The Presentation of a New Method for Image Distinction with Robot by Using Rough Fuzzy Sets and Rough Fuzzy Neural Network Classifier

摘要

Distinguishing different images by robots and classifying them in distinct groups is an important issue in robot vision. In this paper we want to propose a new method for distinguishing images by robot via using Rough fuzzy sets’ decreases method and Rough fuzzy neural network classifier. In this method, the image features like color, texture and shape are excluded and the redundant features are decreased by Rough fuzzy sets method. Then the Rough fuzzy neural network classifier is educated by the use of these decreased features. In next phase, the robot can properly classify the images; it has not seen or the examined images and put them in the correct group by the use of this system. We have compared our proposed method with Johnson decreased method, principal component analysis, and Rough sets, and also we have compared our classifier with the support vector- machine classifier, neural network and K-nearest neighbor. Our tests ’ bed is 1000 images of the COREL image set in ten semantic groups.