Ground Background Clutter Recognition Based on Fully Convolutional Neural Network
- Resource Type
- Conference
- Authors
- Liu, Chang; Lang, Ping; Fu, Xiongjun; Dong, Jian; Li, Mingling; Qi, Xinyue
- Source
- 2021 CIE International Conference on Radar (Radar) Radar (Radar), 2021 CIE International Conference on. :1850-1853 Dec, 2021
- Subject
- Aerospace
Communication, Networking and Broadcast Technologies
Computing and Processing
Fields, Waves and Electromagnetics
Geoscience
Robotics and Control Systems
Signal Processing and Analysis
Adaptation models
Target recognition
Radar clutter
Radar detection
Detectors
Object detection
Robustness
background clutter
clutter recognition
Range-Doppler
fully convolutional neural network
- Language
- ISSN
- 2640-7736
Accurate and robust recognition of background clutter is essential for radar target detection. Aiming at the problems that existing clutter recognition methods have low accuracy and poor robustness, a background clutter recognition method based on novel Fully Convolutional Neural Network (FCNN) is proposed. FCNN is trained and tested based on the simulated ground clutter Range-Doppler (R-D) spectra. Experimental results demonstrate that FCNN is significantly superior to the existing background clutter recognition models in terms of clutter classification accuracy or time complexity. In addition, FCNN can better adapt to clutter recognition under low clutter-to-noise ratio (CNR) scenarios. Finally, we verify the effectiveness of the CFAR detector based on clutter recognition.