摘要
目前图像修复的基本方法主要有两类:纹理合成的方法和基于偏微分的方法。这两种算法都需要复杂的公式,较难理解和实现,为此提出了一种新的图像修复和去噪方法,该方法不需要先验理论。在定义修复区域像素点的优先度基础上提出了一种邻近像素点的图像修复算法;通过检验每一个像素点的被腐蚀程度提出了一种基于邻近像素点的保留图像对象边缘和细节的图像去噪新方法。算法将图像修复和去噪相结合,达到了更好的修复图像的目的。
There are two primary categories of the work that focus on image completion： texture synthesis algorithm and PDE-based method. All of there algorithms are difficult to comprehend and achieve because of complicated formulae, so a new method to image completion without apriorism is proposed. Algorithm of image inpainting based on adjacent pixels is proposed by defining the priority of the pixels belong to the areas being repaired. A new algorithm of remove noise based on reserve adjacent pixels＇ information of edge and detail pass the inspection of each pixel by corrosion is proposed. The algorithm combine image completion and remove, excellent results are achieved in experiments.
出处
《计算机工程与设计》
CSCD
北大核心
2009年第13期3139-3141,共3页
Computer Engineering and Design
关键词
方差
图像修复
图像去噪
噪声检测
像素优先权
varinace
image completion
remove noise
detect noise
pixel＇ s priority
