Study On Detection Methods Of Sea-Surface Weak Targets

Maritime radar has satisfactory detection performance for large and medium-sized seasurface targets with strong echo energy,such as warships and steamships,but its detection ability for weak targets is still poor.Recently,with the rapid development of miniaturization technology and stealth technology,various types of new small targets emerge in an endless stream.These small targets have different characteristics and are merged in complicated sea clutter,showing various degrees of low observability,which causes challenges for maritime radar.Generally,the radar cross section of weak targets is small,which causes target returns to be often submerged in sea clutter.Moreover,target Doppler energy is easy to fall into the main clutter area of sea clutter,which deteriorates the detection performance.In addition,a large number of sea spikes with similar characteristics to small targets will appear in the high-resolution sea clutter.These sea spikes last for several seconds and are easy to be misjudged as targets in short-term observation,increasing false alarm rates.Due to the above factors,the design of weak target detectors in strong sea clutter has become a difficult problem for scholars at home and abroad.This thesis studies the two-stage detection algorithm for sea-surface weak targets in fast scanning mode and the multi-feature detection algorithm for sea-surface weak targets in staring mode.The research results can be applied to coastal surveillance radars,anti-invasion radars for ship,airborne early-warning radars and other radars to improve their detection capabilities.The main contents of this thesis can be summarized as follows:1.In the sea-surveillance scenes,high-resolution sea clutter will exhibit non-Gaussian statistical distribution,temporal non-stationarity,and spatial non-uniformity.These factors will affect the design and parameter selection of sea-surface target detectors,and then affect detection performance.In this thesis,four common clutter models are used to fit the measured sea clutter,and the model parameters are estimated by moment estimation or maximum likelihood estimation.The fitting results are evaluated using the K-S test to determine the best fitting model.In addition,the temporal and spatial correlations of sea clutter are analyzed,and the zero-memory nonlinear transformation method is used to generate correlated K-distributed sequences to simulate sea clutter for subsequent simulation experiments.2.Long-term energy accumulation cannot be achieved in radar fast scanning mode due to the limited number of accumulated pulses in a single scanning frame.On the other hand,the sea spikes are likely to cause false alarms in the high-resolution detection scene.To improve the detection performance,this thesis proposes a framework with two-stage accumulation and dual detection threshold to detect sea-surface weak targets.The proposed detector includes two stages: intra-frame detection and inter-frame detection.In the intra-frame detection stage,the optimal coherent detector in K-distributed sea clutter is used to accumulate the energy of radar returns in a single scanning frame,and a low intra-frame decision threshold is set to ensure that weak target trajectories are not filtered as much as possible.However,it also generates a quantity of sea clutter false alarms,which need to be filtered in the subsequent inter-frame processing.In the inter-frame detection stage,the original dynamic programming-based track-before-detect algorithm is improved.The crossframe accumulation strategy is introduced,which solves the problem of track loss due to the interruption of energy accumulation between frames caused by missed detection in the intraframe detection stage,and improves the detection probability of sea-surface weak targets.In addition,the Doppler guidance strategy is introduced,which uses the Doppler information obtained by intra-frame detection to cut off some unreasonable energy accumulation paths between frames,which effectively reduces the number of false tracks and reduces false alarm rates.Finally,the proposed method is evaluated through simulation experiments and measured data experiments,and it is confirmed that the proposed method can effectively and robustly detect weak moving targets in high-resolution sea clutter.3.Aiming at the problem that it is difficult to control the false alarm rate and make full use of the advantages of multi-features in the feature-based detection method for weak targets,this thesis proposes a multi-feature detection method based on support vector data description to detect sea-surface weak targets.Due to complex characteristics of sea clutter and target returns and the complex interaction between them,single-feature detectors usually have a low detection probability and are not robust enough.In order to improve the separability of sea clutter and target returns in the feature space,this thesis extracts different features of sea clutter and target returns from multiple dimensions such as time domain,frequency domain,fractal dimension,time-frequency domain,etc.The support vector data description algorithm is used to train the hypersphere in the feature space,and the test statistic of the detector is set as the distance from the sample to the center of the hypersphere.The decision threshold can be determined according to the distances from the training samples of the training set to the center of the hypersphere and the preset false alarm rate.The proposed method does not limit the feature dimensions’ number,so it can make full use of difference information of sea clutter and target returns in multi-dimensional feature space and use it more flexibly.And its false alarm rate is easy to control,which meets the actual use requirements.Finally,the proposed method is evaluated through multiple groups of measured data tests,and compared with the existing feature detection methods to verify its performance superiority in detecting floating small targets in sea clutter.