贝叶斯分类器在马铃薯外品质检测中的应用

Application of bayes classifier in detecting potato external quality defects

为了丰富马铃薯外品质的建模方法,结合机器视觉和贝叶斯分类器建立马铃薯外品质的预测模型。先通过机器视觉技术采集虫眼、合格、机械损伤、绿皮、鼠咬5类马铃薯图像,再对马铃薯的灰度图像和彩色图像提取两类3种特征数据,同时结合贝叶斯分类器建立不同马铃薯外品质的预测模型。研究结果表明:利用马铃薯真彩色图像R、G、B三通道获取特征数据,并对其特征数据采用滤波模板进行平滑处理,建立马铃薯贝叶斯分类器预测模型,可以有效提高各类马铃薯的识别率。

To enrich modeling methods of potato external quality, a method was proposed based on machine vision and hayes classifier. The fundamental principle was that extracting of color image from 5 kinds of potato by the machine vision technology included the bugeaten, the mechanical damage, qualified, the green skin, the mouse nips. And then 3 kind of characteristic data were picked up from potato＇s gray image and color image. Finally, the modeling of potato external quality was set up by using characteristic data and bayes classifier. Experimental results showed that it had good recognition effect of potato external quality modeling with gaining characteristic data from R, G, B component images and using the filter template to smooth data.