CAJ | 학술논문

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。
침대랄초협부상견병반도상존재배경복잡차불역분할적문제，대전통적 C －V 모형진행개진，근거랄초병반도상적특정，수선대병반도상적 RGB 안색분량화통도권중계수진행가권처리，접착용차분도상능량진행계산처리，최후，통과수평집연화방정획득병반적분할곡선。실험측시표명，경과개진적 C －V 모형대랄초병반도상분할적효과비전통 C －V 모형급 K －means 방법분할적효과화항조성호。
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.