摘要
针对不同小麦病害有不同的形状特征,利用多重分形分析提取小麦病害图像的8个多重分形谱值作为小麦病害的形状特征参数,并利用这8个特征参数来索引图像数据库作为学习向量量化(LVQ)神经网络的输入,进行样本训练、分类识别。试验结果表明,该算法对小麦病害的识别率可达90.0%以上。
Different wheat diseases have different shape characteristics.Using multi-fractal analysis,eight multi-fractal spectrum values were extracted and used as shape characteristic parameter of wheat diseases,and then they were used to train learning vector quantization neural network.Experimental results showed that the recognition rate of the algorithm on wheat diseases could reach more than 90%.
出处
《湖北农业科学》
北大核心
2013年第7期1669-1671,1675,共4页
Hubei Agricultural Sciences
基金
河南省教育厅科学技术研究重点项目(12A510021)
河南省许昌市科技局计划项目(1101060)
关键词
小麦病害
多重分形谱
智能识别
LVQ神经网络
wheat diseases
multi-fractal spectrum
smart recognition
learning vector quantization neural network
