Hyperspectral imaging for early diagnosis of abiotic stress in strawberry plants

( Mohammad Akbar Faqeerzada ),( Byoung-kwan Cho ) 한국농업기계학회 2022 한국농업기계학회 학술발표논문집 Vol.27 No.1

Strawberry plants year-round produce annual plants exposed to extreme seasonal and environmental changes. Also, strawberry plants are susceptible to intense heat and drought climates affecting plants growth and productivity. Thus, early diagnosis of abiotically stressed plants is essential before symptoms are visible in plants. Advanced optical technologies offer promising approaches for non-destructive visualization of plants' physiological, biochemical, and morphological characteristics in responses due to environmental factors. This work investigated the application of hyperspectral imaging systems for early diagnosis of drought and heat stress in strawberry plants in the range of 400-1000 nm visible near-infrared (VIS/NIR) and 1000-2500 nm shortwave infrared (SWIR). Plants were imaged before and after exposure to heat stress (3 days) and drought (16 days) measured on three alternative days. Initially, plants were segmented from the background, and spectral data were extracted from each leaf's region of interest (ROI). Acquired spectral data were developed in a partial least square discrimination analysis model coupled with multiple preprocessing methods. Both systems presented potentially high accuracy of over 95% in the validation set based on SWIR, While VIS/NIR demonstrated a slightly low accuracy of 92% accuracy. In addition, an image processing algorithm was developed to visualize chemical composition changes in strawberry plants in response to heat and drought stress conditions. In short, this study presented the potential of hyperspectral imaging combined with chemometrics for the early diagnosis of stressed plants.

Postharvest technologies for fruits and vegetables in South Asian countries: a review

Mohammad Akbar Faqeerzada,Anisur Rahman,Rahul Joshi,박은수,조병관 충남대학교 농업과학연구소 2018 Korean Journal of Agricultural Science Vol.45 No.3

Agricultural systems in South Asian countries are dominated by smallholder farmers. Additionally, these farmers have limited access to pre- and post-harvest technologies due to their high initial cost. The lack of these technologies in postharvest handling is responsible for 20% to 44% of fruit and vegetable losses. These high losses are largely the result of a generally weak basic postharvest infrastructure for the preservation of products, which avoids damage from improper handling, transportation, packaging, and storage. High postharvest losses of products negatively affect food availability, food security, and nutrition, as the producer is able to sell less of the farm yield and the net availability of these food commodities for consumption is reduced. An underlying cause of these postharvest losses is the limited awareness and knowledge bases of stakeholders (researchers, farmers, governments, nongovernmental organizations, and merchants) in the traditional supply chains in which these losses occur. The analysis presented in this paper explores the state of postharvest practice in South Asian countries and discusses options for low-cost postharvest technologies in the region that can support small-scale farmers and provide a viable pathway for supply to the market, joining with modern value chains and bringing about individual and regional reduction in postharvest losses of fruits and vegetables. The improvement of basic and simple low-cost technologies through precise research efforts has the potential to prevent such huge losses of products, and help meet the ever-increasing demand for food in South Asian countries.

Hyperspectral Imaging Analysis for External Quality Inspection of Paprika Fruits

( Mohammad Akbar Faqeerzada ),( Ye-na Kim ),( Hangi Kim ),( Byoung-kwan Cho ) 한국농업기계학회 2023 한국농업기계학회 학술발표논문집 Vol.28 No.2

