Improved Data Association Algorithm for Airborne Radar Multi-Target Tracking Via Deep Learning Network
Wenna Li, Ailing Yang, Lianzhong Zhang · IGARSS 2022 - 2022 IEEE International Geoscience and Remote Sensing Symposium · 2022
Traditional data association (DA) algorithms for multi-target tracking need prior information such as target motion model and clutter density. However, this prior information cannot be timely and precisely obtained before tracking. To solve this issue, this paper proposes an improved data association algorithm for multi-target tracking via a deep learning network. First, a dataset is constructed to provide rich offline multi-target data association for network training. Then, the LSTM-DA algorithm is developed to solve the multi-target and multi-measurement matching problem based on the long short-term memory (LSTM) network. The network is composed of two LSTM layers, a masking layer, and three fully connected layers. The experimental results demonstrate that our proposed algorithm outperforms classical data association algorithms and the bi-directional LSTM network.