Information fusion of hyperspectral imaging and electronic nose for evaluation of fungal contamination in strawberries during decay
Information fusion of hyperspectral imaging and electronic nose for evaluation of fungal contamination in strawberries during decay
复制标题
高光谱成像与电子鼻信息融合评价草莓腐烂过程中的真菌污染
DOI：
10.1016/j.postharvbio.2019.03.017
复制
发表时间：
2019-07-01
影响因子：
7
通讯作者：
Tu, Kang
中科院分区：
文献类型：
--
作者：
Liu, Qiang;Sun, Ke;Tu, Kang
研究了基于高光谱成像（HSI）和电子鼻（E-nose）的草莓腐烂过程中微生物含量和品质属性的无损快速检测方法。采用主成分分析（PCA）对HSI和E-nose数据进行降维处理，提取特征信息。建立了草莓微生物含量和品质性状的定量预测模型。结果表明，外观的变化（即，颜色）和内部组成（即，可溶性固形物和可滴定酸）与草莓贮藏期间的微生物含量高度相关。从HSI和E-nose数据集中提取10个基本PC（累积贡献率超过99%），用于直接改进预测模型。基于HSI和E-nose数据的原始信息融合构建的模型并没有提高预测精度。相比之下，基于特征信息融合与基本PC构建的模型具有更好的预测性能比基于单一数据集（HSI或E-nose）构建的模型。最佳预测模型能够预测菌落形成单位，R-P（2）为0.925，RMSEP为0.38 log（10）（CFU g（-1））。本研究表明，这两种传感技术的结合，可以潜在地实现检测草莓的安全性和质量。
The non-destructive method developed based on hyperspectral imaging (HSI) and electronic nose (E-nose) to rapidly detect microbial content and quality attributes of strawberries during decay, was evaluated. Principal component analysis (PCA) was applied to reduce the dimensionality of the data and to extract featured information from the HSI and E-nose data. Quantitative prediction models were developed to forecast the microbial contents and the quality attributes of strawberries. The results showed that the changes in exterior appearances (i.e., color) and interior compositions (i.e., total soluble solids and titratable acidity) of fungi-infected strawberries during storage were highly correlated with the microbial content. Ten essential PCs (with over 99% cumulative contribution) extracted from HSI and E-nose datasets were needed for directly improve the prediction models. The model constructed based on the raw information fusion of HSI and E-nose data did not improve the prediction accuracy. By contrast, the model constructed based on featured information fusion with essential PCs had better prediction performance than that constructed based on single dataset (HSI or E-nose). The best prediction model was able to predict colony-forming units with a 0.925 R-P(2) and RMSEP of 0.38 log(10) (CFU g(-1)). This study illustrates that the combination of the two sensing techniques can potentially be implemented for the detection of safety and quality of strawberries.