摘要
提出了利用支持向量机(SVM)分类的方法对采集图像进行识别。采用计算机图像处理技术针对棉花苗期杂草图像进行分割,提取棉花与杂草的形状特征参数;选取最有效的特征数据组合输入SVM进行分类学习训练,实现杂草的有效识别。结果表明,使用该方法获得的图像识别效率较高,在同等条件下,速度优于人工神经网络。
It was put forward that by using support vector machine（SVM）classification method to identify the image.Computer image processing technology was adopted to segment images of seedling cotton and weeds and extract shape parameters of cotton and weeds.The most effective combination of characteristics data were selected to import SVM to carry classification learning and training,and the weeds were identified effectively.Experimental results showed that the proposed method used to identify weeds was more efficient,and the rate was better than artificial neural network in the same conditions.
出处
《安徽农业科学》
CAS
北大核心
2010年第26期14756-14757,共2页
Journal of Anhui Agricultural Sciences
关键词
颜色特征
形状特征
RTS不变性
SVM
图像识别
Color characteristic； Shape characteristic； RTS invariance； SVM； Image recognition；
