基于深度卷积神经网络的慢动目标检测

Slow moving target detection based on DCNN

针对强杂波背景下慢速运动目标检测性能不足的问题,设计了一种基于深度卷积神经网络(DCNN)的目标检测方法。主要将雷达回波信号距离—多普勒谱作为输入,送入设计的DCNN中,通过学习回波信号中杂波特征,并隐含的去除回波信号中目标成分,得到回波信号的残差谱。然后利用残差谱与回波信号R-D谱进行背景对消以抑制杂波,进而实现对回波信号中运动目标的检测。由于该方法通过学习杂波特性进而进行目标检测,因此适用于未知杂波模型的场景,避免了假设的模型与实际环境不符合的问题。实验验证:该方法相比于传统的杂波抑制目标检测方法,具有较好的性能表现。

To solve the performance problem of slow moving target detection on the condition of strong clutter background, an algorithm based on deep convolutional neural network(DCNN)is proposed.The Range-Doppler spectrum of the echo signal is taken as the input and sent into the DCNN.And the residual spectrum of the echo signal can be obtained by learning the clutter feature in the echo signal and removing the target components implicitly.Then, the background subtraction between the residual spectrum and R-D spectrum of the echo signal is carried out to suppress the clutter, so as to realize the detection of the moving target in the echo signal.Since this method can detect the target by learning the clutter characteristics, it is suitable for the scene of unknown clutter model and avoids the problem of mismatch between the model hypothesis and the actual environment.Experimental results show that this method can achieve better detection performance than traditional methods on slow moving targets detection on the condition of strong clutter background.