Research On Dynamic Programming Track Before Detect Algorithm

With the rapid development of stealth and low-altitude penetration techniques,the surveillance environment becomes more and more complex.The radar system has to face the challenge of detecting and tracking weak target signal in heavy clutter.Compared with the conventional detection methods,the track-before-detect(TBD)technique can achieve better detection performance by accumulating multi-frame raw or low-thresholded data,at the cost of consuming time.Consequently,The TBD method has important theoretical significance and military value.The dynamic programming based TBD(DP-TBD)method has attracted more and more attentions in the radar detection field,due to its simplicity of implement and easiness of extension.In this thesis,supported by the twelfth pre-research project,the DP-TBD method is analyzed and improved from the following aspects:Firstly,for the problem of dim target detection and tracking in heavy clutter,the dynamic and measurement model of dim target in radar system is given.Based on the theoretical basis of DP-TBD,two traditional cost functions are analyzed and compared in simulations.Secondly,in order to improve the detection and tracking performance of the very low observable target,an amplitude association DP-TBD(AADP-TBD)algorithm is proposed.It uses amplitude association information between two consecutive frames to build cost function,and invokes a variable threshold for state backtracking.Compared with traditional DP-TBD algorithms in simulations,the proposed algorithm can greatly improve the detection and tracking performance of the very low observable target.Finally,in order to detect and track the maneuvering dim target in radar system,a modified AADP-TBD method is proposed.It rebuilds the cost function with the feedback of tracking results.The proposed method constructs a feedback function using the Minkowski distance between the candidate state at current time and the estimate state at previous time,and then builds the cost function with the amplitude association information in two adjacent frames.The proposed method can improve the detection and tracking performance of maneuvering weak target effectively.The simulation results verify the effectiveness of the proposed method.