External defects on paprika during post-harvest directly impact its storage and marketability, leading to internal decay and affecting fruit appearance. However, the automatic detection of various paprika defects, including cracks, scars, irregular shapes, color variations, diseases, and bruises, poses a significant challenge due to variations in defect size and visual similarities among different types of defects. In this study, we proposed an accurate and rapid detection method that utilizes hyperspectral imaging in conjunction with multivariate analysis to identify multiple defects in paprika efficiently. Over 500 paprika fruits, each exhibiting varying levels of defects, were subjected to imaging using visible-near infrared (VIS-NIR, 397-1000 nm) and short-wave infrared (SWIR, 1000-2500 nm) systems. Spectral data (2500 spectral) from regions of interest (ROI) on the paprika samples were extracted. Subsequently, a partial least squares discrimination analysis (PLS-DA) model was developed to classify control and defective fruits. Remarkably, developed models achieved over 95% accuracy in predicting defective regions and provided chemical visualization of these regions, enhancing the inspection process. To further enhance the efficient discrimination between normal and defective fruits, we implemented a classification algorithm based on analysis of variance (ANOVA) to determine the optimal wavelength band ratios (F-value). This approach yielded an impressive accuracy rate of over 93% in evaluating the samples. Our findings present a practical and efficient solution for the external quality inspection of paprika, mainly when dealing with naturally occurring defects. Combining hyperspectral imaging, PLS-DA, and ANOVA-based spectral analysis offers a reliable and practical solution for identifying and classifying defects in paprika, ultimately improving product quality and consumer satisfaction.

Prediction of moisture contents in green peppers using hyperspectral imaging based on a polarized lighting system

Mohammad Akbar Faqeerzada,Anisur Rahman,Geonwoo Kim,Eunsoo Park,Rahul Joshi,Santosh Lohumi,Byoung-Kwan Cho 충남대학교 농업과학연구소 2020 Korean Journal of Agricultural Science Vol.47 No.4

Hyperspectral Imaging System for Online Sorting of Adulterated Almond Nuts

( Mohammad Akbar Faqeerzada ),( Mukasa Parez ),( Santosh Lohumi ),( Hoonsoo Lee ),( Hee Young Lee ),( Collins Wakholi ),( Rahul Joshi ),( Byoung-kwan Cho ) 한국농업기계학회 2020 한국농업기계학회 학술발표논문집 Vol.25 No.1

Almonds are nutrient-rich nuts, their health benefits are potentially linked to the high consumption worldwide. Due to relatively higher price, the producers are targeting it as an illegal practice for earning more profit. The most common adulterants are based on superficially matching, which the apricot nut as an adulterant comparatively are inexpensive, almost identical in color, texture, odor, and other physicochemical characteristics with almonds. In addition, apricots nuts contain amygdalin component which is converted to hazardous toxic cyanide in the digestive system. In the past decades, hyperspectral imaging (HSI) has attracted good attention as a rapid, real-time and non-destructive measurement method for food quality and safety analysis. In this study, near-infrared hyperspectral imaging (NIR-HSI system) with the wavelength of 900-1700 nm synchronized to a conveyor belt was used for online detection of added apricot nuts in almond. A total of 448 samples from different varieties of almond and apricot nuts (112x4) were scanned while the samples are moving on the conveyor belt. The spectral data were extracted from each imaged nuts and used for developing a PLS-DA model coupled with different preprocessing techniques. The PLS-DA model showed over 95% accuracy for the validation set. Additionally, the obtained beta coefficient from the developed model was used for pixel-based classification. An image processing algorithm was developed for chemical visualization mapping of almond and apricot nuts. The online classification system feedback with an overall accuracy of 87% for the classification of nuts. The developed online prototype (NIR-HIS) system combined with multivariate analysis exhibit strong potential for classification of adulterated almond, and the result indicates the system can be used effectively for high-throughput industrial classification of adulterated almond nuts in an industrial environment.

High-Throughput Plant Phenotyping for the Chilling Stress of Watermelon Plants

( Mohammad Akbar Faqeerzada ),( Byoung-kwan Cho ) 한국농업기계학회 2021 한국농업기계학회 학술발표논문집 Vol.26 No.2

