Quality Estimation of Canola Using Machine Vision and Vis-nir Spectroscopy

Abstract

Canola is mainly graded either by visual inspection or by smelling. These methods are subjective in nature and are bound to cause errors while deciding the grade of canola. To test canola for amount of erucic acid present the sample needs to be sent to a laboratory for testing through wet chemical analysis. This is a time consuming process. An electronic method that can quantify amount of dockage, presence of distinctly green and heat treated seeds, distinguish samples on the basis of erucic acid, its free fatty acid content and PV, would not only be less time consuming but also would be a more reliable method to grade canola samples. Findings and Conclusions: 1. Canola samples cannot be classified on the basis of total dockage present using L and RGB data obtained from flat-bed scanner. Inclusion of morphological and textural features would improve the classification accuracy. 2. Machine vision can be considered as a potential method to grade canola on the basis of good, distinctly green and heat damaged