摘要
提出了一种基于模糊聚类的黄瓜病害图像自动分割方法,模糊聚类与差分进化算法相结合,在进化过程中根据聚类中心对应的阈值确定模糊聚类中心的个数,由差分进化适应度函数值确定聚类中心是否被选中,以此实现图像的模糊聚类自动分割。经黄瓜炭疽病叶图像、白粉病叶图像、灰霉病叶图像和霜霉病叶图像的实验测试,该方法可以实现无人干预情况下的黄瓜病害图像自动分割,与相同类别个数的FCM算法相比,表现出了更好的性能。
With the combination of fuzzy clustering and differential evolution algorithm, the number of cluster centers was determined according to threshold which responded to cluster center during evolution. The clustering centers were determined whether selected or not by the fitness function values of differential evolution, which realized the automatic segmentation based on fuzzy clustering. After experimental testing with cucumber anthrac- nose leaf image, powdery mildew leaf image, botrytis leaf image and downy mildew leaf image, it was proved that our approach can segment cucumber diseases images automatically without human intervention. Compared with results from FCM with the same amounts of clustering classes, our results show better performance.
出处
《中国农机化学报》
2015年第3期123-126,131,共5页
Journal of Chinese Agricultural Mechanization
基金
天津市高等学校科技发展基金计划项目(20120811)
国家星火科技计划项目(2011GA610012)
关键词
自动图像分割
黄瓜病害
模糊聚类
差分进化
automatic image segmentation
cucumber disease
fuzzy clustering
differential evolution
