Research on an Attitude Controllable Radar Image Generation Method for Noncooperative Targets

Read full text Summarise with VIRAT
Authors: Xiaopeng Xue, Wenxiang Zhong, Degui Yang, Xing Wang, Zhichen Zhao
Cited by: 0
References: 11

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

The generation of radar images of noncooperative targets has always been a hot topic and difficulty in this field. Although the existing methods based on generative adversarial networks can solve the problem of insufficient number of radar images of noncooperative targets, they lack precise control over the target’s pose during image generation. In view of this, inspired by the ControlNet model’s ability to incorporate additional conditions for controllable generation in optical image synthesis, this paper develops a method for generating radar images of noncooperative targets with precise pose control based on the ControlNet model. In this paper, the geometric model of noncooperative targets is first constructed, and a limited number of target radar images are obtained through electromagnetic calculation. Secondly, a frequency domain denoising method is proposed to filter out the stripe noise in the radar images; meanwhile, Lee filter is applied to remove speckle noise from the radar images, thus building a database of radar images for noncooperative targets. Finally, the edge map of the optical images under different poses are used as control conditions to realize the generation of radar images of noncooperative targets with precise and controllable poses based on the ControlNet model. Simulation results demonstrate that this method can generate radar images of noncooperative targets with accurately controlled azimuth and elevation angles. Especially when the number of training samples is limited, the generated results outperform existing algorithms in terms of structural similarity, peak signal-to-noise ratio, and Fréchet inception distance.

Fields of Study & Subjects

Advanced SAR Imaging Techniques Generative Adversarial Networks and Image Synthesis Radar Systems and Signal Processing

Publication details

Journal / source
Space Science & Technology
Volume
6
Publication date
1 January 2026
Language
EN