Plant phenotyping methods are essential tools for the production of demanded crops for a growing population. Access to large-scale high-throughput screening systems are a critical barrier for early quantification of plants states in real-time. For this purpose, we developed an automated high throughput plant screening system which consists of two imaging chambers. The first chamber is equipped with two RGB cameras (top and side view), and a near-infrared hyperspectral imaging (NIR-HSI) system in the wavelength range of 900-1700 nm installed in the second screening chamber. Initially, the system was tested for early detection of chilling stress watermelon plants, a total of 350 watermelon plants were scanned before and after exposure to chilling stress conditions (-5℃) while moving through the chambers on the conveyor belt. An automatic image processing algorithm was developed for image segmentation and data augmentation. The color images applied to a transfer leaning of ResNet50 basis resulted in 90% classification accuracy. While the hyperspectral images were used to extract from each single plant leaves for the development of a partial least square discrimination analysis (PLS-DA) model which displayed over a 95% accuracy for the validation set. The overall results highlight that the high-throughput screen of plants on a combination of machine learning and deep learning has potential to quantify the plants' states under chilling stress condition.

Authenticity Analysis of Almond Powder using FT-NIR Spectroscopy and Chemometrics

( Mohammad Akbar Faqeerzada ),( Santosh Lohumi ),( Byoung-kwan Cho ) 한국농업기계학회 2018 한국농업기계학회 학술발표논문집 Vol.23 No.1

The food adulteration date centuries back today for gaining illegally benefits. Although the majority of adulterated foods do not pose a public health risk, exception related to the cause. The critical point behind adding substance in products are economical purposes to increase the apparent value of the products for reducing production cost for economic gain. It remains a concern among consumers for having fake product in the price of desired commodities or lower quality alternative. A range of conventional methods such as gas chromatography (GC) and, high-performance liquid chromatography (HPLC) are widely used applications for quality analysis of food products. However, these most widely used techniques are personal based, time-consuming, and chemical required. Therefore, food industry required a nondestructive, portable and non-chemical technique for rapid analysis of food products for quality and authenticity evaluation. For this purpose, spectroscopic techniques such as near-infrared, mid-infrared, and Raman spectroscopy have been in use from last several decades. Therefore, in this study, we investigated the potential of FT-NIR spectroscopic technique for authenticity analysis of almond powder which is one of the very common commodity to be adulterated with low prize apricot powder. Thus, the almond powder was adulterated with apricot powder in a concentration range between 0-50% with an interval of 5%. FT-NIR spectra of ten replicated for each concentration were collected and arranged in a matrix. To evaluate the natural relationship among the preprocessed data from different concentration groups, a principal component analysis (PCA) method was applied. As a result, PCA loadings show some important region where the differences lie between almond and apricot powder. Moreover, using the score values of first two principal components, the data can be separated according to their concentration value of apricot. Finally, a multivariate analysis method of partial least square regression (PLSR) was adopted to predict the concentration of apricot powder in almond powder by dividing the data into calibration and validation sets. The PLSR model developed with standard normal variate preprocessed spectra attain highest accuracy (R²_cal=0.997; SEC=0.8% and R²_val=0.991; SEP=1.1%) and lowest error. In addition, the beta coefficient obtained from PLSR model shows unique peaks related to the variation in apricot powder concentration among the different groups of samples. Further, for the reproducibility purpose, the developed model was tested with another variety of adulterated almond powder and similar results (R²>0.99 and SEC<1.8%) was yielded. The result demonstrated the ability of FT-NIR spectroscopy with chemo-metrics for quality analysis of apricot in almond powder.

Hyperspectral Imaging for Green Pepper Moisture Content Prediction Using Polarized Lighting Systems

( Mohammad Akbar Faqeerzada ),( Anisur Rahman ),( Byoung-kwan Cho ) 한국농업기계학회 2017 한국농업기계학회 학술발표논문집 Vol.22 No.2

