Automatic Target Detection Technology Of Ultra-wideband Through Walls Radar

Ultra-wideband through walls radar has good range resolution and penetratingability and achieves a wide range of applications in the anti-terrorism, rapid response,military and disaster rescue. However, currently, the most investigation ofultra-wideband through wall radar systems are limited to the system design as well asthe radar imaging, automatic target detection technology is lagged behind so that theradar image interpretation and interpreted mainly rely on visual inspection of theoperating personnel. But, quick and accurate location and identification of human targetare highly desired in military and disaster rescue, thus this paper takes automatic targetdetection technology as the focus of the study.First of all, we introduce an ultra-wideband through-wall radar system and itsmoving target indicator technology. The moving target indicator, including a couplingalignment, background cancellation, and RFI suppression, is the basis for automatictarget detection. Here, the primary objective of coupling alignment is compensatingsystem waveform jitter and improves the performance of suppressing strong couplingclutter of background cancellation. Background cancellation is a common clutterelimination method. In this article, three kinds of background cancellation methods,accumulated average of background, pulse cancellation, as well as index-weighted, areanalyzed and unified from the time domain and frequency domain respectively andultimately choose index weighting method with satble performance. Finally, accordingto RFI feature in range history image of through wall radar, a mean normalizationmethod along fast time dimension is proposed.When it comes to the preliminary target detection, the CFAR methos are oftenconsiderd for through wall radar. This paper describes a CFAR method along slow timedimension, its adaptive improved method and an OSCFAR method along rangedimenstion based on fourth powers data, repectively. Theoretical analysis andexperimental results show that, OSCFAR based on the fourth power data has a morestable performance on the detection of the target under different motion states andmultiple targets situationTarget detection is only the first step of automatic target detection; there are somefalse alarms and missed alarms in detection image. In order to achieve accurateinformation of the target number and location, the clustering processing and target tracking are further considered. To relieve the computation burden of clustering andtracking, we first eliminate the noise false alarm using binary accumulation, then takeadvantage of the split-level clustering method to suppress sidelobes false alarm andachieve target centroid, finally a target tracking algorithm is proposed to suppressrandom strong clutter false alarms, and distinguish different target trajectorys makingthe automatic detection come ture.