论文部分内容阅读
利用地面成像光谱辐射测量系统(Field Imaging Spectrometer System,FISS)获取了5种玉米籽粒的成像光谱数据,经反射率反演、噪声去除和一阶微分处理后,运用Wilk-lambda逐步判别法进行波段选择并建立判别模型。交叉验证结果表明:(1)平均识别精度为91.6%,除了高油115的识别精度仅有87%外,其他品种的识别精度均在90%以上;(2)在分类方案、光谱波段数、每类样本数量不变的情况下,分类精度受类别数和类别间可分性的影响;(3)随着入选波段数的增加模型识别精度逐步提高。因此成像光谱技术在玉米品种的识别以及质量检测方面具有重要的应用前景。
The imaging spectral data of five kinds of maize kernels were obtained using the Field Imaging Spectrometer System (FISS). After reflectivity inversion, noise removal and first-order differential processing, Wilk-lambda step discrimination Select and establish a discriminant model. The results of cross-validation showed that: (1) The average recognition accuracy was 91.6%, except that the identification accuracy of Gaoyou 115 was only 87%, the recognition accuracy of other varieties was over 90%; (2) In the classification scheme, spectral bands, The accuracy of classification is affected by the number of categories and the separability between categories under the condition of constant number of samples. (3) The accuracy of model recognition is gradually increased with the increase of the number of selected bands. Therefore, imaging spectroscopy has important application prospects in the identification of corn varieties and quality testing.