Nondestructive detection of insect foreign bodies in finished tea products using <scp>THz‐TDS</scp> combination of baseline correction and variable selection algorithms
Xudong Sun, Chao Xu, Jiajun Li, Dongfu Xie, Zhiyuan Gong, Wei Fu, Xinpeng Wang
Abstract
Abstract Current technology of x‐ray imaging detects hard foreign bodies (metal)effectively and extends to plastic, however, soft foreign bodies regarded as undetectable to x‐ray imaging is still a challenge for food safety. Terahertz time‐domain spectroscopy (THz‐TDS) combined with baseline correction and variable selection were employed to discriminate insect foreign bodies in finished tea products. Compared with THz‐TDS absorption coefficient, the accuracy was improved from 88.6% to 97.7% by the use of THz‐TDS time‐domain signals combined with adaptive iteratively reweighted penalized least squares (AirPLS) and Monte Carlo uninformative variable elimination (MC‐UVE) algorithms. Conversely, variable selection algorithms failed to improve the performance of partial least square discrimination analysis (PLS‐DA) model. The results suggested it was feasible to detect insect foreign bodies mixed in the finished tea products using THz time‐domain signals combined with baseline correction and variable selection algorithms. This work provided an alternative method for low‐density food foreign bodies detection using THz‐TDS technology. Practical applications THz time‐domain signals directly applied to calibrate and predict during insect foreign bodies detection without consideration of additional parameter of sample thickness. THz‐TDS system could fix to the tea packaging line for foreign bodies detection and control.