基于支持向量机的玉米叶部病害识别

Corn leaf disease recognition based on suport vector machine method

针对玉米叶部病害图像的特点,提出将支持向量机(SVM)组成的多分类器应用于多种玉米叶部病害识别中。首先利用L ive-W are分割算法分割出玉米叶部病灶,再利用小波特征提取算法提取病灶的特征向量,最后利用支持向量机分类方法进行病害的识别。玉米叶部病害图像识别试验结果表明,支持向量机分类方法适合小样本情况,具有良好的分类能力,适合多种玉米叶部病害的分类。不同的分类核函数的相互比较分析表明,径向基核函数最适合玉米病害的分类识别。

In view of corn leaf disease image characteristics, one multi-classltlcatlon machine is applied in corn Leaf disease recognition. First the algorithm of live-ware segmentation was used to find disease part and the algorithm of the wavelet feature extraction was used to make the corn disease leaf the characteristic vectors, then the support vector machine classification method was applied to recognize the disease. The corn leaf disease image recognition experiment indicates that Support Vector Machine classification method suits the small sample situation and has the better classification ability. The method suits corn leaf disease classification. The different classification kernel functions are compared, and analysis shows that the radial base function most suits the corn leaf disease classification recognition.