高光谱成像识别白胡椒粉末中掺杂的面粉

doi:10.3964/j.issn.1000-0593(2020)09-2851-05

Abstract
References
Related Citation (15)

Download: PDF (2576 KB)
Export: BibTeX | EndNote (RIS)

Abstract White pepper powder is very similar to flour, so it is difficult to distinguish a small amount of flour from white pepper powder by human vision or smell. Hyperspectral imaging technology can not only obtain spectral information but also obtain spatial position information. Therefore, it is possible to predict the content of flour adulteration in white pepper powder and locate the mixing position in white pepper powder by hyperspectral imaging technology. Sixty-two samples are prepared, including 60 samples of pure flour mixed with pure white pepper powder at a ratio of 1% to 60% by weight and a gradient of 1%, in addition, two samples of the pure pepper powder and the pure flour. Each sample was scanned by the hyperspectral image, and a total of 62 hyperspectral data were obtained. Forty-two samples were selected randomly as correction set for partial least squares regression (PLSR) modeling, and the remaining 20 samples were used for prediction set. The pretreatment method of first derivatives was applied to the average spectrum of each sample, and then the PLSR was used to establish a quantitative analysis model for predicting the flour content in white pepper powder. The experimental results show that the root means square error of the correction set is 0.83%, and the root mean square error of the prediction set is 2.73%. The correlation coefficients between the correction set and the prediction set are 0.99 and 0.98 respectively. In order to locate the specific mixing position of flour in the white pepper powder, the Correlation Coefficient Method and the Maximum and Minimum Criterion were proposed. R₁ are used to indicate the correlation coefficient between the sample and the pure flour, and R₂ indicates the correlation coefficient between the sample and the pure white pepper powder. If the location is pure flour, R₁ reaches the maximum and R₂ reaches the minimum. The difference between R₁ and R₂ is calculated to get R, and R is arranged in order from small to large. Using the prediction result of the PLSR regression model as a threshold, the location of R in less than or equal to the threshold value is identified as flour. Then the position of the flour was marked in the adulterated sample so that it could be visually displayed. This research provides a reference for fast, nondestructive and visual identification of white pepper powder adulteration.

Key words：Hyperspectral imaging; White pepper powder; Flour; Partial least squares regression; Quantitative analysis

Received: 2019-08-21 Accepted: 2019-12-30

ZTFLH:

S59

Corresponding Authors: YU Li-min E-mail: yulimin1978@163.com

Cite this article:

URL:

https://www.gpxygpfx.com/EN/10.3964/j.issn.1000-0593(2020)09-2851-05 OR https://www.gpxygpfx.com/EN/Y2020/V40/I09/2851