A Method for Size and Shape Estimation in Visual Inspection for Grain Quality Control in the Rice Identification Collaborative Environment Multi-agent System

Computer vision methods have so far been applied in almost every area of our lives. They are used in medical sciences, natural sciences, engineering, etc. Computer vision methods have already been used in studies on the search for links between the quality of raw food technology and their external characteristics (e.g. color, size, texture). Such work is also conducted for cereals. For the analysis results to meet the expectations of users of the system, it should include not only the attributes describing the controlled products, materials or raw materials, but should also indicate the type of material or species/variety of raw material. However existing solutions are very often implemented as closed source software (black box) therefore the user has no possibility to customize them (for example the enterprise cannot integrate these solutions into its management information system). The high cost of automated visual inspection systems are also a major problem for enterprises. The aim of this paper is to develop a method of estimating the size and shape of a rice grains using visual quality analysis, implemented in the multi-agent system named Rice Identification Collaborative Environment. Using this method will allow statistical analysis of the characteristics of the sample, and will be one of the factors leading to the identification of species/varieties of cereals and determining the percentage of the grains that do not meet quality standards. The method will be implemented as an open source software in Java. Consequently it can be easily integrated into enterprise’s management information system. Because it will be available for free, the cost of automated visual inspection systems will be reduced significantly. This paper is organized as follows: the first part shortly presents the state-of-the-art in the considered field; next, a developed method for size and shape estimation implemented in the Rice Identification Collaborative Environment is characterized; the results of a research experiment for verification of the developed method are presented in the last part of paper.