摘要

   研究评估了使用多光谱成像技术 (405–970 nm)预测猪肉中的铁卟啉（IP）含量的可行性。通过连续投影算法（SPA）鉴别了9个特征波长，建立了偏较小二乘回归(PLSR)校准模型，预测集的决定系数（R2）为0.996。 使用预测模型多光谱图像将IP含量以单像素成像，这是生产过程中有效监控IP含量的分布和实时演化的方式。结果揭示出MSI技术与PLSR校准分析结合的应用潜力，该方法是一种农业食品业中快速无损评估IP含量的方法。

关键词

多光谱成像，铁卟啉，猪肉，偏较小二乘法，连续投影成像，成像

Rapid and Non-destructive Detection of Iron Porphyrin Content in Pork Using Multispectral Imaging Approach

Abstract

The feasibility of using multispectral imaging (MSI) technique (405–970 nm) for predicting iron porphyrin (IP) content in pork was assessed. Based on nine feature wavelengths identified by successive projections algorithm (SPA), a partial least squares regression (PLSR) calibration model was established, leading to a determination coefficient (R2) of 0.996 for the prediction set. Visualization of IP content in each pixel of the multispectral images using the prediction model offers an effective way to monitor the distribution and real-time evolution of IP content during the manufacturing process. The results revealed the potentiality of MSI technique combined with PLSR calibration analysis as a rapid and non-destructive evaluation method of IP content for the agri-food industry.

Keywords

Multispectral imagingIron porphyrin，Pork，Partial least squares regression，Successive projections algorithm，Visualization