摘要： 为检测彩色套印图像是否存在偏差及其偏差参数，提出基于边缘共生条件概率矩阵(ECM)的彩色套印偏差检测算法。该算法检测彩色套印图像的边缘信息，在边缘图像中计算ECM，并提取ECM中的极值点。若存在非(0,0)点处的极值点，则原图中存在套印偏差且极值为其偏差参数；否则，原图中不存在套印偏差。实验表明，该算法的检测结果可以精确到像素级。

关键词: 边缘图像, 扫描向量, 边缘共生条件概率矩阵, 彩色套印偏差, 偏差参数

Abstract: In order to detect whether color printing registration image exists deviation and deviation parameter, this paper presents a detection algorithm of color printing deviation based on Edge Co-occurrence conditional probability Matrix(ECM). It detects the edge information of color printing registration image, calculates ECM of edge image, and extracts the peak value of ECM. If peak value exists except (0,0) point, the color printing registration image exists deviation and the peak value is the deviation parameter; otherwise, the color printing registration image does not exist deviation. Experiments demonstrate that the detection result can accurate to one pixel, and the algorithm is effective and accurate.

Key words: edge image, scanning vector, Edge Co-occurrence conditional probability Matrix(ECM), color printing registration deviation, deviation parameter

中图分类号: 

• TP391.41