基于静态模型的多视角SAR图像目标识别方法
A Method of Multi-look SAR Image Target Recognition Based on Static Model
-
摘要: 传统的多视角SAR图像集合对于目标姿态角具有高度敏感性,因此在用于目标识别时存在一些不足之处。针对该问题提出一种多视角SAR图像的静态建模方法,它将来自一个目标多个视角下的图像信息集成到一个综合的数据结构中,并且该数据结构与目标散射中心有关而与角度无关。然后利用静态模型对不完全姿态角的目标数据进行静态建模,利用模板匹配法对输入多视角图像进行目标识别。理论分析和仿真结果表明,本方法在每个目标只有少量姿态角模板数据可用的情形下比传统模型具有优势。Abstract: Traditional multi-look SAR images are highly sensitive to target aspect angle,which is unfavorable to target recognition.To solve the problem,a static modeling method for multi-look SAR images is proposed.The method integrates images from multiple aspects into a composite data structure.The data structure is reorganized so as to be function of target scatter centers,not of aspects.Then the static model is used to modeling the target data of incomplete aspects,and form the template of different targets.The input multi-look SAR images are classified using template matching algorithm.The theory analysis and simulation result show that the method excels traditional model when a small quantity of SAR images of different aspects are feasible.