Sea Surface Target Detection Technology Based On Feature Transformation Driven Deep Network

Radar target detection under the background of complex sea clutter is one of the important research directions in the field of radar technology.Sea clutter usually presents non-uniform,nonlinear and non-Gaussian characteristics,and the radar echo signal-toclutter ratio is too low,and the target is often submerged and difficult to detect effectively.Therefore,how to effectively detect weak and small targets from strong sea clutter is one of the key issues that needs to be solved urgently in the field of sea surface detection and surveillance,which has important value and significance for strengthening coastal defense construction.1.The method of sea clutter and target feature extraction based on multi-transform domain processing is studied.Starting from the information entropy in the fractional Fourier transform domain,the largest singular value in the singular value decomposition domain,and the low frequency eigenmode function energy ratio in the complex value empirical mode decomposition domain,the sea clutter is analyzed.The different characteristics of the signal and the target signal in different transform domains.Based on this,it provides a basis for distinguishing sea clutter signal and target signal to construct feature space.Finally,the joint distribution of the three characteristics is analyzed,and the different characteristics of the sea clutter signal and the target signal are explored in the high-dimensional space.2.An intelligent detection algorithm of sea surface target based on support vector machine is studied.Using the idea of classification,the sea clutter samples and target samples are classified based on the proposed features to achieve the purpose of target detection,and the false alarm probability index is introduced into the intelligent detection model,so as to solve detecting performance degradation issues because of the model mismatch.3.An intelligent detection algorithm of sea surface targets based on Res Net-50 network is studied.Using the feature expression ability of neural network,combined with the proposed features,the classification of sea clutter samples and target samples is realized,and the false alarm probability index is introduced into the intelligent detection model to realize the target detection method under the condition of constant false alarm rate.Compared with Compared with the traditional CFAR detector,the detection probability of the target is improved.The above algorithms have been verified and compared by IPIX radar measured data.The experimental results prove that the algorithm can effectively improve the detection performance of sea surface targets under the background of sea clutter.