基于支持向量机的水稻稻瘟病图像分割研究

水稻稻瘟病图像的分割是水稻稻瘟病自动分析与识别的关键环节,其分割效果直接影响后续处理。提出一种基于支持向量机的水稻稻瘟病病害彩色图像分割方法。首先选取叶子正常部分的像素点以及颜色相对复杂的病斑像素点作为负训练样本和正训练样本,提取像素R、G、B彩色分量作为特征向量,对支持向量机进行训练,然后在RGB空间利用训练好的支持向量机对待分割图像的所有像素点进行分类,实现水稻稻瘟病彩色图像的分割。为了获得最佳的分割效果,采用网格搜索法对径向基核函数下的不同核参数分割效果和性能进行比较与分析,确定最佳模型参数。利用此模型进行水稻稻瘟病图像分割实验,获得较好的分割精度,结果优于最大类间方差分割算法。

Study on image segmentation of rice blast based on support vector machines

The exact image segmentation of rice blast is a key to analyzing and recognizing rice disease automatically.A novel color segmentation algorithm based on Support Vector Machines(SVM) was proposed.The issue of image segmentation is translated into the issue of classification on RGB space.First the disease part pixels and normal part pixels was selected and used to make the train samples.The sample's values of RGB was used to make the characteristic vectors.The SVM was trained by these samples and was tested by other pixels in the image.In order to acquire the best segmentation result,different classification kernel parameters were compared and analyzed.Finally,the color image was segmented with the trained SVM model.The experimental results show that the accuracy based on this SVM model was better than OTSU.