Research On Ship Target Recognition Based On Structural Features Of High Resolution Range Profile

Radar automatic target recognition is an important research content at home and abroad and becomes one of the most important technologies in the development of modern radar technology.It is widely used in civil and military fields.With the development of electronic and information technology,the requirement of radar detection has been improved from detecting the target to detecting the type of target.Among them,the target recognition based on high resolution range profiles is a very important aspect of radar automatic target recognition,and has made great progress in the identification of ballistic missiles,aircraft and ships.High Resolution Range Profiles reflects the distribution structure of ship targets along the radar line of sight,so it contains a lot of information about target scattering information,structural characteristics and other advantages.And it has become a hot topic in the field of automatic radar identification due to its easy acquisition and processing.In this paper,the structural information of high resolution range profiles is mined to a certain extent,and the extracted structural features are used for identification research.The research mainly focuses on structural feature extraction,feature selection,multi-feature fusion recognition,classifier selection and decision level fusion,and is verified by the measured data of ship target.Firstly,the model of scattering point of ship target and the preprocessing of high resolution range profiles are introduced,and the characteristics of high resolution range profiles,including orientation sensitivity,translation sensitivity and amplitude sensitivity,are analyzed and studied,and the corresponding solutions are given to eliminate the influence on subsequent recognition.After data preprocessing,nine different structural features are proposed for the structural information of the reaction of high resolution range profiles,the actual structural information reflected by each structural feature is analyzed,and the extraction method of structural features are given.Only one structural feature couldn’t distinguish all ship targets,so multi-feature fusion recognition make up the defects of single feature.Nine kinds of structural features are evaluated by the distance of the inner and interclass and mutual information values,and the better combination features,which greatly improves the recognition effect,are obtained.The effectiveness of the algorithm is verified by the measured data.Finally,because of the difference in the classification ability of the classifiers and the influence on the recognition,the method of decision level fusion based on multiple classifiers and multiple features is adopted to improve the recognition effect,and the two methods of Dempster-Shafer evidence theory and the evidence union weighted fusion strategy based on information entropy are used for decision fusion.In view of the difference of classifiers,a multi-classifier selection algorithm based on Dempster-Shafer evidence theory is proposed,and the effectiveness of the algorithm is demonstrated by comparing the combinations of different classifiers and using the measured data.