Research On Detection Technology Of Small Target Using Doppler Information

With the development and wide utilization of stealthy technology, the radar cross section(RCS) of typical military targets, such as aircrafts, missels, tanks and combat ships, has been dropped markedly. As a result, the echo signal from these targets is very weak and usually hidden in strong background clutter and electronic interference, which causes severe challenges to radar detection and tracking. Consequently, small target detection is a crucial problem that is starving for being resolved in the field of radar signal processing. However, the traditional target detection and tracking approaches cannot guarantee reliable detection and tracking performance when encounted with these small targets. Therefore, it is necessary to investigate new theories and methodologies which can improve detection and tracking performances.In this thesis, existing detection and tracking algorithms for small targets are studied, wherein great efforts are devoted to several key issues, such as “track initiation” and “data association”.What is more, a number of improvement measures are put forward in this thesis aimming at the limitations of the algorithms above. The main work and conclusions to this thesis are summarized as following:First of all, the background and the significance of this thesis, latest progress and research status in home and abroad, and classic target detection techniques are intoduced briefly in this dissertation. Meanwhile this thesis analysis the main difficulties in detecting small targets, and introduce the conception and implement approaches of distance condensation.Secondly, several frequently- used track initiation algorithms,such as rule-based method, logical method, modified logical method and method based on Hough transform are introduced in this thesis, and simulation experiments are performed to analyse and compare their performance detailedly. Result of the experiments shows that modified logical method has the best performance in initiating trcak among all of the track initiation algorithms mentioned above, while the method based on Hough transform has the wor st performance. Aimming at the drawbacks of the track initiation algorithms above, many practical improvement measures are put forward in this dissertation, including: take advantage of the dopper information from target to improve traditional correlative wave gate by setting up two-dimensional correlative wave gate base on distance and doppler information; propse a new track initation method based on modified logical method and Hough transform. In order to confirm the correctness and practicality of the improvement measures proposed above, a number of matlab-based simulation experiments are carried out to analyse and demonstrate their performance. Result of the experiments shows that the improved trcak initiation algorithm,which can reduce effectively probability of initiating false tracks and raise probability of detecting real targets, has a better performance than traditional track initiation algorithms, especially in dense cluttered environment.Finally, rationale of nearest neighbor(NN) algorithm and probabilistic data association(PDA) algorithm is introduced in this thesis, and simulation experiments are carried out to analyse and compare their performance. Result of the experiments shows that probabilistic data association(PDA) algorithm has better data association performance than nearest neighbor(NN) algorithm, while its amount of calculation is larger than nearest neighbor(NN) algorithm. Amming at the drawback of probabilistic data association(PDA) algorithm, such as error tracking and miss tracking, in dense cluttered environment, some improvement measures are proposed in this thesis. The main idea of improved probabilistic dada association(PDA) algorithm lies on using doppler information to filter out some false alarm point before data associatio n, thus improving the performance of the improved algorithm. Then simulation experiments are carried out to analyse performance of the improved algorithm above, and result of experiments shows that improved algorithm proposed in this dissertation has bette r target tracking performance than traditional probabilistic data association(PDA) algorithm.