عنوان مقاله [English]
Today, development of intelligent systems in different operations of food and agricultural processing is one of the prime requests in food industry. In this research, a series of machine vision tests, were carried out to investigate the factors influence on walnut kernel discriminant system based on dimension and color (standard method). The research had been focused on discriminating the kernels in 3 quality groups of halves, pieces and small pieces and 3 color groups of Very Light, Light and Light Amber in a common variety. The prediction features were the major and minor diameters of kernel, and color features of Red, Green, Blue, Hue, Saturation, Value, L, a and b channels from three color models and the direction of lighting. In comparison of two lighting methods, back lighting showed more accuracy (94.3%) than the top lighting (91%), however, the difference was not significant. In addition, it was possible to extract the dimension and color data in image capturing in top lighting. The results showed that the accuracy and speed of discrimination based on dimensions was more than that the color identifying. According to the results, it was possible to identify halves with 98.1% accuracy in a mean of 0.31 s. While, the maximum rate of discrimination in color identification was equal to 76.2% in 1.91 s for detecting Light kernels. Due to overlapping of data of different color models, and according to the results of linear discriminant analysis, it is possible to identify Very Light kernels, only by taking account of V parameter (from HSV model) in less than 0.6 seconds with accuracy of 81%. In comparing different color models, the HSV and Lab have the maximum and minimum accuracy in color discriminating of walnut kernels, respectively. The results showed that it is possible to discriminate walnut kernels based on color and dimensional features according to standard method, in less than 2 seconds (for each kernel), under top lighting condition. This information may be used in design and development of walnut grading machines in food industries.