The Detection of Pest Contaminants in Chocolate Using Visible-Near-Infrared Single-Pixel Imaging Technology

Food safety is gaining increasing attention worldwide. Currently, low-density organic foreign objects such as insects are extremely challenging to detect using conventional metal detectors and X-ray inspection systems. This study aimed to develop a visible-near-infrared single-pixel imaging (Vis-NIR-SPI) method to detect small insects inside food. The advantages of Vis-NIR light include its ability to analyze samples non-destructively and measure multiple components simultaneously and quickly, while SPI is robust against dark noise, high scattering, and high equipment costs. The experimental results demonstrated that (1) the newly designed system effectively reduces scattering effects from the highly scattering sample (intralipid 20%), allowing for the capture of information beyond the capabilities of a charge-coupled-device camera; (2) insects positioned behind ham and bread were readily detectable using the imaging reconstruction algorithm; and (3) even for chocolate samples with very high light absorption, only 1 uncontaminated sample out of 100 was mistakenly classified as contaminated, yielding an overall accuracy of 99%. This high level of accuracy underscores the potential of the Vis-NIR-SPI method to provide reliable detection while maintaining sample integrity. Furthermore, this method is cost-effective, offering a practical and efficient solution to improve quality control processes and consumer trust in the food industry.