摘要
针对人类视觉和颜色分类检测问题,提出一种映射色差的颜色距离度量新定义.该映射色差满足距离度量的3个公理条件,并且以两幅图像均值向量之间的L2距离为下界;以映射色差为基础设计了一种表面颜色分类检测新方法,在CIE 1976L*a*b*空间用距离加权重心切分提取主色调,并以上述映射色差距离设计最小距离分类器实现表面检测.实验中对人民币和瓷砖产品表面质量进行自动检测,其检测结果和人的视觉感知特性取得了良好的一致性.该检测方法已应用于人民币印刷质量表面检测系统中.
This paper defines a new measure of color difference, mapping chromatism, aiming to solve color clustering problems. The measure satisfies three conditions of distance axiom. Based on the measure lower bounded by L2 distance between one inspected image and the reference image, a new surface inspection of color clustering is proposed. The inspection extracts dominant colors using radius weighted method in CIE 1976 L ^＊ a ^＊ b^＊ space and clusters the dominant colors to realize surface inspection by minimized distance principle based on mapping chromatism. Experiments performed auto inspection to RMB bills and tiles of surface quality and obtained good results consistent with HVS.
出处
《北京理工大学学报》
EI
CAS
CSCD
北大核心
2010年第1期74-78,共5页
Transactions of Beijing Institute of Technology
基金
中国人民银行印钞造币总公司科研项目(2001L26-g4)
关键词
表面检测
主色调
映射色差
距离加权重心
surface inspection
dominant color
mapping chromatism
radius weighted barycenter
