Rapid Discrimination of Fish Feeds Brands Based on Visible and Short-Wave Near-Infrared Spectroscopy
Fengle Zhu,Shuxi Cheng,Di Wu,Yong He
DOI: https://doi.org/10.1007/s11947-010-0369-6
2010-01-01
Food and Bioprocess Technology
Abstract:Visible and short-wave near-infrared spectroscopy (Vis/SW-NIRS) was investigated in this study to distinguish four different brands of granular and ground fish feeds. A total of 240 samples were prepared for spectra collecting from a field spectroradiometer (325–1,075 nm), 160 of which were randomly selected to create the calibration model, and the rest 80 ones to verify the model. Three types of pretreatments including standard normal variate (SNV), multiplicative scattering correction (MSC), and Savitzky–Golay smoothing were adopted to eliminate the system noises and external disturbances. Then partial least squares analysis was implemented for calibration models. For granular samples, the discrimination ability of SNV and MSC was better than smoothing. However, for ground samples, the recognition ability of smoothing surpassed SNV and MSC. Moreover, for threshold values of ±0.1 and ±0.2, the average discrimination accuracy rates were 85.8% and 97.5%, respectively, for four granular samples and 95.4% and 100%, respectively, for ground samples, which signified the better discrimination performances of ground samples than granular ones. The results indicated that Vis/SW-NIRS technique could be promising to be applied as a rapid and highly accurate way for the qualitative discrimination of fish feeds brands.