Research On Target Tracking Before Multi-radar Plot Fusion

The multi-radar system is developed rapidly for the complex combat environment which contains the low detection rate and high maneuverability targets.The multi-radar system is superior to the single one in detection range,system reliability,robustness,locating accuracy and the ability of target recognition.However,jointly detection and tracking multiple targets using multi-radar system is a tough mission.This work focuses on the algorithm of jointly detection and tracking multiple targets using multi-radar system.The state of art mainly concludes plot fusion and trajectory fusion.However,trajectory fusion barely taking full use of target characters obtained by detection information.Imperfect target detection because of time aligning method,wrong measurement association which comes from dense clutter lead to lots of false trajectories.A Hough transform and MHT based track before detect algorithm is proposed here to solve the multi-radar data processing problem.Plenty of information originated from detection data before the plot fusion has been mostly exploited.Furthermore,the computation is sharply decreased.The algorithm achieves accurate detection and tracking targets under various severe environment.The specific content is as follows:Firstly,the important theories of target tracking before multi-radar point fusion based on Hough transform are reviewed,including the basic theory of track-before-detect(TBD),the basic theory of data association methods,the design framework used for multi-radar data fusion,asynchronous multi-sensor time aligning and space aligning.Secondly,the traditional Hough algorithm is not suitable for multi-radar model of asynchronous sampling.A 3-dimensional Hough transform(3DHT)based TBD algorithm is proposed,which fully utilizes the time information of measurements to solve the multi-radar model,and reduces the computational complexity of Hough transform by Robert's optimal line representation and optimized parameter discretization way,through the "voting" mechanism and the "non-maximum suppression" idea to improve the detection model,while increasing the detection probability and reducing the false alarm rate.Finally,Because the Hough transform based track-before-detect method are designed to detect the targets in a straight-line constant-velocity model,this paper presents a novel method for detecting and tracking multiple maneuvering targets using Hough transform and multiple hypothesis tracking(MHT).This method consists of two stages.In stage one,the measurements in multiple scans are partitioned into overlapped time windows.Tracklets in each time window can be detected by the 3DHT.In stage two,the tracklets are associated to get the entire trajectories with the MHT.Output the target track.This paper simulates the target linear motion model and maneuvering target movement in a multi-radar environment.The simulation results fully verify the effectiveness and superiority of the proposed method.