Research On Small Ship Target Detection Method For Reconnaissance Radar

Facing the more and more complex marine environment,as well as the upgrading technology of weapons which are under the sea and made it as a cover has become a seriously threat to the coastal areas.At the same time,which means that the diversity of targets require the performance of modern radar detection getting higher.The technology of small target detection under the sea,as the hotspot and difficulty in radar technical attracts people’s attention consistently.In this paper,a new multi-parameters combined binary detection method is proposed for the small ship target.We use IPIX datasets and another X-band datasets to prove it working well.The mainly content of this paper can be divided into three parts:Firstly,sea clutter modeling and statistical characters analysis.IPIX datasets has been used to be modeled under different polarizations and sea state.In this paper,six models has been used for fitting analysis,including Rayleigh,Weibull,Log-normal,K distribution,Gamma,generalized Gamma distribution.Expect for K distribution,generalized Gamma distribution can degenerate into another four distributions under some special conditions.The generalized Gamma distribution as a tri-parameters model is flexible to describe the complex high-resolution radar sea clutter.We use two kinds of estimation method to estimate the parameters of the six models,such as high order moment estimation and maximum likelihood estimation.Chi-square and K-S test are used to evaluate the goodness-of-fit for the six models.Then,we choose the best one as the CFAR detector for the first-threshold of multi-parameters combined binary detection method.Experiment on IPIX datasets show that the model shape parameters extract from statistical characters analysis and the correlation time of clutter sequence can effectively distinguish the target cell from pure clutter cell.Secondly,multi-parameters combined binary detection.Because of the limitation of single parameter’s detection performance,we propose a multi-parameter combined binary detection algorithm which refers to double threshold detection in this paper.Cell average constant false alarm rate is used for the first threshold detection,according to the clutter modeling results select the appropriate detector.Operating a second threshold detection on the cells which are renumbered after passing the first threshold.The parameters used as the second threshold includes fractal property--Hurst exponent,recurrence quantification analysis(RQA)measures,statistical characters parameters.Setting another single threshold for each parameter,then make a statistical analysis for the detection result.According to the result make a final decision.IPIX datasets are used to verify it can work well.Limited by the number of data cell,when we use CFAR to pro-cess the datasets,it will cause a great CFAR loss,so we another X-band measured datasets to verify its effectiveness.Thirdly,track-before-detect based on dynamic programming.Using the correlate time of target is stronger than the noise background in the multi-frame.The energy of target is greater than the noise background after accumulating multi-frame.So we use this kind of feature to reach the goal for weak target detection.According to the optimal function determine the location of the target,while using the test results backtracking track.The validity of the method is verified by simulation and semi-measured data.