Compressed Sensing Techniques for Ultrasonic Imaging of Cargo Containers

Abstract

This contribution presents a compressed sensing (CS)-based ultrasonic imaging system for fast, low-cost inspection of metallic cargo containers. The idea is to detect the footprint of metallic objects within the container that can be used to conceal smuggling goods. This ultrasonic technology can complement currently deployed X-ray-based radiographic systems and millimeter-wave scanners, thus increasing the probability of detection. The proposed hardware consists of an array of acoustic transceivers that is attached to the metallic structure of the metallic cargo container to create a guided acoustic wave. Variations in the thickness of the metallic structure create reflections that can be located by backpropagating the measured reflected wave. Aiming to reduce the number of acoustic transceivers, this contribution evaluates the feasibility of applying CS techniques in the proposed acoustic imaging system. It has been observed that in the majority of the cases, the acoustic images retrieved by the cargo inspection system are sparse, that is, only those image pixels corresponding to discontinuities in the metallic plate (due to gaps, joints, placement of a metallic object on it) are different from zero. Thus, sparsity condition, which is one of the CS requirements, is satisfied for this particular application. A simulation-based example resembling a real case of cargo inspection is considered for validation purposes. A comparison between standard backpropagation and CS for different number of samples is presented, proving that CS is able to recover the acoustic image with as few as 10% of the samples required by Nyquist sampling rate.