Comparison of k-NN, SVM, and NN in Pit Pattern Classification of Zoom-Endoscopic Colon Images using Co-Occurrence Histograms

Abstract

Co-occurrence histograms are used as features to classify magnifying endoscope imagery with k-NN, SVM, and NN classifiers. In the k-NN classification case these histograms may improve the classification accuracy of simple ID color histograms up to 10% in the 2 classes case and up to 5% in the 6 classes case. The classification results of SVM and NN classifiers have turned out to be noncompetitive and do not improve the classification result of ID color histograms.