Research On Dim Target Detection And Tracking Algorithm In Complex Environment

In modern warfare, in order to achieve effective strike, remote targets should be detected and tracked to ensure enough time for command-control system. Because of multipath error, small target imaging area and so on, remote target signal power becomes weaker and more difficult to be detected and tracked. Meanwhile, with the developments of stealth technology, low-altitude penetration technology, anti-radiation technology and electromagnetic interference, the radar cross section becomes small and the target echo becomes weak. Furthermore, the target signal is also influenced by complex background, such as high false alarm and high clutter. Thus, detecting and tracking dim targets in complex environment becomes a technique challenge. Compared with classic detect before track (DBT) technique, track before detect (TBD) method is an effective solution. It makes soft decision on the measurement data. Unlike DBT, TBD detections are not declared at each scan. Instead, a number of scans of data are processed, and then the estimated target track is acquired when the detection is declared. In this thesis, research and application about detecting and tracking dim targets in complex environment are summarized in several key bullets below:1. For detecting and tracking a maneuvering dim target with fluctuating amplitude, a multiple model particle filter track before detect (MMPF-TBD) algorithm based on sliding window is proposed. The traditional MMPF-TBD needs more particles when the target maneuvers and fluctuates strongly. The new algorithm applies sliding window to determine whether the particles are affected by the estimation of the target. When the value exceeds threshold, new particles are added in accordance with the state estimation of the previous moments. With the new algorithm, the diversity and effectiveness of the particles are protected, and the detection probability and tracking accuracy are improved.2. For multiple radar target detection and tracking, a novel particle filter based track before detect (PF-TBD) algorithm is proposed to detect multiple targets with unknown maximum target number. At each scan, in order to reduce the influence of invalid models and states, the new algorithm adaptively updates the state set and event model set according to target estimation results. After filtering, sliding window method is applied to determine whether the targets appear or disappear, and the target number and state are estimated. Simulation results show that the new algorithm can effectively detect and track multiple targets in real time, and improve detection performance.3. For traditional TBD methods assume a single sensor system, a new particle based TBD algorithm is proposed for multiple asynchronous radar system. According to the difference of radar sampling rates and detection coverage, this algorithm designs a classification criterion to divide particle set into two parts. One part of the particles is used to estimate the target state, and the other part is used to preserve adequate particles in each radar detection coverage. Simulation results confirm the efficiency of the new algorithm.4. For traditional tracking techniques have advantages when estimating multiple maneuvering target states, an improved dynamic programming track-before-detect algorithm (DP-TBD) called penalty DP-TBD (PDP-TBD) is proposed. By PDP-TBD, the advantages of traditional tracking techniques are introduced to TBD frames. The performances of tracking techniques are used as a feedback to the detection part, and the feedback is constructed by a penalty term in the merit function. With this feedback, the algorithm combines traditional tracking techniques with DP-TBD and it can be applied to simultaneously detect and track maneuvering dim targets. Meanwhile, a reasonable constraint that a sensor measurement can originate from one target or clutter is proposed to minimize track separation.5. In real environment, many uncertain problems exist, for example, the uncertainty of target movement, the uncertainty of the origination of measurements, the uncertainty of the interval of two successive sampling scan, the uncertainty of target number and so on. As a solution, a multi-target tracking algorithm based on multi-model is proposed. The simulation and experiment confirm that the algorithm can effectively deal with100targets, and estimate theses targets track number and states.