Hyperspectral imaging to identify flour doped in white pepper powder
author: Gavin
2022-01-06
Pepper powder can not only be used as a seasoning, but also has high medicinal value. As the economic value of pepper increases, adulterated pepper products have also been reported on the market in recent years. The color of white pepper powder is very close to that of flour. If a small amount of flour is mixed into white pepper powder, it is difficult to distinguish by human vision or smell. Therefore, it is necessary to use hyperspectral imaging technology to predict the content of flour in white pepper powder and locate the flour Incorporation site in white pepper powder. This study provides a reference method for realizing the rapid, non-destructive and visual identification of the true and false white pepper.
Principle
Hyperspectral imaging can not only obtain material spectrum information, but also spatial location information. Therefore, the use of hyperspectral imaging technology provides the possibility to locate the adulterants. The near-infrared spectrometer combines the partial least squares method and the PLSR method to establish a quantitative analysis and prediction model for the content of flour in white pepper powder. The spectral data was substituted into the regression model of PLSR to obtain the concentration value of the white pepper powder doped with flour. Finally, the correlation coefficient method and the maximum and minimum discriminating criteria were used to mark the position of the flour in the original spectral image.
Solution
Spectral collection uses Optosky ATH1010, AATH1010 is low cost hyperspectral camera for industrial application, It it employs CMOS sensor and the spectral range is 380-1000nm。
Principle
Hyperspectral imaging can not only obtain material spectrum information, but also spatial location information. Therefore, the use of hyperspectral imaging technology provides the possibility to locate the adulterants. The near-infrared spectrometer combines the partial least squares method and the PLSR method to establish a quantitative analysis and prediction model for the content of flour in white pepper powder. The spectral data was substituted into the regression model of PLSR to obtain the concentration value of the white pepper powder doped with flour. Finally, the correlation coefficient method and the maximum and minimum discriminating criteria were used to mark the position of the flour in the original spectral image.
Solution
Spectral collection uses Optosky ATH1010, AATH1010 is low cost hyperspectral camera for industrial application, It it employs CMOS sensor and the spectral range is 380-1000nm。
(1).Prepare 60 parts of white pepper, each weighing 500g, use a pulverizer to pulverize the white pepper particles, and then sift through a 60-mesh sieve to obtain a pure white pepper powder sample. The finished product of pure flour was purchased from the supermarket. Use an analytical balance with an accuracy of 0.1 mg to take out 5,000 mg of powder from the prepared white pepper powder sample, and mix the pure flour into the pure white pepper powder according to a gradient of 1% to 60% to 1%; stir the sample with a mixer Evenly mix the flour and white pepper powder to obtain 60 sample powders with different doping concentrations,then take one sample each of pure white pepper powder and pure flour, which constitutes a total of 62 samples. Put the sample into a sample cell with a size of 6.80cm×6.80cm, and use a spatula to repeatedly scrape the surface of the sample cell to make the surface of the sample flat.
(2).In order to reduce the influence of the moisture content of the sample on the spectrum, the sample was dried by a BH-2 blast dryer for 1 hour and then taken out to prepare for hyperspectral scanning.
(3).Spectral collection uses Optosky ATH1010,The computer controls the operation of the camera and the sample stage through the HyperSpectral Image software. The obtained hyperspectral pseudo-color image is shown in Figure 1(b). Figure 1 (b) is a pseudo-color image of RGB three primary colors synthesized by three wavelength points of 680, 550 and 470 nm respectively.
(4).The reflectance curve after calibration of the whiteboard and blackboard is shown in Figure 3. As the wavelength increases, the total reflectance of the sample The body presents a slow upward trend; because the white pepper powder has a greater absorption of the spectrum than flour, as the sample doping concentration increases, the reflectance ratio becomes larger.
(5).In the preprocessing of the spectrum, common methods include decentralization, standard normal transformation, multivariate scattering correction, and the application of the first derivative and second derivative of the Safezky-Golay smoothing. And a compound pretreatment method that combines the above methods in pairs. This work will analyze and compare the effects of eight methods, including no pre-processing, Mean, SNV, MSC, MSC+Mean, MSC+SNV, SAVG1 and SAVG2, on the establishment of the PLSR prediction model. The prediction performance of the PLSR model is evaluated according to the root mean square error of the correction set, the root mean square error of the prediction set, the correlation coefficient of the correction set, and the correlation coefficient of the prediction set. Under different principal components, the smaller the RMSEP value, the larger the RP value, indicating that the prediction performance of the model is better.
Test result
For 42 samples in the calibration set, partial least square regression (PLSR) was used to establish a prediction model for identifying the flour content in white pepper powder, and then the prediction model was used to test the 20 samples in the prediction set. In order to obtain the best prediction model, the original spectral data were analyzed and compared through different prediction processing methods such as Mean, SNV, MSC, MSC+Mean, MSC+SNV, SAVG1 and SAVG2. As shown in Table 1.
Figure 6(a) is a grayscale image of a sample mixed with 25% flour at a wavelength of 550nm. Figure 6(b) shows the position of the flour in the original gray-scale image of Figure 6(a) after using the correlation coefficient method and the maximum and minimum discrimination criteria. The white display part in the figure represents the flour. In order to show the white pepper powder mixed with flour more clearly, the figure 6(b) is converted into a binary image. Black represents white pepper powder and white represents flour. The result is displayed after magnification 7 times. As shown in 7(b). The results of other doping ratios are shown in Figure 7 (a-f).
Conclusion
Hyperspectral scanning was performed on 60 samples in which pure flour was mixed into pure white pepper powder according to a ratio of 1% to 60% and a gradient of 1%. Calculate the average spectrum of each sample. After Mean pretreatment, use the PLSR method to establish a quantitative analysis model for predicting flour content. The RMSEC of the calibration set is 0.83%, and the RMSEP of the prediction set is 2.73%. The prediction sets all have higher correlation coefficients RC = 0.99 and RP = 0.98. PLSR modeling has a better predictive effect. The main reason is that white pepper powder and flour have different light absorption degrees. The light reflection of flour is stronger, while the light absorption of white pepper powder is stronger. Therefore, as the concentration of flour mixed increases, the reflectance of the spectrum becomes larger, showing a stronger correlation coefficient. Experiments show that the use of hyperspectral imaging technology can not only predict the content of flour mixed in white pepper powder, but also locate the position of flour in white pepper powder through spectral data of spatial location points. In the future, we will further study the method of using hyperspectral imaging to identify multiple substances mixed in pepper powder.
Related products
Classification of vegetation diseases and insect pests based on hyperspectral remote sensing data
FieldSpec Spectroradiometer:ATP9110-25
Hyperspectral Camera:ATH1010
Related articles
(1) BansalS, Sinh A, Mangal M, et al. CriticalReviewFoodScience and Nutrition, 2017, 57(6): 1174.[3] Gorgani L, Mohammadim, Najafpur GD, et al. Comprehensive Reviews In Food Science and Food Safety, 2017, 16(1):124.
(2) Hamrapurkar PD, Jadhav K, Zine S. Journal of Applied Pharmaceutical Science, 2011, 1(3): 117.
(3)Paravathy V A, SwethaVP, SheejaTE, et al. Food Biotechnology, 2014, 28(1): 25.
(3).Spectral collection uses Optosky ATH1010,The computer controls the operation of the camera and the sample stage through the HyperSpectral Image software. The obtained hyperspectral pseudo-color image is shown in Figure 1(b). Figure 1 (b) is a pseudo-color image of RGB three primary colors synthesized by three wavelength points of 680, 550 and 470 nm respectively.
(4).The reflectance curve after calibration of the whiteboard and blackboard is shown in Figure 3. As the wavelength increases, the total reflectance of the sample The body presents a slow upward trend; because the white pepper powder has a greater absorption of the spectrum than flour, as the sample doping concentration increases, the reflectance ratio becomes larger.
(5).In the preprocessing of the spectrum, common methods include decentralization, standard normal transformation, multivariate scattering correction, and the application of the first derivative and second derivative of the Safezky-Golay smoothing. And a compound pretreatment method that combines the above methods in pairs. This work will analyze and compare the effects of eight methods, including no pre-processing, Mean, SNV, MSC, MSC+Mean, MSC+SNV, SAVG1 and SAVG2, on the establishment of the PLSR prediction model. The prediction performance of the PLSR model is evaluated according to the root mean square error of the correction set, the root mean square error of the prediction set, the correlation coefficient of the correction set, and the correlation coefficient of the prediction set. Under different principal components, the smaller the RMSEP value, the larger the RP value, indicating that the prediction performance of the model is better.
