基于多物理场的舰船目标识别方法

摘要

针对现有舰船目标识别方法多是基于舰船单一物理场信号,识别性能较差,已不能适应现代战场环境的复杂性、多变性的问题,提出了基于多物理场的舰船识别方法.该方法通过小波分解提取舰船多物理场信号频段能量特征,并分别构建了基于特征层融合和决策层融合的特征融合模型,使不同物理场特征之间得以相互校准、补充和验证确认,实现单一物理场特征无法实现的识别功能.仿真验证结果表明,所提出的方法实现了92.65%的舰船目标识别率,与基于单一物理场的舰船识别率相比,提高了4.12%.%Existing ship identification methods are usually based on single ship physical field,which will limit the model performance and cannot adapt with the complex and changeable modern warfare.To handle this issue, we proposed a new method for ship recognition based on the combination of multi-physical fields.The wavelet decomposition algorithm is used to extract the features of each field.We construct feature-level feature fusion model and decision-level feature fusion model,which can verify and correct the features extracted from multi-physical fields.Experimental parts demonstrate that the proposed method achieved a best recognition accuracy of 92.65%,can got a improvement of 4.12% compared to methods based on single ship physical field.