Separating Agricultural Goods with Image Processing and Fuzzy Inference Systems

Abstract

Fuzzy logic inference systems that give effective results about the model in systems that are difficult to establish a model; respectively, it starts with defuzzification the data structurally. Later, it enables the development of approaches to model solutions by extracting rules and then clarifying data. Membership functions, inference methods and performances used are important points that affect the validity and reliability of the system. In the production sector, where modeling is partially difficult, classification of products according to their quality allows the application of fuzzy inference systems. Classification according to quality parameters has been studied extensively in the literature. Recently, artificial intelligence studies have made image processing-based quality classifications possible. There are many membership functions and inference methods with different structures. In this study, image processing data have been classified using artificial intelligence-based image processing algorithms and Fuzzy Inference System (FIS) Algorithms. With this study, it is aimed to bring a new approach to the literature in the quality processes of agricultural products by combining fuzzy logic algorithms and image processing technology. Performance analyzes of fuzzy logic inference parameters have been made with Python under different operating conditions. The obtained results have been examined and interpreted. Satisfactory results have been obtained in the first phase analyzes and studies are continuing. This study has been developed by IND Information Technologies and Fersan within the scope of TÜBİTAK 1507 project.