Virtual Grader for Apple Qualityassessment using Fruit Size and Illumiation Features
The present paper reports on the development of an intelligent virtual grader for assessing apple quality using machine vision. The heart of the proposed virtual grader was executed in the form of K-Nearest Neighbor (K-NN) classifier designed on the architecture of Euclidean distance metric. KNN classifier is executed for this particular application due to its robustness to the noisy environment. The present study revealed that fruit surface illumination is one of the major deterministic parameters affecting accuracy substantially while assessing apple quality based on fruit size. The performance of the proposed virtual grader was examined experimentally under different conditions of fruit surface illumination. An industrial grade camera connected to an image grabber was used to implement the proposed industrial-grade virtual grader using machine vision. Results of this study are quite promising with an achievement of 99% efficiency at 100% repeatability when fruit surface is exposed to an optimal value of 310 lux. However, such an attempt has not been made earlier.
