Joint wall clutter mitigation and data-driven model for through-wall detection. Issue 18 (16th September 2016)

• Record Type: Journal Article
• Authors: Wang, F. F.; Zhang, Y. R.; Zhang, H. M.
• Is Part Of: International journal of remote sensing. Volume 37:Issue 18(2016)
• Page Start: 4486
• Page End: 4499
• Publication Date: 2016-09-16
• Subjects: Remote sensing -- Periodicals; Télédétection -- Périodiques; 621.3678
• DOI: 10.1080/01431161.2016.1213924
• Languages: English
• ISSNs: 0143-1161

Abstract

ABSTRACT: Wall effects such as the strong front wall return and the propagation of the signal as it travels through the wall must be taken into account in through-wall detection. First of all, compensation of through-wall propagation effects such as attenuation and dispersion should be considered in imaging procedure assuming effective removal of the strong front wall return (called wall clutter). However, the wave backscattered from the target behind the wall is usually submerged in the wall clutter. In addition, the wall parameters are unknown in practical situation. Therefore, how to detect a target behind the wall without prior knowledge of wall characteristics is becoming a new challenge that is difficult to solve. In this article, a novel method jointing wall clutter mitigation and data-driven model (DDM) is proposed to address this new challenge. First, the total echo is decomposed into wall, target, and noise subspaces using singular value decomposition (SVD), while the image entropy is introduced as an evaluation criterion for selecting the sub-spaces boundary, and therefore the wall clutter is mitigated. Then, through-wall detection problem is cast as a regression one using DDM model and the regression problem is solved by means of Support Vector Machine (SVM). It is demonstrated that the proposed method can efficiently detect the target behind the wall without prior knowledge of its characteristics.