Sea Surface Target Detection Based On The Combination Of Multi-polarization Features

In recent years,the development of radar target detection technology is booming,and it has been widely used,and it is of great significance in both military and civilian.Since the marine environment is affected by many factors,target detection technology has become a hot topic in this field.Traditional sea surface target detection is based on statistical theory for modeling and analysis,which has certain limitations.Therefore,this thesis considers from the perspective of the target polarization direction,introduces the incoherent polarization target decomposition,and completes the target detection based on the detection statistics obtained by the decomposition.The research content of this thesis is as follows:We firstly describe the basic theory of polarization target decomposition,analyze the polarization of radar electromagnetic waves secondly,introduce the characteristics and forms of several polarization matrices thirdly,recommend the representative expressions of coherent decomposition and incoherent decomposition in polarized target decomposition and analyze the characteristics and applications of each decomposition lastly.Then we use the polarized and the non-polarized features of target to achieve target detection,extract two polarized features,that is the relative volume scattering power and the relative average amplitude.We firstly compare the one-dimensional space distribution of the proposed two features,and then propose a two-feature-fusion detector based on the two-dimensional convex hull learning algorithm.Compared with the four single feature detectors,that is the relative volume,dihedral angle,surface scattering power and relative average amplitude,our proposed detector can obtain better detection performance.Finally,two target decomposition methods are used to extract different polarized features,such as,the average scattering energy in the Cloude decomposition and the two relative scattering powers in the Freeman decomposition.Compared the three-dimensional space distribution of the three extracted features,it can be seen that the distinction between the target and the clutter in the three-dimensional space is obvious.Therefore,a three-feature-fusion detector is proposed by using three-dimensional convex hull training algorithm.Compared with the classic CFAR algorithm,the three-feature-fusion detector can obtain good detection performance.