基于灰度-梯度共生矩阵的大米加工精度的机器视觉检测方法  

A MACHINE VISION DETERMINACTION METHOD OF RICE MILLING DEGREE BASED ON GRAY-GRADIENT CO-OCCURRENCE MATRIX

作　　者：万鹏[1,2] 龙长江[1] 

机构地区：[1]华中农业大学工程技术学院 [2]吉林大学生物与农业工程学院,长春130022

出　　处：《粮食储藏》2010年第4期48-51,共4页Grain Storage

摘　　要：提出了基于灰度-梯度共生矩阵结合机器视觉技术检测大米加工精度的方法。使用实验碾米机制备不同加工精度的大米样品,设计大米加工精度机器视觉检测系统获取不同加工精度大米样品图像,再利用灰度-梯度共生矩阵提取大米图像的纹理特征,采用逐步判别分析法构建Fisher判别函数组对大米样品的加工精度进行检测。试验结果表明:该方法对4种不同加工精度大米样品检测的平均准确率达到94.00%。A detecrmination of the rice milling degree was proposed based on gray-gradient co-occurrence matrix with machine vision.The rice of samples different milling degree were prepared by using an experimental mill machine,and a machine vision system detecrmiting the milling degree of rice was designed to obtain the rice kernel image of the different milling degree,the texture features of the rice image were obtained using gray-gradient co-occurrence matrix,and the Fisher discriminant functions constructed with stepwise discriminant analysis were used to detect the milling degree of the rice samples.The test results show that the average accuracy rate of the different milling degree of 4 rice samples detected with the method is 94.00%.

关 键 词：大米 加工精度 机器视觉 灰度-梯度共生矩阵 FISHER判别 

分 类 号：TS212[轻工技术与工程—粮食、油脂及植物蛋白工程]