Research On Through Wall Radar Extended Moving Target Detection And Tracking Algorithm

In recent years,the crime rates and terrorist incidents are rapidly growing which caused by the booming of urban population.Therefore,law enforcement officials need to know the movements of criminals indoor in buildings and securely rescue hostages.Ultra wide band radar has the characteristics of low power,high resolution and strong penetration.It is widely used in street combat,geological engineering,archaeology,post-disaster relief and life detection.Through-wall radar with the ultra wide band technology can accurately locate the indoor target position in the building and provide a good target trajectory for the user's.Through-wall radar have two core steps: one is target detection,and the other is target tracking.This paper proposes the following solutions to the problems encountered in the target detection and target tracking of through-radar.1.Antenna coupling and the wall reflection are strong in the echo signal of the through-wall radar which results the low signal-to-noise and low visibility of the target.In order to solve above problem,the paper proposes an adaptive exponential elimination algorithm before imaging.Firstly,using the previous moment echo and the corresponding index factor which takes advantage of the feedback of wall amplitude signal,decide the stationary clutter background.Then,the static clutter such as antenna coupling and wall reflection of echo signal at the current moment can be eliminated,and the signal-to-noise ratio of moving target at the current moment can be enhanced,the detection algorithm's workload can be reduced.2.According to the problem that strong clutter signal after the imaging of the echo signal leads to the low signal-to-noise of the target,the traditional moving target detection algorithm cannot completely eliminate the false alarm phenomenon caused by multipath and indoor random noise of the image.In this paper,an improved dynamic threshold ViBe detection algorithm is proposed based on the feature that the target position changes between frames.Firstly,a dynamic threshold is added in the pixel classification step of ViBe detection to eliminate part of multipath and make it adaptable to the changing problem of moving target pixel value in the through-wall radar image.Then,the moving target information is extracted to completely eliminate the "ghost" phenomenon caused by strong multipath clutter with three frame differences algorithm.Finally,the method of the connected domain is adopted to eliminate the isolated noise caused by indoor random noise in the image.The detection method caneffectively eliminate false pixel caused by indoor random noise,provide accurate and real-time target location information and has certain robustness through experimental analysis.3.Since through-wall radar extended target has multiple measured values while the traditional multiple target tracking algorithm considering the target has only one single measured value,the traditional multiple target tracking algorithm cannot handle multiple extended target measurements matching problem when the targets' tracking threshold overlap.This paper proposes a improved ML-PDA extended target tracking method.Firstly,Gaussian weight sequence is used to obtain the measurement of the target from the detection image.Then the measurements of the target threshold overlapping region are matched with the extended target through the K-Modes clustering algorithm.Finally,by using the distribution information of multiple measurements of the extended target,the likelihood ratio function of the target motion state is calculated and the real position of the target is estimated.Comparing the tracking algorithm in this paper with the JPDA algorithm in simulation experiment,the former can adapt to the measured data of the extended target of the though-wall radar.Moreover,the tracking algorithm does not need Kalman filter,which ensures the real-time performance of the algorithm.By processing the measured data of the though-wall radar,the results show that the method can identify the target,improve the accuracy of trajectory,provide the position information of the target in real time and gets robustness.