基于多模态数据融合学习网络的微弱目标群检测方法

Low Signature Target Group Detection Based on Multimodal Data Learning Fusion Network

微弱目标群检测是目标检测领域中的难点,随着传感技术发展使得基于多模态数据融合的目标检测方法成为可能。传统的方法常常关注于人工设计的多模态数据融合层面,一般采用信号级及决策级的融合,无法充分利用多模态内在融合特征。为此,为了充分利用深度卷积网络的内在多模态特征表达能力,提出一种端到端多模态数据学习融合网络,能够融合学习可见光、红外和多普勒脉冲雷达数据。在FLIR_ADAS数据集上验证算法识别微弱目标群的性能,实验结果表明本文算法能够显著提高微弱目标群检测性能。

Low signature target group detection is a difficult problem.With the development of sensor technology,the target detection method based on multimodal data fusion becomes possible.Traditional methods focus on the artificial design of multi-modal data fusion level,by using signal level and decision-making level fusion,and neglecting multi-modal inherent fusion features.In order to make full use of the inherent multimodal feature expression ability of deep convolution network,a point to point multimodal data learning fusion network is proposed,which can fuse data from visible,infrared and Doppler pulse radar data devices.The experimental results on FLIR_ADAS dataset show that the proposed algorithm can significantly improve the performance of low signature target group detection.