识别不同水稻株型的高光谱模式方法的建立

提出了一种用高光谱技术快速识别不同水稻株型的新方法。首先在试验田内选择33个不同的水稻品种，测定了每个品种的14个株型特征参数，并采用荷兰Avantes公司的AvaSpec-2048便携式光谱仪采集不同株型水稻的高光谱数据。通过聚类分析，将所有水稻品种分为差异较大的3个株型类别。再采用平均平滑法和标准归一化方法对光谱数据进行预处理，对光谱数据主成分分析并获得各主成分数据。最后将主成分数据作为BP神经网络的输入变量，株型类别作为输出变量，建立了三层人工神经网络识别模型，并用模型对预测样本进行预测。结果表明，预测准确率为100%。该方法实现了对不同水稻株型的快速、无损识别。

ESTABLISHMENT OF PATTERN RECOGNITION METHOD OR DIFFERENT RICE TYPE BASED ON HYPERSPECTRAL DATA

Fourteen plant types’parameters and hyperspectral data of 33 different rice varieties in the field were measured by traditional field methods and Netherlands Avantes’s AvaSpec-2048 portable spectrometers， respectively. Based on cluster analysis method,all rice varieties were divided into three different plant type categories. Then, hyperspectral data were processed by average smoothing method, the standard normalization method and principal component analysis, and all the main components were obtained. Finally, using principal component data as the BP network input variables, plant type category as output variables, the identification model of three-layer artificial neural network was established, and unknown samples were predicted by BP models. The results showed that prediction the accuracy rate was 100%, and the method could discriminate different plant type of rice rapidly and non-destructively.