Identification of a Landmark in a Roentgenographic Cephalogram by Employing the Wavelet Neurons

This paper describes an identification method of a landmark in a roentgenographic cephalogram by employing the input-correlated wavelet neurons. For the purpose of improvement of identification precision, a novel pattern matching method, named "wavelet neuron matching (WNM)," is proposed in this paper. Furthermore, the "weighted window", which is proposed in this paper, enables us to consider the orthodontists' knowledge on local information as precisely as possible. The effectiveness and the validity of the proposed method have been verified by the experiments to identify a landmark called Menton.