基于边缘与非边缘分开处理的印刷品缺陷检测

摘要

针对印刷品缺陷检测问题，通过对经典缺陷识别算法的研究，提出了一种边缘点与非边缘点分开处理的印刷品检测缺陷识别算法。算法以统计过程原理为出发点，根据边缘点和非边缘点不同的像素特征，对非边缘点采用统计阈值法处理，对边缘点采用邻域平均值与梯度值相结合的算法处理。该算法经过实验证明，能够有效地剔除伪疵点，提取出真正的缺陷。%The classic defects recognition algorithm is studied with an aim at printing defects detection , and this paper puts forward a recog-nition algorithm that processes edge and non-edge points separately in print defects detection .The algorithm is based on statistical process princi-ple according to the characteristic difference of pixel edge points and non -edge points; the statistical threshold method is used for non-edge points, and the combination of field average value and gradient value is used for edge points .The algorithm was proved by experiment and it can effectively eliminate the false defects in the process of extracting the real defects .