火力与指挥控制2023，Vol.48Issue(12)：34-40,7.DOI:10.3969/j.issn.1002-0640.2023.12.005
基于生成对抗网络的空中红外小目标仿真模型
Infrared Aerial Small Target Simulation Model Based on Generative Adversarial Network
摘要
Abstract
The infrared aerial small targets can be applied in such military fields as target detection and target tracking,etc.but it is costly to obtain them.In the WGAN model,the noise reduction algorithm is introduced to improve the generated image quality and the Wasserstein distance penalty term is improved.An infrared aerial small target simulation model based on generative adversarial network(IST-WGAN)is proposed.The model is trained on the self-built aerial infrared small target datasets.The infrared images generated by the model are verified by subjective discrimination,FID score and small target detection algorithm.The results show that the proposed model can generate the effective infrared small target samples.关键词
距离惩罚/降噪算法/生成对抗网络/红外小目标Key words
distance penalty/noise reduction algorithm/generative adversarial network(GAN)/infrared small targets分类
信息技术与安全科学引用本文复制引用
李兴海,臧绍飞,马建伟,马晓毓..基于生成对抗网络的空中红外小目标仿真模型[J].火力与指挥控制,2023,48(12):34-40,7.基金项目
国家自然科学基金(62073124) （62073124）
河南省高等学校重点科研基金资助项目(22A120005) （22A120005）
