CSBE-SCGAB

Classification of date varieties using statistical classifiers

Authors: Annamalai, M, A-Y Rashid, G Thomas
Description: Quality assurance is becoming an important criterion in the success of marketing in both domestic and international sector. Computer vision technology provides an accurate, objective and online quality measurement system. Color Image based classification procedures were developed to discriminate three date varieties: Khalas, Fard and Madina. The classification efficiency of linear discriminant function (LDF) and quadratic discriminant function (QDF) was determined with 65 color features. The average classification accuracies of LDF were 97%, 94% and 88%, while using all 65 features, top 10 contributing features and top 5 contributing features, respectively. Similarly, it was 98%, 93% and 88% while using 65 features, top 10 and top 5 features, respectively with QDF. Implementation of automated computer vision system would be beneficial in the packaging of dates with varietal purity.
Keywords: computer vision, dates variety, statistical classification
Technical field: technical_fields_app4
Session name: Food and bioprocess engineering
Date: 2012
Identifier: CSBE12132
Coverage: Oman

Search a publication

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