基于灰色系统理论的图像修复研究  

Research on Image Restoration Based on Grey System Theory

作　　者：曾卫波 邢永康[1] 石杨[1] 

机构地区：[1]重庆大学计算机学院,重庆400030

出　　处：《计算机工程》2014年第8期217-223,共7页Computer Engineering

基　　金：国家自然科学基金青年科学基金资助项目(60403009);中央高校基本科研业务费专项基金资助项目(CDJXS12180008;CDJZR10180021)

摘　　要：针对结构图像修复中修复质量和时间难以平衡的问题,提出一种基于灰色理论和蚕食控制的快速修复方法。该方法以待修复区域的已知邻域信息为基础,采用基于均值生成时序的单因子系统云灰色预测模型SCGMmv(1,1),挖掘不同方向灰度序列的变化趋势,对单个未知像素点的像素值进行预测,并在修复过程中模仿蚕食桑叶对填充顺序进行控制,在保证边缘结构修复质量基础上,避免了复杂的修复优先级计算。实验结果显示,该算法不仅能有效延伸边缘和平滑区域的拓扑结构,而且能满足实时性要求,获得了较好的视觉效果。For structure image restoration,a fast method based on gray theory and nibble control is proposed.It adopts a single-factor cloud grey forecasting model based on mean-value generating time series to mine the tendency of grayscale sequences in different direction and to predict the grayscale value of single unknown pixel using the known information around the area to be restored.The restoration process is the same as the silkworms eat mulberry leaves.It replaces the complex priority calculation and guarantees the quality of edge structure restoration.Experimental results show that the algorithm not only extends topology of edge and smooth area,but also meets the real-time requirements and a good visual effect.

关 键 词：图像修复 灰色预测 修复过程 偏微分方程 SCGMmv(1 1)模型 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]