Discrimination of Viable and Non-Viable Corns Using FT-NIR Technique Sampling Both Sides of the Kernel

( Collins Wakholi ),( Byoung Kwan Cho ) 한국농업기계학회 2015 한국농업기계학회 학술발표논문집 Vol.20 No.2

Rapid assessment of corn seed viability using short wave infrared line-scan hyperspectral imaging and chemometrics

Wakholi, Collins,Kandpal, Lalit Mohan,Lee, Hoonsoo,Bae, Hyungjin,Park, Eunsoo,Kim, Moon S.,Mo, Changyeun,Lee, Wang-Hee,Cho, Byoung-Kwan Elsevier Science B.V., Amsterdam. 2018 Sensors and actuators. B Chemical Vol.255 No.1

Current State of Postharvest Fruit and Vegetable Management in East Africa

Wakholi Collins,조병관,모창연,김문성 한국농업기계학회 2015 바이오시스템공학 Vol.40 No.3

Purpose: Fruit and vegetable production is a fast-growing sector in East Africa, and it bears considerable local andinternational market potential. In an effort to analyze the challenges within this sector and suggest possible solutions, thisstudy reviews the postharvest handling technologies commonly used with fruits and vegetables in East Africa. Methods:During the course of this study, small-scale farmers were identified as the most prominent producers of fruit and vegetablecrops in the region. Results: We found that many of these small-scale farmers employed relative simple and inexpensivetechniques in handling their limited volumes of produce. Several factors could be addressed to reduce postharvest losses,including weak policies, inferior infrastructure, and poor market strategies. However, the lack of basic knowledge (includingdemographic, scientific, and economic knowledge) among the stakeholders (e.g., researchers, farmers, governments,nongovernment organizations, and merchants) on how to develop, implement, use, and sustain the recommended handlingtechnologies is probably the most problematic. Conclusions: We recommend that high priority be placed on closing theknowledge gap, which could enhance the efforts of all the stakeholders to address and reduce postharvest losses.

Development of an Online measurement of corn seed viability using hyperspectral imaging

( Collins Wakholi ),( Youngwook Seo ),( Byoung-kwan Cho ) 한국농업기계학회 2016 한국농업기계학회 학술발표논문집 Vol.21 No.2

The ability to classify seeds according to their respective traits like viability, variety etc. is key to regulation and ensuring seed quality and purity. Since the mass inception of genetically modified (GM) seeds to the market, many undesirable effects have been observed and are able to pose danger to the natural ecosystem and to humans at large. On top of the undesirable effects, there is a need to keep the natural gene pool as pure as possible from mixing with GM genes thus calling for careful caution during the handling of GM organisms. Currently many regulations have been imposed on the trade and import of GM seeds. In some societies for instance, the imported GM seeds have to be either killed or traded under very careful certification conditions. Clearly, there is a need to develop cutting edge technology to detect Viable GM seeds before import or trade. The main purpose of this research was to develop a working online sorting system based on Short Wave Infrared (SWIR) hyperspectral imaging (HSI) that is able to detect and sort GM corn seeds according their viability. In this study, hyperspectral image data from 600 corn seeds (50% microwave treated, 50% as control) was collected, analyzed and used to build classification models to classify between treated (non-viable) and non-treated (control/viable) corn groups. Three classification models were tested which included Partial Least Squares-Discriminant Analysis (PLS-DA), Support Vector Machines (SVM) and Linear Discriminant Analysis (LDA). The PLS-DA model coupled with some image processing tools (2-D median filter and morphological operations) performed the best and was applied to the online sorting system and tested its performance at different conveyor belt speeds until optimum. The overall system classification accuracy from the online tests was found to be 84% and 81% for white and purple corn respectively at conveyor belt speed of about 72 mm/s. The overall system accuracy was found to be lower than that produced from the offline spectral analysis possibly due to effects like ambient temperature, electronic noise, test sample dissimilarity, etc. The results show that the hyperspectral based online sorting prototype is able to be used for non-destructive real-time inspection of corn seed viability which could be faster and cheaper alternative to the conventional measurement methods which are expensive, time consuming and representative.

Review : Agricultural Process and Food Engineering ; Current State of Postharvest Fruit and Vegetable Management in East Africa

( Collins Wakholi ),( Byoungkwan Cho ),( Changyeun Mo ),( Moon S. Kim ) 한국농업기계학회 2015 바이오시스템공학 Vol.40 No.3

Purpose: Fruit and vegetable production is a fast-growing sector in East Africa, and it bears considerable local and international market potential. In an effort to analyze the challenges within this sector and suggest possible solutions, this study reviews the postharvest handling technologies commonly used with fruits and vegetables in East Africa. Methods: During the course of this study, small-scale farmers were identified as the most prominent producers of fruit and vegetable crops in the region. Results: We found that many of these small-scale farmers employed relative simple and inexpensive techniques in handling their limited volumes of produce. Several factors could be addressed to reduce postharvest losses, including weak policies, inferior infrastructure, and poor market strategies. However, the lack of basic knowledge (including demographic, scientific, and economic knowledge) among the stakeholders (e.g., researchers, farmers, governments, nongovernment organizations, and merchants) on how to develop, implement, use, and sustain the recommended handling technologies is probably the most problematic. Conclusions: We recommend that high priority be placed on closing the knowledge gap, which could enhance the efforts of all the stakeholders to address and reduce postharvest losses.

Beef Carcass Yield Estimation Using 3D Reconstruction From Structured Light and Multi-view Geometry

( Collins Wakholi ),( Kyungdo Kwon ),( Changyeun Mo ),( Wang-hee Lee ),( Byoung-kwan Cho ) 한국농업기계학회 2018 한국농업기계학회 학술발표논문집 Vol.23 No.1

Reconstruction of 3D models from real objects is gaining popularity in computer vision and graphics nowadays. This is because it has numerous applications in areas such as industrial inspection, autonomous driving, reverse engineering, multimedia. However, 3D reconstruction is complex and often limited by factors such as sensor specifications, computer resources and expertise. While much efforts have been done to refine the reconstruction algorithms, not much has been done to optimize the scanning technologies. Advances in 3D scanning technology have introduced new techniques to obtain depth information from a scene such as multi-view geometry, structured light and laser scanning. Obtaining 3D information using structured light scanning is become popular because of its precision while using multi-view geometry is relatively low cost but still lacks in accuracy. Since all 3D scanning methods have advantages, the pursuit for accurate resultant 3D models at low cost calls for integration of the scanning technologies. This study therefore aims to develop a 3D system using 1) multi-view structured light scanning, MVSLS and 2) multiple view geometry for accurate beef carcass yield estimation. For the MVSLS-based system, a trinocular system (involving two cameras matched to a single projector) was used and a 360-degree scan of the samples was done in intervals of 45 degrees. The multiple view photogrammetry-based 3D scanning system employed two cameras and independent light sources to capture a 360-degree scan of the samples in intervals of 10 degrees. Using multi-view reconstruction algorithms and data from both setups, point clouds were obtained from matched features that were reconstructed into visual 3D models. Evaluation and comparison of the developed models was done based on the accuracy of volume estimation and reproduction of planar surface texture. The 3D model from best method was selected to estimate the volume and subsequently weight and price of the beef carcass. This study serves as a precursor to the development of a real-time beef carcass yield estimation method for industrial use.

Comparison between Tetrazolium and Germination Test Results for Melon Seed Viability Determination using Hyperspectral Imaging

( Collins Wakholi ),( Byoung-kwan Cho ) 한국농업기계학회 2018 한국농업기계학회 학술발표논문집 Vol.23 No.1

Tetrazolium (TZ) test is one of the most reliable and widely used method in research to determine seed viability. The TZ test is able approximate seed viability through determining whether a seed’s tissue is still biochemically active or not. Despite the accurate prediction of viability by germination tests, the TZ test is preferred because it is relatively quick compared to germination tests that take anywhere between days to weeks to acquire results. The objective of this study was to compare the TZ test with germination tests (proved to give a realistic picture of seed viability) results using Short wave infrared, SWIR Hyperspectral imaging, HIS data and multivariate data analysis. For this study, damage-free melon (cucumis melo) seeds were used and were randomly divided into two groups of 1000 seeds each. Hyperspectral imaging data of both groups were collected using the same settings and under same environment. One group was subjected to germination tests while the other to TZ test to determine the viability of the seeds. The performance of the multivariate classification models achieved based on germination tests and TZ test results were compared. A clear difference in the classification model performances for germination and TZ test based models were observed. The results of this study serve as a key foundation in the development of a real-time sorting system based on HSI for melon seeds.

Determination of Viability of Seedless Watermelon (citrullus lanatus) Seeds Using Hyperspectral Imaging and Chemometrics

( Collins Wakholi ),( Byoung-kwan Cho ) 한국농업기계학회 2017 한국농업기계학회 학술발표논문집 Vol.22 No.2

Hyperspectral imaging (HSI) has emerged as a powerful technique in non-destructive quality and safety inspection in the food industry. This is attributed to its ability to simultaneously detect the composition (chemical, structural and functional information) and distribution of particular traits in a given sample. Hyperspectral imaging (HSI) technology has been successfully employed in the non-destructive determination of internal quality parameters of seeds with a promise of revolutionizing industrial sorting applications. Despite the limitations, this study focuses on employing short wave-infrared (SWIR) HSI to predict the true viability of seedless water melon (citrullus lanatus) seeds. For this study, data from 1248 damage-free seeds were taken using a SWIR HSI camera (wave range 1000~2500 nm). Afterwards, the seeds were subjected to germination tests to determine their real viability. The data were modeled using multivariate data analysis methods (partial-least square discriminant analysis, PLS-DA and support vector machines, SVM) to optimally discriminate viable from non-viable seeds. Model building was done using 70% of the data, the rest were used for validation. The PLS-DA model was able to achieve the best accuracy of 74% and 72% in calibration and validation respectively. The low accuracy achieved was probably due to the complexity of modeling the natural causes of loss of seed viability or even due to the thick seed coat (deters spectral signal penetration). The future of this research will focus on improving classification accuracy and implementing the resultant model to a real-time, high-speed sorting system based on HSI detection technology.

Erratum to: Nondestructive Estimation of Lean Meat Yield of South Korean Pig Carcasses Using Machine Vision Technique

Lohumi, Santosh,Wakholi, Collins,Baek, Jong Ho,Kim, Byeoung Do,Kang, Se Joo,Kim, Hak Sung,Yun, Yeong Kwon,Lee, Wang Yeol,Yoon, Sung Ho,Cho, Byoung-Kwan Korean Society for Food Science of Animal Resource 2019 한국축산식품학회지 Vol.39 No.3

On-line Spectral Measurement System for Pineapple Quality

( Semyalo Dennis ),( Collins Wakholi ),( Byoung-kwan Cho ) 한국농업기계학회 2021 한국농업기계학회 학술발표논문집 Vol.26 No.2

Translucency is a physiological disorder undesirable in Pineapples. In this condition, the internal flesh of pineapple fruit shows water-soaking appearance, is prone to diseases, and is susceptible to mechanical damage, thus significantly reducing its quality. The study was aimed at developing a on-line spectral measurement software to identify the translucency level in pineapples. The software was developed in C++ programming language for fast processing during on-line measurements and analysis. A total of 172 pineapple fruits were used for on-line spectral data collection with a spectral range of 200-1100 nm. The reference data for translucency level was obtained from the RGB color images of internal flesh of pineapple fruit. The level of translucency was graded on a scale of 1 (no translucency) to 3 (very translucent). A partial least squares-discriminant analysis (PLS-DA) model was created using a C++ custom developed software. A sample size of 138 and 34 pineapples were used as the calibration and validation sets respectively. The PLS-DA classification accuracy achieved for the calibration and validation set was 79% and 78% respectively. The study indicates that there is potential to identify the level of translucency for intact pineapples in real-time.