基于高光谱技术地不同霉变程度籼稻快速鉴别.docVIP

基于高光谱技术的不同霉变程度籼稻快速鉴别 龚中良1 郑立章1 文 韬1,( 李立君1 谢洁飞1 马 强3 （中南林业科技大学机电工程学院1，长沙 410004） （华南农业大学工程学院 南方农业机械与装备关键技术教育部重点实验室2，广州 510642） （中南林业科技大学理学院3，长沙 410004） 摘 要 为解决快速、无损鉴别籼稻霉变程度问题。该文利用高光谱技术采集正常、轻度霉变、中度霉变和重度霉变共4类籼稻样本的光谱数据，经不同预处理后，通过连续投影算法（SPA）（MLR-DA）竞争性自适应重加权PLS-DA）建立鉴别模型。研究结果表明，SG-SPA-MLR-DA和RAW-CARS-PLS-DA模型的验证集相关系数RP均大于0.950。其中，RAW-CARS-PLS-DA模型的预测效果最优，其验证集相关系数RP为0.969，均方根误差RMSEP为0.269，对未知籼稻样本的总体分类准确率为93.33%。该模型对不同霉变程度籼稻具有较强的鉴别能力，故该方法可为快速、无损鉴别籼稻霉变程度提供技术支持。 关键词 高光谱技术 预处理 霉变 籼稻 模型 鉴别 中图分类号：S12；S 文献标识码：A1003-0174（ ） - - Identification of the different degree of moldy indica rice based on hyperspectral technology Gong Zhongliang1 Zheng Lizhang1 Wen Tao1,2 Li Lijun1 Xie Jiefei1 Ma Qiang3 (School of Mechanical and Electrical Engineering, Central South University of Forestry and Technology1, Changsha 410004) (Key Laboratory of Key Technology for South Agricultural Machinery and Equipment, Ministry of Education, Engineering College, South China Agricultural University2, Guangzhou 510642) (School of Science, Central South University of Forestry and Technology3, Changsha 410004) Abstract In order to solve the problem of fast and nondestructive identification of moldy. The spectral data of four kinds of moldy indica rice was collected by using the hyperspectral technology, these moldy indica rice included the normal indica rice, mildly moldy indica rice, moderately moldy indica rice and seriously moldy indica rice. After preprocessing with five methods, the successive projections algorithm (SPA) was proposed to extract the characteristic wavelength, then, the identification models of different degree of moldy indica rice were established by using the multivariate linear regression-discriminate analysis (MLR-DA). Meanwhile, the competitive adaptive reweighted sampling (CARS) was used to obtain characteristic variables, the identification models were established by using partial least squares regression - discriminate analysis (PLS-DA). The results show that