| Maritime surveillance attracts more and more attentions around the world with the rapid development of human activities.High frequency surface wave radar(HFSWR),which is a useful tool for monitoring the sea,attracts widely attention by scientists and engineers.However,there are some shortcomings when using HFSWRs to detect ships and classify multiple targets for the exsitence of significant sea clutter and the low resolution.To improve the performance of target detection in sea clutter as well as multi-target detection,advanced signal processing techniques in time-frequence domain and spatial-temporal domain are studied by this dissertation to take advantages of the diversities of target signal and sea clutter in varied domains.Multi-frequency detection and data fusion techniques are denoted to elevate the detection capability as well.Based on previous researches of target detection by HFSWRs,further researches are conducted on radio scatter theories,radar measurement principles,statistical models of target and sea clutter,nonlinear tracking filters and other related techniques.Various frameworks of target detection and tracking are compared with each other as well.The detailed contents are as follows:First,classical target detection and tracking methods are stuied and introduced,and several methods to improve the range and Doppler frequency measuring are compared.Research foundations,including the scatter theories of target in high-frequency band,range and bearing measurements,the fluctuant models of target,the sea clutter models and non-linear tracking filters,are introduced.The shortcomings of conventional methods for target detection and tracking are demonstrated by the analysis of radio scatter priciples in HF band,the equation of radar detection as well as the analysis of target signal and sea clutter.Furthemore,we try to improve the range and Doppler resolution by multi-frame accumulation and super-resolution algorithms.The tracking performance of entended Kalman filter and unscented Kalman filter are evaluated by simulations.All these researches lay a solid foundation for target detection and sea clutter suppression.Then,the study of sea clutter suppression when target signal and sea clutter are mixed in the Doppler domain is conducted.The narrow-band sea clutter model is introduced and implemented first,then the SVD based sea clutter rejection,the loop iteration clutter suppression method and the orthogonal projection method are introduced and analyzed.To overcome the shortcomings of classical methods,independent component analysis is proposed and varified by simulations.Furthermore,taking avantages of the linear feature of fractional Fourier transform(FRFT)and classification capability of singular value decomposition,sea clutter suppression by SVD-FRFT filtering is introduced and varified.The problem,sea clutter rejection for mixed target signal and sea clutter,is solved well both in simulation and experiments,showing the effectiveness of SVD-FRFT filtering.Next,target detection in sea clutter is studied.Target detection by time frequency analysis is introduced first,then signal features of target and the first-order sea clutter are studied,and modeling of radar echo is implemented.A novel target detection method based on FRFT to improve target detection is proposed.The gain of target signal and sea clutter by FRFT is analyzed,while the operation conditons are studied.With the analysis of estimation of FRFT orders,we also propose using bandwidth-kurtosis method to solve the most difficult problem when using FRFT to improve target detection by HFSWRs.Then the sparse-recovery-based mutiple signal classification(SR-MUSIC)method is proposed to improve target detection by HFSWRs.The shortcomings of the classical cascaded temporal-spatial estimation are shown by the studing of maneuvering characters of target and distributions of sea clutter in spatial-temporal domain.The deficiency of traditional spatial-temporal estimation is also considered,and the multiple-measurement-vector based sparse recovery theory is introduced.Qualitative analysis of spatial-temporal estimation is given.The SR-MUSIC approach could effectively detect a target masked by sea clutter as well as multiple adjacent targets and distinguish them from each other.Results obtained using real data with opportunistic targets validate our approach.Therefore,the proposed 2-D SR-MUSIC approach improves target detection and outperforms conventional cascaded methods.Then,enhanced target detection by multi-source data fusion and time-varied tracking filtering based on multi-frequency high-frequency surface wave radar is proposed.Enhanced target detection by multi-site data fusion is studied first,while data features of targets are analyzed.A novel data fusion method by multiple-dimension linear weighting is proposed.The system performance is defined in terms of time on target,track fragmentation,tracking area and accuracy.The nonlinear filter,unscented kalman filtering,is denoted here to decrease the variance of measurements.The effectiveness of the data fusion and tracking procedures is shown by experiments.Furthermore,the classical framework of target detection and tracking by HFSWRs are shown.The whole processing chain,including the pulse compression,matched filtering,interference suppression,Doppler processing,constant false alarm rate detection,spatial spectrum estimation,data fusion and tracking filtering,are discussed.The shortcomings are shown by real data analysis.To improve the target detection ability,the framework based on spatial-temporal estimation is proposed and varified by comparsion between real measurements and data of automatic identification system.Detection performance based on consistency,data acquisition rate and so on are shown and discussed.Finally,conclusions are achieved.The necessarities of multi-frequency measuring and data fusion,time-frequency analysis,spatial-temporal estimation for improving target detection are shown clearly from the study on both radar system design and multi-dimension singal processing(range-doppler domain,time-frequency domain,spatial-temporal domain).Further researches on efficient sparse recovery algorithms,FRFT order estimation and higher order frequency modulated features of target are valuable. |
PDF Full Text Request