1. College of Information and Electrical Engineering，Shenyang Agricultural University，Shenyang 110866，China
2. Beijing Research Center of Intelligent Equipment for Agriculture，Beijing 100097，China
3. National Research Center of Intelligent Equipment for Agriculture，Beijing 100097，China
4. Key Laboratory of Agri-Informatics，Ministry of Agriculture，Beijing 100097，China
5. Beijing Key Laboratory of Intelligent Equipment Technology for Agriculture，Beijing 100097，China
Abstract：Near infrared hyperspectral imaging technology not only can acquire the image information of the sample, but also contain the spectra information about each pixel. Due to the abundant information that the method provides, it has been applied to detect food safety. This study adopted near infrared hyperspectral technology to detect azodicarbonamide in flour. Hyperspectral images of pure azodicarbonamide, pure flour, and azodicarbonamide-flour mixture samples with different concentrations of azodicarbonamide were collected, by comparing the average diffuse reflectance spectra of pure azodicarbonamide and pure flour, the four absorption bands with high difference were found at 1 574.38, 2 038.55, 2 166.88 and 2 269.91nm. Second derivative was used for each pixel in the mixture sample ROI. Azodicarbonamide pixels and flour pixels were detected by three spectral similarity analysis methods: spectral angle mapper, spectral correlation angle, and spectral correlation measure. The results showed that the average spectra after pretreatment cannot effectively detect azodicarbonamide in flour. Spectral similarity analysis of single pixel spectra can be used to classify azodicarbonamide pixels and flour pixels in mixture samples. The validation of the classification results showed the correct classification of azodicarbonamide pixels and flour pixels. This study could provide a method support for the detection of additives in flour based on near infrared hyperspectral imaging technology, and provide a reference for the detection of adulterants in food.
