基于神经网络的大豆叶片病斑的识别与研究

综合运用计算机数字图像处理技术与人工神经网络技术,建立了一个多层BP神经网络,实现了大豆叶片中病斑的自动识别与特征计算。首先通过计算机视觉技术采集叶片图像。其次,采用BP神经网络完成了对病斑图像的识别。最后,运用数字图像处理技术完成了对病斑区域相关特征值的计算。实验证明,该方法能有效地识别出病斑区域,识别率可达100%。该研究为将来病种的识别提供了理论依据。

Investigation and Recognition on Diseased Spots of Soybean Laminae Based on Neural Network

A multi-layer BP neural network, using computer digital image processing and artificial neural network was established in this paper, which could identify the area of diseased spots of soybean laminae. Firstly, obtained the image of soybean laminae, and then recognized the diseased spots through neural network. At last, computed the feature parameter of diseased spots using the technology of digital image processing. The experiment showed that the diseased spots could be recognized correctly and the accuracy could reach to 100%.This research might provide theoretical foundation for the recognition of the category of diseases.