Precise And Intelligent Detection Of Sea-surface Weak Targets Under The Complex Sea Conditions

With the increasing population and rapid development of technology,the activity of marine exploitation is becoming more and more intensive.Sea-surface weak target detection technology is not only an important part of ocean cognition and ocean utilization,but also the foundation of the strategy of “Building An Ocean Power”.However,the accurate seasurface weak target detection is facing several challenges under the complex sea conditions.In order to solve these challenges,this paper analyzes the characteristics of the sea clutter and the design of the detector,including the sea clutter modeling,the design of the constant false alarm rate(CFAR)detector and the intelligent detector.The contents of this thesis are listed as follows.1.Accurate cognition of sea clutter under the complex sea condition is studied.Specifically,we develop a kernel density estimation(KDE)based framework to model the sea clutter distributions without requiring any prior knowledge.In this framework,we jointly consider two embedded fundamental problems,the selection of a proper kernel density function and the determination of its corresponding optimal bandwidth.Regarding these two problems,we adopt the Gaussian,Gamma,and Weibull distributions as the kernel functions,and derive the closed-form optimal bandwidth equations for them.To deal with the highly complicated equations for the three kernels,we further design a fast iterative bandwidth selection algorithm to solve them.Our proposed KDE approach can not only reduce the modeling error but also improve the detection probability.2.A new-type sea-conditions-adaptive CFAR detector is studied for detection of weak targets within sea clutter.Based on the proposed sea clutter models,we adopt the temporal correlation(TC)as the metric to put forward a new-type background level estimator,referred to as the cell-averaging temporal-correlation CFAR(CATC-CFAR)detector.In the CATC-CFAR,the correlation length is a key parameter,which is difficult to determine as it intimately related to detection environments,such as wind speed and significant wave height.To this end,we elaborately select the de-correlation time as the correlation length and model it as a linear-regression problem with detection environments taking into account,and utilize the gradient descent algorithm to derive the optimal coefficients.Our proposed detector can significantly improve the detection probability in both low signal to clutter ratio(SCR)and low false alarm rate(FAR)cases.3.An intelligent detector based on the 3-D feature space is studied.We devote to establishing a discriminative multidimensional feature space and taking it as the input of the machine learning based classifiers.We first exploit the concept of the fractal theory to extract three representative features in the time and frequency domains and construct a3-D feature space.We then combine the constructed feature space with the decision tree approach to design an environment-adaptive detector.Most importantly,we modify the decision tree based detector to a FAR-controllable detector to meet the requirements of different detection applications.Our proposed detector can significantly improve the detection probability under different sea surface environment.In conclusion,this thesis devotes to first realizing accurate cognition of sea clutter under the complex sea conditions,and then designing the sea-conditions-adaptive CFAR detector and the intelligent detector that is based on the 3-D feature space,so as to achieve precise and intelligent detection of sea-surface weak targets in complex detection environments.