Research Of Ship Target Detection Algorithms For HFSWR Based On Time-frequency Analysis

The 21st century is the century of ocean, then how to protect and develop the ocean has become the focus of world development. High Frequency Surface Wave Radar (HFSWR) can detect ship targets and low-altitude aircrafts beyond the horizon by using diffraction characteristics of high frequency electromagnetic wave propagation along the sea surface, thus having the characteristics of far detection range, high detection accuracy, and less investment etc. It can realize full-time and real-time monitoring of maritime space, especially those with significant value for both military and civilian monitoring of ship targets in marine exclusive economic zone (EEZ). Therefore, how to improve the detection accuracy of moving objects has always been one of the biggest challenges in HFSWR maritime surveillance currently. Based on the continuity feature of moving targets, the dissertation investigates the effective ship target detection algorithms from time-frequency analysis technique. The following research work has been accomplished.(1) A time frequency analysis algorithm combined Hilbert Huang Transform with wavelet packet decomposition. In traditional HHT, the empirical mode decomposition (EMD) is first performed to obtain the signal intrinsic mode function (IMF) components, and then each IMF component is processed by using Hilbert spectrum analysis to obtain the time-frequency distribution diagram. In order to overcome the end effect and mode interference problem, the proposed algorithm extends the data sampling time to avoid error caused by end point effect, and employs the band-pass filter characteristics of wavelet packet to suppress the mode interference problem of EMD. Simulation results show that the improved algorithm has good time-frequency aggregation and adaptability, and can get satisfied time-frequency distribution. In addition, the application of the algorithm to the radar signal analysis is investigated, and some existing problems are presented.(2) A target ship detection algorithm based on wavelet spectrum zoom and reassigned scalogram for HFSWR. The practical target radar data is usually centralized within ±0.3Hz frequency bands, especially in the Doppler shift range of sea clutter. Therefore, in order to improve the detection accuracy of moving targets for HFSWR, a target detection algorithm based on spectrum zoom and reassigned wavelet scalogram time-frequency analysis is presented in the dissertation. Firstly, the received signal of HFSWR is processed by the spectrum zoom technique to improve the frequency resolution for the subsequent time-frequency analysis. Then the Morlet wavelet based time-frequency analysis method is used to extract the distribution features of targets in the time-frequency plane. Moreover, the wavelet scalogram is reassigned to improve the time-frequency concentration and suppress cross term interference. Finally, the suspected target area is detected accurately from the obtained time-frequency distribution map. Then, the proposed algorithm is applied to the actual radar data, and the experimental results using AIS data verify the effectiveness of the algorithm. The proposed algorithm can not only effectively detect moving objects with very small differences among their Doppler frequencies, but also extract objects with Doppler frequencies near the sea clutter, thereby providing a fine and accurate detecting and analyzing algorithm to some suspect object areas which are not identified by the conventional target detection algorithms. Moreover, it can improve the accuracy of moving target detection under sea clutter background.