Illumination is a key subsystem for good quality image acquisition. When the green pepper image was acquired using direct lighting system, provides the specular reflection from the shiny surface features of the objects and its intensity fluctuates with time. These specular pixels, typically with reflectance values approaching or exceeding 100%, produce highly invalid spectral responses. Hence, it is not recommended to include the specular pixels in the data set for any spectral processing. These phenomena can be accomplished by placing a polarizing filter at the light source and another on the camera lens. Therefore, the objective was set to feasibility study VIS/NIR hyperspectral imaging along with polarized lighting system to predict the moisture content of green pepper. After acquiring the spectral images, each green pepper was used for moisture content (w.b) determination by the gravimetric method using a drying oven at 105°C for 24h. The mean spectra from three portions (5×5 pixels) on each 100 matured green peppers were extracted. A multivariate calibration model was developed by using partial least squares (PLS) regression with different preprocessing methods. The result showed that the smoothing preprocessed spectra based model resulted in better performance for the prediction of moisture content compared to other models, with a determination coefficient (R²_pred) of 0.81 and standard error of prediction (SEP) of 0.71%. The regression coefficients yielded by the best PLS regression model were used to select feature wavelengths for creating the mapping image of the moisture content. Accordingly, the polarization filter helps for uniform illumination and removing of gloss and artifact glare from the green pepperimages. Finally, conclude that the hyperspectral imaging with a polarized lighting system and appropriate multivariate analysis has the potential for the prediction of moisture contentin green pepper.

Hyperspectral imaging for predicting the allicin and soluble solid content of garlic with variable selection algorithms and chemometric models

Rahman, Anisur,Faqeerzada, Mohammad A,Cho, Byoung‐,Kwan John Wiley Sons, Ltd 2018 Journal of the Science of Food and Agriculture Vol.98 No.12

<P>CONCLUSION: The present study clearly demonstrates that hyperspectral imaging combined with an appropriate chemometrics method can potentially be employed as a fast, non-invasive method to predict the allicin and SSC in garlic. (C) 2018 Society of Chemical Industry</P>

Quality Analysis of Stored Bell Peppers Using Near-Infrared Hyperspectral Imaging

Rahman, Anisur,Faqeerzada, Mohammad Akbar,Joshi, Rahul,Lohumi, Santosh,Kandpal, Lalit Mohan,Lee, Hoonsoo,Mo, Changyeun,Kim, Moon Sung,Cho, Byoung-Kwan American Society of Agricultural and Biological En 2018 Transactions of the ASABE Vol.61 No.4

<P>Abstract. The objective of this study was to predict the moisture content (MC), soluble solids content (SSC), and titratable acidity (TA) content in bell peppers during storage (18°C, 85% relative humidity) over 12 days, based on near-infrared hyperspectral imaging (NIR-HSI) in the 1000-1500 nm wavelength range. The mean spectra of 148 mature bell peppers were extracted from the hyperspectral images, and multivariate calibration models were built using partial least squares (PLS) regression with different preprocessing spectra techniques. The most effective wavelengths were selected using the variable importance in projection (VIP) technique, which selected optimal variables for the target quality parameters of bell peppers from a full set of variables. Subsequently the selected variables were used to develop a PLS-VIP model for simplifying the prediction model. The MC, SSC, and TA content in bell peppers during storage changed from 90.7% to 93.0%, from 6.1%Brix to 7.3%Brix, and from 0.222% to 0.334%, respectively. The PLS regression model with MC, SSC, and TA content resulted in coefficients of determination (R<SUP>2</SUP>pred) of 0.83, 0.85, and 0.7, with standard errors of prediction (SEP) of 0.08%, 0.075%Brix, and 0.013%, respectively, using SNV preprocessed spectra for MC and TA content and Savitzky-Golay (S-G) second-order derivatives preprocessed spectra for SSC of bell peppers. By contrast, the prediction results yielded R<SUP>2</SUP>pred of 0.69, 0.75, and 0.68, respectively, with SEP values of 0.103%, 0.107%Brix, and 0.011% when the PLS-VIP model was employed. The PLS-VIP model simplified the calibration model by selecting the most important variables in terms of their responsiveness to bell pepper quality properties. The results revealed that HSI coupled with multivariate analysis can be used successfully to predict the MC, SSC, and TA content in bell peppers. Keywords: Fruit quality, Hyperspectral imagery, Image analysis, Spectral analysis, Stored bell pepper.</P>