基于改进C-V模型的辣椒病斑图像分割  

Segmentation of Capsicum Disease Spot Image Using Improved C-V Model

作　　者：王建玺[1] 王启明[2] 

机构地区：[1]平顶山学院软件学院,河南平顶山467000 [2]平顶山学院计算机科学与技术学院,河南平顶山467000

出　　处：《山东农业科学》2015年第10期116-118,124,共4页Shandong Agricultural Sciences

基　　金：河南省科技厅科技发展计划项目(142102210226)

摘　　要：针对辣椒叶部常见病斑图像存在背景复杂且不易分割的问题,对传统的C-V模型进行改进,根据辣椒病斑图像的特征,首先对病斑图像的RGB颜色分量和通道权重系数进行加权处理,接着用差分图像能量进行计算处理,最后,通过水平集演化方程获得病斑的分割曲线。实验测试表明,经过改进的C-V模型对辣椒病斑图像分割的效果比传统C-V模型及K-means方法分割的效果和抗噪性好。In view of complex background and difficulty in segmentation of capsicum leaf disease spot images,the conventional C -V model was improved.According to the features of capsicum disease spot im-age,firstly,the RGB color components and channel weight coefficient of disease spot image were weighted pro-cessing;then they were calculated using difference image energy;finally,the disease spot segmentation curve was got through the level set evolution equation.The experimental results showed that the improved model had better segmentation effect and noise immunity than C -V model and K -means model on capsicum disease spot image.

关 键 词：辣椒病斑 图像分割 C-V模型 加权颜色 水平集 

分 类 号：S436.611.1[农业科学—农业昆虫与害虫防治]