摘要
利用便携式光谱仪采集水稻和杂草稻叶片的光谱信息,光谱的波段范围是350～2 500nm。同时,利用求一阶导数来对原始光谱数据进行预处理,对预处理后的光谱数据进行主成分分析,优选出4个波段点:525,722,1 392,1 882 nm。在每个波段点处选择10个特征波段,组成40×48的矩阵作为神经网络的输入模型。利用Matlab建立BP神经网络的识别模型进行训练,对于训练好的模型进行仿真以及模型检验,其识别精度可以达到90%以上。
In this paper,the portable spectrometer was used to collect the spectral information of the rice and weedy rice.The range of the spectral band is 350-2500nm.After the analysis,930-969nm band of the derivative spectra were chose as the useful information.Then use the Matlab to establish a BP neural network.Choose 48 samples of rice and weedy rice to train the BP-NN model.After the simulation of the BP-NN model,two batch of samples was used to test the model.The overall classification can reach 90%.This research demonstrated that the artificial neural network and the spectral information can be used to identify the weedy rice from the rice.
出处
《农机化研究》
北大核心
2013年第1期156-158,163,共4页
Journal of Agricultural Mechanization Research
基金
镇江市农业科技支撑计划项目(NY2011002)
江苏省自然科技基金项目(BK2009200)
江苏省普通高校科研创新计划项目(1221200026)
关键词
杂草稻
光谱
人工神经网络
识别
weedy rice
spectroscopy
artificial neural network
identification
