基于神经网络的椭圆扩展目标形态估计
Shape estimation for elliptic extended target based on neural network
陈训成 1戚国庆 1亓俊杰 2李银伢 1盛安冬1
作者信息
- 1. 南京理工大学 自动化学院,南京 210094
- 2. 上海精密计量测试研究所,上海 201109
- 折叠
摘要
针对复杂环境下由稀疏量测引起的椭圆扩展目标形态估计精度低的问题,提出了一种基于神经网络的形态估计方法.利用神经网络对目标量测进行处理,估计出椭圆扩展目标的轴长,然后结合卡尔曼滤波算法实现目标的跟踪.仿真实验结果表明,通过与基于随机矩阵、乘性误差以及卷积神经网络等模型的算法相比,所提算法的跟踪性能有显著改进.
Abstract
Aiming at the problem of low accuracy of elliptic extended target shape estimation caused by sparse measurement in complex environment,a neural network based shape estimation method is proposed.The neural network is used to process the target measurement,and the axis length of the elliptic extended target is estimated.Then the target tracking is realized by combining Kalman filter algorithm.Simulation results show that the tracking performance of the proposed algorithm is significantly improved compared with the existing algorithms based on random matrix,multiplicative error and convolutional neural network.
关键词
扩展目标跟踪/神经网络/卡尔曼滤波/形态估计Key words
extended target tracking/neural network/Kalman filter/shape estimation引用本文复制引用
基金项目
国家自然科学基金(62171223)
国家自然科学基金(61871221)
出版年
2024
NETL
NSTL
万方数据