基于SVM的储粮害虫图像识别分类

Image Recognition and Classification of the Stored-grain Pests Based on Support Vector Machine

粮虫图像识别属于小样本、参数多和特征之间混合度大的分类问题,因此分类器的设计是自动检测系统的关键环节。为此,采用网格搜索法,以SVM交叉验证训练模型的识别率为判别准则,对支持向量机分类器的参数和进行优化。应用SVM分类器对粮仓中危害严重的9类粮虫进行了自动分类,识别率达到93%以上。结果证实了基于SVM的分类器可进一步提高粮虫识别的精度。

The image recognition of the stored - grain pests is few - sample, multi - feature and multi - compound degree of various pests. The design of the classifier is a very important part for the stored - grain pests＇ detection system. The grid search was proposed, to optimize parameters and in the classifier based on support vector machine, by the judgment rule that is the recognition ratio of the cross - valid train model. The nine categories of the stored - grain pests in grain - depot were automatically recognized by the classifier based on support vector machine, the correct identification ratio was over 93 %. The experiment showed that it was practical and feasible.