Application Case of the CHNSpec FS-22 Hyperspectral Camera in Non-Destructive Detection of Beef Freshness

As a high-protein, low-fat premium food ingredient, beef freshness directly affects food safety and commercial value. Traditional detection methods such as high-performance liquid chromatography–mass spectrometry (HPLC-MS) and gas chromatography–mass spectrometry (GC-MS), although highly accurate, suffer from limitations including complex operation, high cost, and destructive sampling. In recent years, hyperspectral imaging technology has gradually gained attention in agricultural product quality inspection due to its advantages of rapid, non-destructive analysis and the ability to simultaneously acquire spectral and image information.

The FS-22 hyperspectral camera launched by Hangzhou CHNSpec Technology Co., Ltd. features a spectral range of 400–1000 nm, a spectral resolution of 5 nm, and 300 acquisition bands. Supported by a line-array InGaAs detector, it enables stable acquisition of surface spectral information of objects. The device has demonstrated good applicability in quality inspection within agriculture and food-related fields.

I. Research Background and Methods

In a recent study on beef freshness, a research team constructed a hyperspectral imaging system in which the FS-22 hyperspectral camera was used as the core acquisition device. The system was also equipped with halogen-tungsten light sources and professional analysis software to collect spectral information from fresh beef samples.

A total of 112 beef samples were prepared and stored at 4 °C for 14 days to simulate quality changes during actual circulation. Spectral data were collected weekly, and pH value, total volatile basic nitrogen (TVB-N), and color parameters (L, a, b*) were measured simultaneously to establish correlation models between spectral information and quality parameters.

II. Role of the FS-22 in the System

Within the system, the FS-22 hyperspectral camera was responsible for collecting spectral reflectance information of beef samples in the visible–near-infrared range. With uniform illumination on the sample surface, the camera received reflected light signals, which were dispersed by a grating to form continuous spectral curves. Researchers extracted spectral data from regions of interest of each sample, which were preprocessed and used for subsequent modeling.

The study also employed the CHNSpec DS-100 portable color difference meter to objectively measure beef color parameters, providing reference values for model development.

III. Data Processing and Model Construction

The research team applied SG+MSC methods to preprocess the original spectra to reduce noise interference. Subsequently, the competitive adaptive reweighted sampling (CARS) algorithm was used to extract characteristic wavelengths related to quality parameters, and partial least squares regression (PLSR) prediction models were established.

The results showed that models built based on spectral data acquired by the FS-22 demonstrated high consistency between predicted and measured values for five key parameters: pH, TVB-N, L, a, and b*. The coefficients of determination for the test set reached 0.9539, 0.9660, 0.9266, 0.8683, and 0.9018, respectively. The models exhibited good stability in independent validation, with a beef freshness classification accuracy of 92.6%.

IV. Conclusion

Hyperspectral imaging technology provides new approaches for food quality and safety inspection. The application of the FS-22 hyperspectral camera in this beef freshness detection study demonstrates its practical value in non-destructive quality inspection of agricultural products. With further development of spectral data processing and modeling techniques, hyperspectral imaging is expected to have broader application prospects in the food industry and agricultural inspection.

Product Recommendation：FigSpec FS-22 Imaging Hyperspectral Camera

Product Features

• Spectral range: 400–1000 nm

• Spectral resolution: 5 nm

• Number of spectral channels: 600

• Image resolution: 1920 × 1920