[Discrimination of pork storage time using near infrared spectroscopy and Adaboost+OLDA].
[Discrimination of pork storage time using near infrared spectroscopy and Adaboost+OLDA].
复制标题
DOI：
--
复制
发表时间：
2012-12
期刊：
影响因子：
--
通讯作者：
Xiaohong Wu;K. Tang;Jun Sun
中科院分区：
文献类型：
--
作者：
Xiaohong Wu;K. Tang;Jun Sun
猪肉的储存时间与其新鲜度密切相关。借助近红外漫反射光谱仪收集猪肉样品数据。使用正交线性判别分析（OLDA）算法来提取特征。进一步将Adaboost算法引入OLDA中，在OLDA和Adaboost的基础上提出了一种新的算法Adaboost+OLDA。为了考察Adaboost+OLDA算法的分类率和计算时间，在实验中将经典特征提取方法（PCA+LDA和OLDA）与Adaboost+OLDA进行了比较。实验结果表明，Adaboost+OLDA可以高效计算，提高了OLDA的泛化能力。 Adaboost+OLDA的平均分类率超过95%。
Pork storage time is closely related to its freshness. With the help of near infrared diffuse reflectance spectroscopy, pork sample data were collected. The orthogonal linear discriminant analysis (OLDA) algorithm was used to extract features. Furthermore, by introducing Adaboost algorithm to OLDA, a new algorithm, named Adaboost+OLDA, was proposed based on OLDA and Adaboost. To investigate the classification rate and the computational time of Adaboost+OLDA algorithm, the classical feature extraction methods (PCA+LDA and OLDA) were compared with Adaboost+OLDA in the experiments. Experimental results showed that Adaboost+OLDA could be computed efficiently and in improved the generalization ability of OLDA. The average classification rate of Adaboost+OLDA is more than 95%.