基于共聚焦拉曼光谱技术的苹果轻微损伤早期判别分析

苹果在采摘、分拣、储存和运输过程中容易受到挤压、振动和碰撞而损伤，轻微损伤早期肉眼很难识别，轻微损伤部位易被病原微生物入侵而导致自身和周围水果腐烂，因此，苹果轻微损伤的早期快速准确地判别能有效地降低经济损失，对苹果的采后处理和储存具有重要意义。本研究应用拉曼光谱结合化学计量学方法对苹果早期轻微损伤进行快速识别。采用Savitzky-Golay(SG)卷积对原始拉曼光谱进行平滑去噪，用自适应迭代重加权惩罚最小二乘(airPLS)算法进行基线校正，用非线性的支持向量机(SVM)回归算法建立分类判别模型，采用KS法划分训练集和验证集后，基于线性和多项式核函数建立SVM分类模型的分类准确率可达到97.8%。结果表明，拉曼光谱技术结合化学计量学方法可快速识别苹果的早期轻微损伤，展示了拉曼光谱技术用于判别苹果早期轻微损伤的应用前景。

Detection of Early Tiny Bruises in Apples using Confocal Raman Spectroscopy

Mechanical damage of apple can result from compression, vibrations and collisions during harvesting, handling, transport operation and storing process. The part of tiny bruise apple is unable to be identified by the naked eye and is more likely to be invaded by micro-organism and pathogen, which will not only cause the affected fruit to rot, but will also affect other intact fruit. Therefore, it is significant for the postharvest treatment and storage to a identify the early tiny bruise of apple quickly and accurately, which can reduce economic losses. Raman spectroscopy combined with chemometric methods was used to rapidly classify apple flesh with early tiny bruising. SG (Savitzky-Golay) was used to smooth spectroscopy. AirPLS (adaptive iteratively reweighted penalized least squares) was used to correct the baseline of spectroscopy. After using KS method to divide training set and verification set, classified models were developed with non-linear support vector machine (SVM) regression which were based on the linear and polynomial kernel functions. The classification accuracy rate was 97.8%. The results showed that Raman spectroscopy combined with chemometric methods can quickly identify the early tiny bruise of apple, demonstrating the application prospect of Raman spectroscopy to discriminate the early tiny bruise apple.