CSBE-SCGAB

Detection of fungal infection in pulses using near infrared (NIR) hyperspectral imaging

Authors: Karuppiah, K., D. Jayas, N. White
Description: Chick pea is one of the major pulse crops in Canada. Fungal infection remains a major concern during storage of pulses, particularly Aspergillus and Penicillium species. Infections in pulses by fungi cannot be detected visually until the infection is severe. Chick pea samples were artificially infected with Aspergillus and Penicillium species and images were taken at intervals of two, four, six, eight and ten weeks of infection using a NIR hyperspectral imaging system in the wavelength range of 960 to 1700 nm. The spectral and histogram (spatial) features of these hypercubes were extracted and classification models were developed to predict infection levels using supervised training algorithms namely linear discriminant analysis (LDA) and quadratic discriminant analysis (QDA). For Aspergillus species both LDA and QDA showed 100% classification accuracy for the healthy samples and for the infected samples the accuracies ranged from 80 to 95% for each stage of infection. For Penicillium species LDA and QDA showed 100% classification accuracy for the healthy samples and for the infected samples the accuracies ranged from 85 to 95%.
Keywords: pulses,fungal infection, NIR hyperspectral imaging
Technical field: Information Systems Engineering
Conference name: CSBE/SCGAB 2015 Annual Conference, Edmonton, AB, 5-8 July 2015.
Session name: Poster
Citation: Karuppiah, K., D. Jayas, N. White. 2015. Detection of fungal infection in pulses using near infrared (NIR) hyperspectral imaging. CSBE/SCGAB 2015 Annual Conference, Edmonton, AB, 5-8 July 2015.
Publisher: Canadian Society for Bioengineering
Date: 2015-07-05
Publication type:
  • Poster
Type: Text.Article
Identifier: CSBE15100
Coverage: Canada
Language 1: en
Rights: Canadian Society for Bioengineering

Search a publication

Search for:
Find entries that have:
Date:
From:
To:
Publication type: