Computer Vision, Pattern Recognition, Image Processing and Graphics, National Conference on
Download PDF

Abstract

We propose a dental classification system to effectively classify molar teeth from premolar teeth in dental bitewing radiographs. Our system includes a novel image enhancement method that combines homomorphic filtering to reduce the uneven exposure problem, and both adaptive contrast stretching and adaptive morphological transformations to accentuate the texture of gums and pulps. We also propose using relative length/width ratios of both boundaries of a complete tooth and a pulp, as well as the relative crown size as three features for SVM classifier. The experimental results show that our classification system can classify both molars and premolars in both maxilla and mandible with an accuracy rate of 93.9%, 95.7%, 98.6%, and 91.9%, respectively from 45 dental bitewing images. The results also show that our system correctly classifies two images that were reported misclassified and has higher premolar classification rates in both jaws than the methods presented in a published article.
Like what you’re reading?
Already a member?
Get this article FREE with a new membership!

Similar Articles