基于光谱技术的土壤快速分类方法研究.pdfVIP

浙江大学学报( ) 36(3) : 282~ 286, 2010 Jour nal of Zhej ian g Un iversity (Agric1 Life Sci1 ) :10082 9209(2010)05 DOI: 10.3785/j. issn. 10082 9209. 2010. 03.007 1 2 3 3 3 王遵义, 金春华 , 刘飞 , 王艳艳 , 鲍一丹 ( 1., 315100; 2. , 315211; 3. , 310029) : 针对浙江省4 种典型土壤, 研究应用可见2 近红外 谱近红外 谱和中红外 谱3 个波段范围 进行土壤快速分类的方法. 在获取 谱信息的基础上, 采用不同 谱建模方法以提高检测精度, 简化分 析计算; 并分别采用主成分分析结合人工神经网络(PCA2ANN/BP) 偏最小二乘法(PLS) 和偏最小二 乘法结合人工神经网络(PLS2ANN) 3 种方法进行建模.结果表明: 中红外 谱波段对土壤分类的效果 不理想, 而可见2 近红外 谱近红外 谱波段均能较好地进行土壤分类; 在可见2 近红外波段, PLS2 ANN 模型对土壤的分类效果优于PCA2ANN/ BP 和PLS模型, 为土壤快速准确分类提供了一种简便可 行的方法. : 谱技术; 土壤; 分类; 人工神经网络; 主成分分析; 偏最小二乘法 : S 155; O657.33 : A 1 2 3 3 3 WANG Zun2yi , JIN Chun2hua , LIU Fei , WANG Yan2yan , BAO Y2i dan ( 1. Zhej iang Wangli University , Ningbo, Zhejiang 315100 , China; 2 . College of Lif e cience and Biotechnology,N ingbo University , N ingbo, Zhej iang 315211, China ; 3 . College of Biosystems Eng ineering and Food cience, Zhej iang University, H angzhou 310029 ,China ) Rap id discr imina tion of soil var iety b ased on sp ectr oscop ic t echn iques. Journal of Zhejiang University (Agric1 Life Sci1 ), 2010,36(3):2822286 Ab stract Three spectral regions were investigated for the discrimination of four main soil varieties in Zhejiang Province. Thethree spectral regions included the visible2near infrared (Vis2 NIR), near infrared (NIR), middle2 infrared (MIR) spectroscopy. After spectral collection, different calibration methods were applied to improve the detection precision and to reduce the computational complexity. Principal component analysis (PCA) combined with