Wavelet based segment detection and feature extraction for 3D T-ray CT pattern classification

Abstract

This paper explores three dimensional (3D) Terahertz (T-rays) computed tomographic (CT) classification based on T-ray functional imaging techniques. The target objects are separated by their refractive indices, which are indicated by the intensity in the images. Segmentation techniques are employed to identify the position of each pixel belonging to the different classes. Wavelet methods are applied to the detected T-ray pulsed responses for feature extraction. A Mahalanobis distance classifier is selected for the final classification task. This paper presents T-ray CT classification techniques that allow analysis of measured T-ray transmission image statistics and that automatically identify materials within a heterogeneous structure.