Test result
For 42 samples in the calibration set, partial least square regression (PLSR) was used to establish a prediction model for identifying the flour content in white pepper powder, and then the prediction model was used to test the 20 samples in the prediction set. In order to obtain the best prediction model, the original spectral data were analyzed and compared through different prediction processing methods such as Mean, SNV, MSC, MSC+Mean, MSC+SNV, SAVG1 and SAVG2. As shown in Table 1.
Figure 6(a) is a grayscale image of a sample mixed with 25% flour at a wavelength of 550nm. Figure 6(b) shows the position of the flour in the original gray-scale image of Figure 6(a) after using the correlation coefficient method and the maximum and minimum discrimination criteria. The white display part in the figure represents the flour. In order to show the white pepper powder mixed with flour more clearly, the figure 6(b) is converted into a binary image. Black represents white pepper powder and white represents flour. The result is displayed after magnification 7 times. As shown in 7(b). The results of other doping ratios are shown in Figure 7 (a-f).
Conclusion
Hyperspectral scanning was performed on 60 samples in which pure flour was mixed into pure white pepper powder according to a ratio of 1% to 60% and a gradient of 1%. Calculate the average spectrum of each sample. After Mean pretreatment, use the PLSR method to establish a quantitative analysis model for predicting flour content. The RMSEC of the calibration set is 0.83%, and the RMSEP of the prediction set is 2.73%. The prediction sets all have higher correlation coefficients RC = 0.99 and RP = 0.98. PLSR modeling has a better predictive effect. The main reason is that white pepper powder and flour have different light absorption degrees. The light reflection of flour is stronger, while the light absorption of white pepper powder is stronger. Therefore, as the concentration of flour mixed increases, the reflectance of the spectrum becomes larger, showing a stronger correlation coefficient. Experiments show that the use of hyperspectral imaging technology can not only predict the content of flour mixed in white pepper powder, but also locate the position of flour in white pepper powder through spectral data of spatial location points. In the future, we will further study the method of using hyperspectral imaging to identify multiple substances mixed in pepper powder.
Related products
Classification of vegetation diseases and insect pests based on hyperspectral remote sensing data
FieldSpec Spectroradiometer:ATP9110-25
Hyperspectral Camera:ATH1010
Related articles
(1) BansalS, Sinh A, Mangal M, et al. CriticalReviewFoodScience and Nutrition, 2017, 57(6): 1174.[3] Gorgani L, Mohammadim, Najafpur GD, et al. Comprehensive Reviews In Food Science and Food Safety, 2017, 16(1):124.
(2) Hamrapurkar PD, Jadhav K, Zine S. Journal of Applied Pharmaceutical Science, 2011, 1(3): 117.
(3)Paravathy V A, SwethaVP, SheejaTE, et al. Food Biotechnology, 2014, 28(1): 25.
Airborne hyperspectral image inversion of soil heavy metal content Empirical model selection and feature extraction
Hyperspectral imaging system detects and characterizes the spectralum of microplastics
Related Article
Discover how UAV‑based hyperspectral imaging and deep learning can monitor SDI, TOC, and TEP – key membrane fouling indicators for seawater desalination during harmful algal blooms. Based on a recent Water Research study, this article explores how Optosky's ATH9010 enables spatial risk mapping and proactive intake management.
Applications | ATH9010 Enables Water Quality and Membrane Fouling Monitoring for Desalination During Algal Blooms
Choosing between 785 nm and 1064 nm for Raman? This guide explains the physics of fluorescence, compares signal strength and fluorescence suppression, and provides a step‑by‑step decision tree. Real‑world drug detection and pesticide examples show why wavelength matters – and how dual‑wavelength coverage offers the ultimate solution.
785 nm or 1064 nm? Fluorescence Is the Dividing Line
Inspired by the 2026 Science paper from Zhang Jun’s team on a video‑rate on‑chip hyperspectral microsystem, Optosky’s ATH series moves beyond lab‑only tools. We deliver tailored, real‑world hyperspectral systems that integrate compact hardware, edge AI, and application‑specific algorithms—turning raw spectral data into actionable decisions, right where you need them.
From Science to the Field: Custom Hyperspectral Solutions by Optosky
See how 40 Optosky NY3300Pro multispectral units were deployed across a large‑scale demonstration farmland in Northwest China. With 9‑band imaging, solar power, 4G transmission, and automated data analytics, this solution enables precise growth monitoring, early pest detection, and water‑fertiliser optimisation – a true leap toward smart agriculture.
Case Study | Bulk NY3300Pro Installation Transforms Smart Farming