基于CV-ANN技术的花椒色素类成分含量快速无损检测研究

Study on Fast and Non-destructive Detection of Pigments Content of Zanthoxylum bungeanum by CV-ANN Technology

目的:通过探索花椒颜色和色素类成分含量的相关性,为花椒品质的快速评价提供参考依据。方法:采用机器视觉(CV)技术客观量化不同来源花椒的颜色;分别测定花椒样品中黄酮、多酚、花青素以及叶绿素的含量;采用SPSS 20.0软件分析花椒颜色与色素类成分之间的相关性;借助matlab工具箱采用人工神经网络(ANN)训练建模,根据花椒的颜色值预测花椒色素类成分含量。结果:花椒黄酮、多酚和花青素含量与L*值呈显著负相关,与a*值呈显著正相关;叶绿素含量与a*值呈显著负相关,与b*值呈显著正相关;花青素含量与b*值呈显著负相关。人工神经网络模型中,花椒黄酮、多酚、花青素和叶绿素的预测值与实测值相关系数分别为0.9881,0.8797,0.9842,0.9854。结论:花椒颜色与色素类成分含量密切相关,通过机器视觉-人工神经网络联用技术能够实现花椒色素类成分的快速无损检测,达到快速评价花椒品质的目的。

Objective:Explore the correlation between color and pigments content of Zanthoxylum bungeanum and provide a reference for fast evaluation of its quality.Methods:Computer vision(CV)technology is used to objectively quantify the color of Zanthoxylum bungeanum from different sources.The content of flavonoids,polyphenols,anthocyanins and chlorophyll in Zanthoxylum bungeanum is determined respectively.SPSS 20.0 software is used to analyze the correlation between color and pigments content of Zanthoxylum bungeanum.The artificial neural network(ANN)training model is adopted with the matlab toolbox to predict the pigments content of Zanthoxylum bungeanum according to color value.Results:The content of flavonoids,polyphenols and anthocyanins in Zanthoxylum bungeanum is negatively correlated with L*value and positively correlated with a*value.Chlorophyll content is negatively correlated with a*value and positively correlated with b*value.There is a significant negative correlation between anthocyanin content and b*value.In the artificial neural network model,the correlation coefficients between predicted value and measured value of flavonoids,polyphenols,anthocyanins and chlorophyll are 0.9881,0.8797,0.9842,0.9854 respectively.Conclusion:The color of Zanthoxylum bungeanum is closely related to the pigments content,CV-ANN technology pachieves fast and non-destructive detection of pigments content of Zanthoxylum bungeanum and fast evaluation of the quality of Zanthoxylum bungeanum.