Sea-surface Floating Small Target Detection Method Using Polarization Features

The detection of sea-surface floating small target in sea clutter is a hotspot and difficult problem in radar community.The echo of radar emitting electromagnetic wave to the seasurface contains sea clutter.When detecting sea-surface targets,many radars often work on high resolution mode in order to reduce the power level of the sea clutter.Because of spacetime nonstationarity of high resolution sea clutter and the complex influence of wave to the motion characteristics of sea-surface small targets,the echo of sea-surface small target is difficult to be coherent integrated effectively,which degrades sea clutter suppression and the performance of target echo coherent integration obviously.Feature detection technology based on multiple features is considered to be one of the effective methods to detect seasurface floating small targets.Then extracting salient features that can distinguish sea clutter and target is the key step of feature detection technology.The rapid development of fully polarized radar and the continuous improvement of hardware processor performance make joint detection using multi-polarization channel radar data possible.There are obvious differences between sea clutter and floating target echo on polarization.Thus,the joint detection of sea-surface small floating targets using high resolution radar data is a potential research direction.This thesis researches feature extraction and the utilization of polarization information of feature based sea-surface floating small target detection using four polarization channel sea clutter data of IPIX radar database.Thesis is divided into the following three parts.In the first part,the polarization theory of electromagnetic wave is introduced,including polarization representation and polarization description of scatterers.The definition and solution of polarization scattering matrix,polarization coherence matrix and polarization covariance matrix are introduced in detail.Then four classical polarization target decomposition theories are described.Three classical decomposition algorithms: Cloude decomposition,Pauli decomposition and model decomposition based on unitary transformation are reviewed.In the second part,the theoretical analysis shows that the Freeman-Durden three-component decomposition method can describe the target polarization scattering mechanism and polarization characteristics well and is more suitable for floating target detection in sea clutter.Use the decomposition method above to extract three polarization features: the energy corresponding to the relative volume scattering mechanism,the energy corresponding to the relative dihedral scattering mechanism and the energy corresponding to the relative surface scattering mechanism.According to the separation of clutter with target feature vectors and clutter-only data feature vectors in 3D feature space,the fast convex hull learning algorithm is used to obtain the decision convex region.Finally the detection is realized.This method has better detection performance than traditional detection method based on feature because of the selection of appropriate polarization features.In the third part,two traditional sea-surface target detection methods based on feature are reviewed: the tri-feature-based target detection method and the detection method based on normalized Doppler spectrum.In the two detection methods discussed above use little polarization information.As for the problem,the new polarization features are obtained by fusing amplitude spectrum and Doppler power spectrum respectively.The detector based on four polarization channel fusion features and double-feature detector based on normalized Doppler spectrum are proposed.The experimental results of real sea clutter data show that the proposed improved methods based on four polarization channel fusion have better detection performance and robustness.