Spatial-temporally Non-coherent Detection In Sea Clutter Background

Target detection in sea clutter is always a main task for marine surveillance radar, which has received wide attention. The non-stationary and non-Gaussian characteristics of sea clutter highly affect the performance of target detection, which limits the target detection in sea clutter. Therefore, an analysis and research of properties of sea clutter is necessary. Recently, the Pareto-distributed clutter model has been increasingly used to analyze, model and simulate sea clutter. Moreover, moving target detection under the Pareto-distributed clutter has been an important subject in practical applications. This paper mainly deals with the parameter estimate of sea clutter under the model of Pareto distribution and the spatial-temporally non-coherent detection algorithm which matches the shape parameter.At first, the characteristics and statistical model of sea clutter are analyzed. Several common amplitude distribution models for sea clutter are reviewed and their application ranges as well as advantages and disadvantages are interpreted. Then, the spatial-temporal correlation of sea clutter is given and the compound-Gaussian model is introduced.Secondly, the generalized Pareto intensity model of sea clutter is presented and its parameters are estimated. Two parameters estimation methods, the moment estimation method and the bi-percentile method, are mainly introduced. Due to the abnormal value samples, a bi-percentile method is adopted and applied to the non-Gaussian estimate for airborne sea clutter in a large scene and the perceived effectiveness of non-Gaussian method is verified by simulated data. Simulated experiments show that bi-percentile method with the ability to resist abnormal samples performs better than the traditional moment method.Finally, the basic structure of traditional non-coherent CFAR detection is reviewed at first and the implied conditions and limitations to realize its CFAR properties are illustrated. Then, due to the limitations of the traditional non-coherent CFAR detectors, a spatial-temporally non-coherent detection method matching the shape parameter is proposed, that is, median ratio detection method in Pareto distribution. Moreover,a detailed introduction of proposed method is given and the setting and calculation of CFAR threshold are completed. It is proved that its threshold is only determined by the shape parameter of Pareto distribution, the number of reference cells, the number of integrated pulses and the false alarm rate while independent of scale parameter. Then, the problem of selection for spatial-temporal reference cells is researched and several different selection modes for spatial-temporal reference cells are given. Finally, the experimental results indicate that the median ratio detection method under the Pareto distribution has the constant false alarm rate(CFAR) property in the whole scene and enables to resist the abnormal interference. Moreover, it is proved that the proposed method of selection for spatial-temporal reference cells can improve the detection performance.