Detection Of Fumigated And Dyed Chinese Wolfberry By Hyperspectral Imaging

Chinese wolfberry is a kind of legal medicine and food in China. It’s a traditional Chinese herbal medicine and nutritional supplements. Chinese Wolfberry has various beneficial effects on people. In recent years, with the enhancement of people’s health consciousness and its affordable price, the market demand of Chinese wolfberry has been increased rapidly, which leads to some unscrupulous traders enhance a fancy appearance of the Chinese wolfberry with chemical methods to sell them at a higher price, which is Sulphur Fumigation and chemical staining of the Chinese wolfberry. There are no efficient and quick methods of nondestructive testing. Therefore, the paper proposes the use of hyperspectral imaging detection method, in order to accurately distinguish normal dry Chinese wolfberries from those chemical processed ones. This method will bring the availability for the future development of Chinese wolfberry sulfur fumigation and stained multispectral online judging system. The main methods and conclusions are as follows:(1)Take the same batch of Chinese wolfberry, and then drying, normal sulfur fumigating and chemical dyeing them. Next, divides them into Normal-Fuming Group and Normal-Colored Group and use hyperspectral imaging system to obtain high spectral image data samples in the range of 408-1013 nm, finally, take the average spectra of the region which the computer interested in.(2)Analyze the normal- fumigation Chinese wolfberry samples, through principal component analysis, spectral screening 650 nm and 1000 nm are two important wavelength, and the two band spectral values of the divisions to get the band ratio spectra, make use of discriminant analysis discriminant analysis of spectral band ratios by Fisher, and the single band image of the two special bands are divided and get the band ratio image, finally the application of the mask was obtained after removing the background band ratio image. The results show that the visible and near infrared hyperspectral imaging technology can be a very good test to distinguish normal sulfur fumigation and modular. The correct detection rate reaches 100%. Only 2 of the 953 bands were used in the study. The amount of data was reduced greatly.(3)The identification of normal colored wolfberry sample, two samples of the region of interest extraction in spectral data by using MATLAB software are proceed, using different spectral pretreatment methods and different number of principal components of Lyceum barb arum and chemical staining were compared and analyzed, then builds the qualitative discrimination model of Lyceum barb arum. The results show that, using the discriminant analysis method combined with the original spectra were used to be established DA model of the optimal level of correction. The method of correct classification is 100%.