摘要
提出一种遗传算法(GA)优化支持向量机(SVM)的方法,用于印刷品套准状态识别。对采集到的印刷标识图像进行纹理分析,提取灰度共生矩阵的惯性矩、能量、相关性和熵等4个特征参数,计算各参数在0°,45°,90°和135°方向上特征值的均值和方差,生成8维纹理特征数据;建立以径向基为核函数的SVM模型,采用GA对SVM的惩罚参数与核函数参数进行寻优选取,完成识别。结果表明:GA-SVM模型可有效识别印刷套准,满足实时性要求。
A genetic algorithm(GA) optimization support vector machine(SVM) method is proposed for printing registration state recognition.Texture analysis on collected printing marks images is carried out and the moment of inertia,energy,correlation,and entropy of gray level co-occurrence matrix are extracted. The mean and variance values of gray level co-occurrence matrix are calculated and the parameters are calculated to generate 8-dimensional texture feature data. SVM model using radial basis Kernel function is established. GA is used to optimize and select the penalty factor and kernel function parameters of SVM,thus complete the recognition.Results show that the GA-SVM model can effectively recognize the printing registration and meet the real-time requirements.
作者
王世辉
王仪明
武淑琴
焦琳青
李林会
WANG Shi-hui;WANG Yi-ming;WU Shu-qin;JIAO Lin-qing;LI Lin-hui(Beijing Key Laboratory of Digital Printing Machinery,Beijing Institute of Graphic Communication,Beijing 102600,China)
出处
《传感器与微系统》
CSCD
2018年第11期142-144,共3页
Transducer and Microsystem Technologies
基金
北京市教委科技计划重点项目暨北京市自然科学基金重点资助项目B类(KZ201510015016)
关键词
遗传算法
支持向量机
印刷
套准识别
纹理分析
genetic algorithm
support vector machine (SVM)
printing
registration recognition
texture analysis