repair time,complex iteration and unnatural edge transition in curvature driven diffusions（CDD）image inpainting model,an improved CDD image inpainting model algorithm was proposed.The proposed model was simplified by the curvature of operations to reduce the running time of the repair algorithm.The rate of pixel’s gray value and the weighted smoothing function were introduced to overcome the phenomenon that the edges of an image’s diffusion restrain each other.The problem of blurred edge was effectively solved through the above steps.By comparing the peak signal to noise ratio（PSNR）and the repair time,experimental results show that the repair time of the improved CDD model is shorter than that of the traditional CDD model,and it also has clearer image edge and better visual effects.

关 键 词：CDD模型 灰度差值比率 权重平滑函数 偏微分方程 峰值信噪比 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]