摘要
对包含结构和纹理信息的缺损图像进行了修复。记录与已知像素点像素值的差值小于某一阈值的像素点,采用曲线拟合方法将这些像素点连接成平滑的曲线,并且将此曲线的延伸方向作为等值线上延伸的方向,采用由外向内并且边缘优先为最高优先等级的优先程度逐层填充像素值的方法,将空间连续性方向延伸和图像润饰的思想相结合,使修复区域逐步愈合。该算法能保持图像边缘,结构填充合理。
To restore the degrade images including structure and complex texture, the arithmetic is built. Firstly, when the difference of the current pixel value and the known pixel value is less than a definite value, sign them, and link these pixels by curve fitting, so the direction of the curve extended is the space succession direction. Then from out to inner with edge priority and prior degree sequence, calculate pixel values. Combine this method with the interrelated idea of image retouching, fill in the missing region one by one. The algorithm can preserve image edges, make image structure more logical.
出处
《计算机工程与设计》
CSCD
北大核心
2009年第4期925-927,937,共4页
Computer Engineering and Design
关键词
阈值
曲线拟合
边缘优先
空间连续性
图像润饰
图像边缘
definite value
curve fitting
edge priority
space succession
image retouching
image edge
