摘要
划痕是旧视频中常见的问题,如何快速、准确的检测出划痕对它的进一步修复尤为重要.通过分析现有垂直划痕检测算法的优缺点,针对检测中存在的问题,提出了一种基于混合模型的视频划痕检测算法,将Sobel算子、形态学闭运算、广义模型检测等算法有效的结合起来对划痕图像进行检测,实验证明,这个算法对各类划痕的检测都有效,并解决了如何区分划痕与垂直的自然线条的问题,避免了误检,相对于几个典型的算法,该算法的准确率更高并且有更强的鲁棒性.
Scratch is a common problem in old video. How to detect scratches quickly and accurately is especially important for its further repair. This paper reviews the existing algorithms for vertical line scratches detection.In order to make a further research on vertical line scratches detection,we take advantage of them and propose a new method.To be specific,we use the Sobel operator,the closing operator,and the generalized model to detect the scratches.And solved the problem of distinguishing scratches from vertical natural lines.This method is more effective for the degrade film which has many alone scratches. The experimental results show that the improved method can detect the scratches more effectively.
出处
《吉林化工学院学报》
CAS
2017年第11期63-67,共5页
Journal of Jilin Institute of Chemical Technology
关键词
信息处理技术
划痕检测
混合模型
SOBEL算子
Information processing technology
scratch detection
synthesis algorithm
Sobel operator
