Separation of touching almonds and their quality classification by combining image processing and artificial neural networks techniques

The quality evaluation of agricultural products is one of the key factors in promoting their quality. In this study, a method based on combined image processing technique and artificial neural network was presented. Separation of touching almonds under different positions is a very important step in design of grading devices. In this study, an image processing algorithm based on extracting critical points in the image of almonds and drawing segmentation lines between them is presented. In the next step, the feature vector which includes 6 shape features, 36 color features and 36 texture features was composed. PCA method was used to reduce the dimension of the feature vector. The quality classification of almond in different classes was carried out by artificial neural networks (ANNs). Among different ANN structures, the 18-7-7-4 topology was the most optimum classifier (total accuracy was obtained 96.92%).