Development of Artificial Vision System for Quality Assessment of Oyster Mushrooms

In this paper, we have illustrated a simple and effective method of assessing the fresh and deteriorated oyster mushroom samples. Analysis of correlation coefficients is done very effectively here to identify the interrelationship of different color layers of the RGB and HSV color map of the samples, as this degrades progressively. These correlation features are further analyzed using two supervised learning models incorporating support vector machine (SVM) and artificial neural network (ANN) to develop a simple classifier model to identify the two different classes of the mushroom images. The proposed work is simple as it includes a simple computational supervised model, and, more importantly, it is very effective as it explores the interrelation between the different layers of two different color maps of the image samples. The highest classifier accuracy achieved using ANN and SVM models exceeds 95% and 98% respectively; and the mean accuracy level from fifty such observations stands 94.7% and 93.4%, which are high considering contemporary researches. High accuracy of classification, simplicity of analysis and image acquisition through smartphones makes the proposed work suitable for implementing in application-based software for smartphone users which would further enhance the wide applicability of the proposed mushroom classifier scheme.