牛奶三维荧光光谱特征提取.pdf

中国乳品工业 rpgy@ 牛奶三维荧光光谱特征提取 1 1 1 2 1 陈雪 ，郭荣 ，陈旭 ，韩奕奕 ，雍克岚 （ 上海大学生命科学学院，上海 ； 上海市乳品质量监督检验站，上海 ） 1． 200444 2． 200072 摘 要：测量了 个牧场连续 天合格奶样的三维荧光光谱，使用 对三维荧光光谱进行了归一化、平滑滤波和去除散射效应等预处 7 7 Matlab 理，然后对处理后的数据矩阵依次进行了奇异值分解，提取其第一主成分作为特征光谱。 结果表明：该特征光谱可区分 个牧场来源的 7 合格奶样，且可判断出该牧场合格奶样质量是否稳定。 各牧场奶质的稳定性，相关系数、第一第二大奇异值比值与第一主成分特征光谱 直观结果一致。 关键词：牛奶；三维荧光光谱；奇异值分解；相关系数 中图分类号： 文献标识码： 文章编号： （ ） TS252.1 A 1001-2230 2009 07-0013-03 Research on feature extractions of 3D spectrum of milk 1 1 1 2 1 CHEN Xue , GUO Rong , CHEN Xu , HAN Yi-yi , YONG Ke-lan （ ， 1.Life School, Shanghai University, Shanghai 200444 China; 2．Dairy Quality Supervision and Inspection Bureau of Shanghai ，Shanghai 200072 ，China ） Abstract 3D fluorescence spectrum of qualified milk from seven farms were measured in seven days and their feature extraction methods were explored. With simple ordering Rayleigh scattering data as zero, the obtained 3D spectrum was processed by singular value decomposition. The result showed that the singular value decomposed spectrum could be used as a characteristic spectrum to differentiate the qualified milk from seven farms and analyze its quality stability. For milk quality stability of seven farms, correlation coefficient, singular value decomposition and character spectrum share the same result. ； ； ； Key words