摘要
综合运用计算机数字图像处理技术与人工神经网络技术,对叶片色度值进行了计算;同时,建立一个多层BP神经网络,实现了对大豆叶片中病斑的自动识别,运用区域标记法对病斑的特征参数进行计算。实验证明,该方法能有效地识别出病斑区域,识别率可达92.1%,计算的特征参数为病种的识别提供了理论依据。
This paper calculates the value of leaf color and establishes a multi - layer BP neural network, using computer digital image processing and artificial neural network, which can identify the area of diseased spots of soybean laminae. Meanwhile, calculates the feature parameters using the technology of signature. The experiment improves that the diseased spots could be recognized correctly and the accuracy could reach to 92.1%. The feature parameters may provide theoretical foundation for the recognition of the category of diseases.
出处
《农机化研究》
北大核心
2009年第6期41-43,共3页
Journal of Agricultural Mechanization Research
关键词
病虫害
图像处理
色度值
病斑
神经网络
damage by disease and insect
image processing
value of leaf color
diseased spot
neural network
