Real-time inspection of pork quality attributes using dual-band spectroscopy. (November 2018)

• Record Type: Journal Article
• Authors: Wang, Wenxiu; Peng, Yankun; Sun, Hongwei; Zheng, Xiaochun; Wei, Wensong
• Is Part Of: Journal of food engineering. Volume 237(2018)
• Page Start: 103
• Page End: 109
• Publication Date: 2018-11
• Subjects: Portable device -- Real-time detection -- Response correction -- Dual-band Vis/NIR spectroscopy -- Quality attributes -- Pork; Food industry and trade -- Periodicals; Food -- Analysis -- Periodicals; Aliments -- Industrie et commerce -- Périodiques; Aliments -- Analyse -- Périodiques; Aliments -- Recherche -- Périodiques; 664.005
• DOI: 10.1016/j.jfoodeng.2018.05.022
• Languages: English
• ISSNs: 0260-8774

Abstract

Abstract: The real-time determination of meat quality is important to both consumers and meat producers. In this study, a portable optical device was developed based on dual-band visible/near-infrared (Vis/NIR) spectroscopy to realize the real-time and simultaneous detection of multiple quality attributes of pork. A 'response correction' algorithm was proposed to fuse two separate spectral regions into a contiguous one that covers the entire Vis/NIR region. Then, partial least square regression (PLSR) models based on single and dual spectral regions for each attribute were developed and their results were compared. Finally, an improved competitive adaptive reweighted sampling algorithm (ICARS) was proposed to select specific wavelengths for each attribute. The optimized calibration models using these selected informative variables exhibited better performance than those with full band spectra. The results demonstrated that the developed device in tandem with the proposed algorithm could help realize the real-time detection of multiple quality attributes of meat. Highlights: A portable detection device was developed based on dual-band Vis/NIR spectroscopy. A 'response correction' algorithm was proposed to fuse two separate spectral regions. PLSR models were established for multiple quality attributes of pork meat. Feature wavelengths were selected for each attribute. Real-time detection of pork was realized by the device and the algorithm.