基于机器视觉的番茄脐腐病的鉴别  

Tomato blossom-end rot identification based on machine vision

作　　者：孟祥佳[1] 杨断利[1] 籍颖[1] 李文志[1] MENG Xiangjia;YANG Duanli;JI Ying;LI Wenzhi(Graduate School,Agricultural University of Hebei,Baoding 071000,China)

机构地区：[1]河北农业大学研究生学院,河北保定071000

出　　处：《现代电子技术》2017年第6期100-102,108,共4页Modern Electronics Technique

基　　金：河北省科技厅自筹项目(13227431)

摘　　要：利用计算机视觉技术,通过图像处理和模式识别方法,对番茄脐腐病的自动识别进行研究。首先利用中值滤波法对病症图片进行处理,去除噪声,其次采用Otsu算法对病斑进行分割。通过提取病斑区域的颜色特征和形状特征,同时采用贝叶斯判别方法和支持向量机方法实现特征参数的提取。实验结果表明,贝叶斯判别方法对训练样本和实验样本的判别准确率分别达到100%和92%,高于支持向量机方法,实现了番茄脐腐病的准确识别。The computer vision technology,image processing and pattern recognition method are used to study the automate identification of the tomato blossom?end rot.The median filtering method is adopted to process the disease image and eliminate the noise.The Otsu algorithm is employed to segment the scab.The Bayesian discriminate method and support vector machine method are used to extract the feature parameters respectively according to the color feature and shape feature extracted in the scab area.The experimental results show that the discriminate accuracy of the former(Bayesian discriminate method)to the training sample and experiment sample can reach up to100%and92%,which is higher than that of the latter(support vector machine method),and can recognize the tomato blossom?end rot accurately.

关 键 词：机器视觉 番茄脐腐病 特征提取 模式识别 

分 类 号：TN911-34[电子电信—通信与信息系统] TP391.4[电子电信—信息与通信工程]