A Fuzzy Decision Tree Classifier for Recognition of Fuzzy Image Regions

Document Type : Original Article

Authors

1 College of Engineering, Al-Azhar University, Cairo, Egypt.

2 Egyptian Armed Forces.

Abstract

Abstract:
In this paper we present a framework for image segmentation and fuzzy region
recognition based on fuzzy decision tree (FDT) classifier. The inference of the fuzzy
decision tree starts by construction of crisp classification and regression tree (CART),
then fuzzification of decision boundary at each node. Fuzzy regions in digital images
are extracted by fuzzy c- means algorithm. Then fuzzy regions are matched with fuzzy
tree models constructed for sample images from the application domain. This procedure
was applied on MRI sample images for identification of normal and abnormal tissues. It
was also applied for diagnosis of pathological brain cases by identification of shape and
tissue of fuzzy regions. Experimental results show low error rate for the cases under
study. Comparison of error rates obtained by this FDT classifier and the CART
classifier shows a slight difference of 0.1 % between the two classifiers.

Keywords

Files

Share

How to cite

Statistics