MULTI RESOLUTION BASED IMAGE REGISTRATION TECHNIQUE FOR MATCHING DENTAL X-RAYS

Abstract

This paper presents a new technique to assess the bone formation after periapical dental surgery. The proposed technique consists of two main stages: image registration and spectral subtraction stages. Image registration is used to avoid projection errors produced due to nonstandardization of X-ray scanners. Wavelet coefficients are used instead of grey values for registering the images. Coarse to fine strategy with four levels of resolutions is used to speed up the process. The second stage is the spectral subtraction stage. It is used to yield the difference image between pre- and post-operative images which represents the bone gain or bone loss with light and dark areas, respectively. Algorithm has been applied on a number of pre- and post-surgery intra oral periapical (IOP) dental X-ray images. Mean and root mean square error (RMSE) are computed to assess the quality of registration technique. The technique presented here is compared with grey level based method; results show that proposed technique outperforms conventional grey level method based on dyadic sampling.