Object recognition using artificial neural networks

In recently, the studies on the object recognition by using artificial neural networks are very increased. In the solution of object recognition problems under comlicated and fuzzy data, the success of the decision principle of the artificial neural networks by learning has been effective on these studies. Artificial neural networks has the peculiarity that compares given objects with the objects which it has learned before and finds the similiriaties between them to classify the objects into certain groups. The object recognition is not a different process from the objects classification in basic. In this thesis; basic object recognition approaches are examined and the recognition of the distorted (incomplete or noisy) objects is implemented by using supervised and unsupervised artificial neural networks models. In addition, an object recognition system, regardless rotation and position variation is developed. The developed system is set up on the derivation of object's feature vector after some preprocesses and the steps of object recognition by classifying and according to the obtained vector. The object's feature vector extraction by using both local and global methods are classified with the artificial neural networks and successfull results are obtained. KEY WORDS : Object Recognition, Artificial Neural Networks, Pattern Classification, Feature Exraction.