首页> 外文会议>International conference on communications, signal processing, and systems >A Deep Learning Method of Moving Target Classification in Clutter Background
【24h】

A Deep Learning Method of Moving Target Classification in Clutter Background

机译:杂波背景下运动目标分类的深度学习方法

获取原文

摘要

The Doppler spectrums of radar echoes of targets can reflect the change of the instantaneous velocity of targets. Therefore, it can be used for analyzing the motion state of the target and classifying them. Besides, deep learning is widely used in the classification of images. This paper proposes a deep learning based method of classifying targets in sea clutter. First, we introduce the motion model of targets and analyze their Doppler spectrum, based on which, we stimulate the time-frequency images of targets' radar echoes. Since clutters in echoes usually obey Weibull distribution, we add Weibull clutter (Mezache and Soltani) to a novel threshold optimization technique for far-away detection in Weibull clutter using fuzzy neural networks, 2007, [1]) to the echo signals. Then we classify targets with different networks using NVIDIA DIGITS, based on the images and analyze the results of classification.
机译:目标雷达回波的多普勒频谱可以反映目标瞬时速度的变化。因此,可以用于分析目标的运动状态并对其进行分类。此外,深度学习被广泛用于图像分类。本文提出了一种基于深度学习的海杂波目标分类方法。首先,我们介绍了目标的运动模型并分析了它们的多普勒频谱,在此基础上,我们刺激了目标雷达回波的时频图像。由于回波中的杂波通常服从威布尔分布,因此我们将威布尔杂波(Mezache和Soltani)添加到一种新颖的阈值优化技术中,以使用模糊神经网络对威布尔杂波进行远距离检测,2007,[1])。然后,我们使用NVIDIA DIGITS根据图像对具有不同网络的目标进行分类,并分析分类结果。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号