基于Android的苹果叶部病害识别系统设计  

Design of apple leaf disease recognition system based on Android

作　　者：屈赟[1] 陶晡[2] 王政嘉[1] 王树桐[2] 

机构地区：[1]河北农业大学教务处,河北保定071001 [2]河北农业大学植物保护学院,河北保定071001

出　　处：《河北农业大学学报》2015年第6期102-106,共5页Journal of Hebei Agricultural University

基　　金：河北农业大学青年基金项目(QN201238);河北省教育厅项目(SZ131019)

摘　　要：为快速便捷地解决苹果病害识别问题,本研究设计了基于Android的图像识别系统。采用最大类间方差法(Otsu)对病斑图形分割,提取了病斑的颜色特征,纹理特征和形状特征、运用支持向量机(SVM)对病斑进行了分类,并在服务器端建立了苹果叶部病害特征库。手机客户端采集5种苹果病害图像,上传到服务器端进行识别,并将识别结果反馈给客户端,平均正确识别率为85.33%,测试效果良好。In order to quickly and easily solve the identify problem of apple diseases, the image recognition system based on Android was designed in this study. Using Otsu method （Otsu）, we can divide the lesion graphics, extract color characteristics, texture, and shape features of lesion, classify the lesion by support vector machine （SVM）, and establish apple leaf disease feature library in the server terminal. Mobile client acquires five kinds of disease image, up- loads it to the server terminal. The terminal will recognize it and put the results back to the client. The average correctidentification rate is 85.33%, and the testing result is good.

关 键 词：ANDROID 苹果病害 图像识别 CANNY算子 支持向量机 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]