摘要
[目的]探究快速检验草莓的成熟度、提高草莓采摘自动化水平的方法。[方法]采集350～2 500 nm波段草莓的光谱信息,提取光谱的一阶导数值并进行主成分分析,取2个主成分并获得了主成分中贡献率最大的6个峰值点,将各个峰值点两侧各取5个波段点作为特征点。随机选取58个样本组成60×58的矩阵作为训练集,导入用Matlab建立的人工BP神经网络中进行训练。[结果]利用测试集进行识别模型的检验,识别正确率达到93.1%。[结论]利用近红外光谱对草莓成熟度进行识别是可行的。
[Objective] To explore the method for the rapid detection of strawberries＇ maturity level and the improvement of the automatic strawberry picking level.[Method] The strawberries＇ spectral information around 350-2 500 nm was collected,and the first derivative of the strawberries＇ spectral information was drawn for carrying out a principal components analysis.The six points which contributed the largest in the principle components were obtained.The five additional feature points aside each crest feature points were picked out as main feature points,and 58 strawberries were selected randomly to made up a 60×58 array as training set.The array was imported to BP neural network which was built of Matlab for training.[Result] To test the neural network model by the strawberries samples of test set,the distinguishing rate could reach up to 93.1%.[Conclusion] To identify the maturity degree of strawberries by near infrared spectra is practicable.
出处
《安徽农业科学》
CAS
2012年第10期6292-6294,共3页
Journal of Anhui Agricultural Sciences
基金
江苏省高校自然科学研究项目(09KJD210003)
关键词
神经网络
主成分分析
近红外光谱
成熟度识别
草莓
Neural network
Principal components analytic
Near infrared spectra
Recognition of maturity level
Strawberry
