Infrared Image Generation Method Based on Improved CycleGAN

doi:10.16182/j.issn1004731x.joss.25-0998

Abstract

Abstract:

To address the problems in current infrared image generation such as insufficient contrast between target and scene, excessively large discrepancies from real scenes, indistinct thermal source features, and great difficulty in constructing measured infrared image datasets, an improved CycleGAN-based infrared image generation method was proposed. By optimizing the network structure of the generator and adding a non-local module and a CBAM convolutional attention mechanism into the generator, the extraction capability of CycleGAN for infrared features was enhanced, enabling the network to capture the subtle features of targets more accurately; a perceptual loss function was introduced to improve the detail retention and perceptual quality of the generated infrared images and enhance the stability of model training, realizing high-quality image conversion from visible light to infrared image. Experimental results demonstrate that the infrared images generated based on this method not only effectively improve image quality and highlight thermal source features but also significantly reduce the cost and difficulty of traditional infrared image acquisition, providing a more convenient and efficient solution for the acquisition and analysis of infrared images in practical applications.

Key words: infrared image, CycleGAN network, convolutional attention module, cross-modal conversion

CLC Number:

TP389.1

Jin Qiqi, Zhang Xiang, Gao Li, Zhang Lin, Yan Junliang, