خط مشی دسترسیدرباره ماپشتیبانی آنلاین
ثبت نامثبت نام
راهنماراهنما
فارسی
ورودورود
صفحه اصلیصفحه اصلی
جستجوی مدارک
تمام متن
منابع دیجیتالی
رکورد قبلیرکورد بعدی
Document Type:Latin Dissertation
Language of Document:English
Record Number:55403
Doc. No:TL25357
Call number:‭MR50640‬
Main Entry:Chelladurai Vellaichamy
Title & Author:Identification of fungal infection in wheat using thermal imaging techniqueChelladurai Vellaichamy
College:University of Manitoba (Canada)
Date:2008
Degree:M.Sc.
student score:2008
Page No:110
Abstract:Wheat is the major cereal crop grown in Canada, and about 70% of its production is exported to various countries. Growth of fungi on the grain is the most common cause of spoilage of stored grain. The traditional fungal detection methods such as plate agar and microscopic detection techniques require about a week for detection and quantification. Early detection of fungal infection is necessary to carry out control methods to minimize the storage losses. The feasibility of the infrared thermal imaging system to identify the fungal infection in stored wheat was studied. Thermal images of bulk wheat grains infected by Aspergillus glaucus, Aspergillus piger and Penicillium spp. were obtained using an un-cooled focal planar array type infrared thermal camera after heating grain with a plate heater and cooling in ambient air for 180 and 30 s, respectively. In total, twelve temperature features were derived from heating and cooling data. Ten-way, three-way, pair-wise and infection level based classification models were developed by linear and quadratic discriminant analyses using the derived temperature features. Classification accuracies of 90-100% and 95.5-99% were obtained for healthy and fungal-infected samples, respectively, using pair-wise linear discriminant analysis (LDA) classifier from a three-week old infection. In quadratic discriminant analysis (QDA), classification accuracies were 86-100% and 94.5-98.69% for healthy and infected samples, respectively. In pair-wise comparisons between healthy and for each of the fungal species infected samples, the LDA classifier yielded an accuracy of >91% for healthy, >95% for A. glaucus -infected, and >94% for A. niger -infected samples from the 3 rd week of infection onwards. When classifying A. glaucus, A. piger and Penicillum -infected grains at different infection periods, both classifiers gave relatively low accuracies (25 to 71.9%) for leave-one-out and bootstrapping validation methods. Most of the misclassification happened between the fungal species at the same level of infection. The early detection of fungal infection, i.e., at 3 weeks of fungal growth helps to carry out the control measures to prevent the grain deterioration and quality loss. The results prove that the thermal imaging system has the potential for application in the grain industry to detect the kernels infected by fungi and the level of infection (low or high).
Subject:Applied sciences; Biological sciences; Aspergillus glaucus; Aspergillus niger; Penicillium; Agronomy; Agricultural engineering; 0359:Agronomy; 0539:Agricultural engineering
Added Entry:University of Manitoba (Canada)