AN IMAGING APPROACH FOR INTRAWALL INSPECTIONS

In the last years, the growing demand for "physical" security assessment techniques has increased the attention on noninvasive diagnostic methods capable of detecting the presence and characterizing the morphology of non-accessible objects concealed into buildings or other man-made structures. Among the other applications, one which is particularly relevant is the intrawall imaging, where the pursued goal is typically that of detecting targets (f.i., weapons and/or drugs) hidden into a cavity or a gap embedded into a wall. The problem of imaging non-accessible objects can be addressed by means of microwave imaging methods, wherein one aims at retrieving the unknown features from the measures of the backscattered fields corresponding to known incident fields. However, the mathematical difficulty of the underlying inverse scattering problem entails that suitable processing tools have to be developed in order to achieve reliable results. In this communication, we propose a novel and effective microwave method for intrawall imaging, in which one decomposes the problem into two parts. First, the measured data are processed to retrieve the unknown location and shape of the cache wherein the target are possibly hidden, then, by taking advantage of this acquired knowledge, the same data are exploited to reveal the presence and possibly the shape of the concealed objects. The method we propose is based on the Linear Sampling Method (LSM) [1]. Among the several strategies proposed to tackle the problem of determining the geometry of nonaccessible targets, the LSM is worth being considered, due to its general applicability and computational effectiveness. As a matter of fact, LSM can work for single or multiple dielectric and/or metallic targets and provides their geometrical features through an indicator function, which attains low values inside the objects and high values elsewhere. Such an indicator is readily obtained in each point of the investigated domain by solving a linear system, whose kernel is given by the measured scattered fields data and whose right-hand side is the Green's function pertaining to the considered reference geometry (homogeneous space, layered medium, etc.). However, despite its advantages, LSM cannot discern between electrically homogeneous and inhomogeneous targets. Therefore, as the whole target constituted by the unknown cache with possibly hidden inclusions indeed corresponds to an electrically inhomogeneous object, one can easily understand that LSM is not suitable for those intrawall imaging applications wherein one both aims at discovering caches and revealing their content.