摘要
以发酵醋醅为研究对象,应用高光谱图像技术对其图像和光谱信息进行研究,以期对发酵状况快速预测。首先通过主成分分析(PCA)对其图像信息进行PCA;然后利用预处理后的光谱信息结合全光谱偏最小二乘(PLS)、区间偏最小二乘法(i PLS)和联合区间偏最小二乘法(si PLS)建立总酸、p H及不挥发酸含量的快速预测模型,选择最优模型。结果表明,依据图像信息的不同主成分,优选出3幅特征图像,提取每幅图像的对比度、相关性、角二阶矩和一致性等4个基于灰度共生矩阵的纹理特征变量,利用K-最邻近法(KNN)建立发酵醋醅的识别模型,预测集识别率达到90.04%,能很好的预测醋醅发酵状况;优选出si PLS模型最优,预测集总酸、p H值和不挥发酸的RMSEP分别为0.75、0.05和0.3,能够实现重要理化指标的快速预测。因此利用高光谱图像技术可快速预测醋醅发酵状况,为优化工艺操作和提高发酵质量等提供有效、快速地检测手段。
The objective of this study was to investigate the imaging and spectral characteristics of fermented vinegar grains by using the hyperspectral imaging technique （HPIT）, to rapidly predict the fermentation status. Initially, a primary component analysis （PCA） was performed. The pre-treated spectral information was then combined with the partial least squares （PLS）, interval PLS （iPLS）, and synergy interval PLS （siPLS） of the full spectrum to establish rapid prediction models for total acid content, pH, and non-volatile acid content, in order to select the best prediction model. Three characteristic images were chosen based on the different main components represented by the imaging data. We extracted four characteristic variables （contrast, correlation, angular second moment, and homogeneity） by texture analysis, based on gray level co-occurrence matrix, The K-nearest neighbor （KNN） method was used to establish a recognition model for fermented vinegar grains, with a predicted recognition rate of 90.04%, which would enable a good prediction of the vinegar grain fermentation status. The synergy interval partial least squares （siPLS） model displayed the best performance, predicting the total acid content, pH value, and root mean squared error of prediction （RMSEP） of non-volatile acids to be 0.75, 0.05, and 0.3, respectively. This finding indicated that the model could rapidly predict important physical and chemical parameters. Therefore, it would be feasible to use HPIT for the rapid prediction of fermentation quality of vinegar grains. This study provides an effective and rapid means of detection to improve the process operation and quality of fermentation.
出处
《现代食品科技》
EI
CAS
北大核心
2014年第12期119-125,共7页
Modern Food Science and Technology
基金
国家863科技计划项目(2011AA100807)
全国优秀博士基金资助项目(200968)
国家自然科学基金(61301239)
新世纪优秀人才项目(NCET-11-00986)
江苏省杰出青年基金(BK20130010)
关键词
固态发酵
高光谱图像技术
图像分析
光谱分析
solid-state fermentation
hyperspectral image technique
imaging analysis
spectral analysis
