Research On Air Target Structure Identification And Feature Extraction Method Based On ISAR Image

The development of Inverse Synthetic Aperture Radar(ISAR)technology has transformed radar technology into visualization,which has effectively improved the ability of radar to acquire information on targets and scenes.As a two-dimensional high-resolution rangeazimuth image of a target,ISAR image contains the refined structural information of the target,and is an important means to achieve target feature extraction and classification and recognition.Most of the traditional ISAR image interpretation methods rely on manual interpretation and have low efficiency.For some complex ISAR images,it is limited by factors such as personnel experience,which makes it difficult to identify targets.Therefore,how to quickly and accurately realize the automatic target interpretation of ISAR images is an urgent problem that needs to be solved in the current radar target recognition field.Aiming at the above problems,starting from the target ISAR signal model,this paper studies the method of target identification and physical feature extraction.The main contents are summarized as follows:(1)A method for identification and feature extraction of typical parts of aerial targets that combines image processing and signal processing is proposed.Firstly,the ISAR image quality is improved by image processing,and the target support area is extracted.Then,aiming at the problem that it is difficult to effectively identify the typical parts of the target by methods such as principal component analysis(PCA),it is proposed to use the tracking information to estimate the projection direction of the air target body in the ISAR image.And using the orientation as prior information,the projection orientation of the wing in the ISAR image is located.The method accurately identifies the target fuselage and wing components in the support area according to the projection pointing,and extracts the relevant geometric dimension features of the target.The experimental results based on the simulation and measured data show that the proposed method can automatically identify the fuselage and wing components of the air target,and at the same time realize the accurate extraction of the two-dimensional size features of the target.(2)The three-dimensional attitude information of the aircraft target is of great value for tasks such as flight state monitoring and action intention analysis.To this end,this paper proposes a joint estimation method of 3D attitude and size of target and payload based on ISAR images.The method first extracts the projected physical features of the aerial target through the convex hull detection method,establishing an optimization model for the difference between the theoretical value of the projected feature corresponding to the 3D model of the target and the observed value of the extracted projected feature.And by using the gradient method to solve the optimization model,the joint estimation of the three-dimensional pose and size of the target is finally achieved.In addition,in order to mine more refined structure information of the target from the ISAR image,a 3D attitude and size estimation method for the payload carried by the target is also proposed.The experimental results based on simulation and measured data show that the proposed method can automatically extract the threedimensional attitude and physical size features of the target and the load,and achieve a high estimation accuracy.