Ultrasonic scatterer structure classification with the generalized spectrum

Ultrasonic back-scattered echoes resulting from the structures within a scanned object contain information of potential diagnostic value. The most common nondestructive evaluation (NDE) techniques use large-scale changes in the back-scatterer coefficients to reveal boundaries between materials with different density/elasticity properties or defects in homogeneous material regions. Less common techniques consider small-scale scatterer characteristics that give rise to textures and other features not readily seen in the A-scan envelope or intensity image. This paper considers applying the generalized spectrum (GS) for classifying small-scale scatterer structures into three broad categories, diffuse, specular, and regular. The GS distinguishes between stationary (diffuse scattering) and certain classes of nonstationary processes based on a statistical characterization of the phase spectrum, and the GS can be normalized to limit variations due to frequency selectivity of the scatterers and the ultrasonic propagation path. This paper explains how the GS can be applied to classify scatterer structures over small sections of the ultrasonic A-scan and demonstrates its classification performance with simulations. The significance of the approach to NDE applications, such as flaw detection in homogeneous material and material characterization in more complex material, is also discussed.