基于遗传神经网络的杂草与作物识别方法的研究

Investigation on the Recognition of Weed and Crops Based on Genetic Neural Network

综合运用计算机数字图像处理技术与人工神经网络技术,建立了一个多层前馈神经网络,实现了大豆与杂草的识别。首先,通过计算机视觉技术采集大豆与杂草的叶片图像作为测试样本,之后应用遗传神经网络建立对杂草图像的识别模型。研究结果表明,这种图像处理技术与神经网络的结合的方法,识别作物区域的准确率可达90%以上。

A multi-layer feed forward neural network used computer digital image processing and artificial neural network was estab- lished, which could identify the area of soybean leaf and weed. Firstly, the image of soybean and weed was obtained by the technique of visual computer as samples, and then recognize the weed＇ area through neural network. The results showed that the method of integrat- ing of image processing and neural network, the exact rate of identifying the corps in the region over 90 percent.