Research On Sea-surface Weak Target Detection Using Spatial Correlation Of Radar Echoes

Sea-surface weak target detection is one of the key technologies for marine exploration and development,and it is also an important foundation for building a marine monitoring network and constructing a strong marine nation.High-performance sea-surface weak target detection requires comprehensive analysis and utilization of radar echoes.However,most existing methods only focus on the time-frequency characteristics of radar echoes while ignores their spatial characteristics,which severely limits the detection performance.Therefore,to further improve the detection performance for sea-surface weak targets under various conditions,this paper fully exploits the information contained in the physical space dimension of radar echoes and further applies them into detection process.The main work of this paper is summarized as follows.First,by using the spatial correlation of sea clutter,this paper proposes an improved adaptive normalized matched filter(ANMF)detection scheme.In detail,this paper modifies the original binary detection problem according to the spatial correlation of sea clutter texture.Then,under this modified detection problem,a new ANMF decision formula is strictly derived.Based on this decision formula,a texture spatial correlation analysis step is added to original ANMF detection process to filter out reference cells that have the same texture as the cell under test to provide additional prior information for target detection.Simulation based on realistic radar datasets demonstrates the proposed detector outperforms conventional ANMF detector in different signal-to-clutter ratio(SCR)and Doppler frequency cases.In particular,the proposed detector can achieve a performance improvement of up to 10%in low SCR cases(less than-5 d B).Second,by using the spatial correlation of the target echoes,this paper proposes a feature detection method based on spatial correlation.The echoes of low-speed floating targets usually fall into the wider sea clutter bandwidth in the Doppler domain,which will cause the performance degradation of the ANMF detector.To solve this problem,this paper first constructs a new feature of average spatial correlation coefficient(ASCC)to describe the strong spatial correlation between the echoes of the target cell and the echoes of the neighboring cells.Subsequently,a feature-based detector is designed based on this feature and tested on the measured dataset.Results show that,compared with commonly used time-frequency features,using ASCC feature can significantly improve the detection performance at low SCR(less than-2 d B)and low false alarm rate(less than 10^-3)cases.The performance improvement can reach up to 34%and 60%,respectively.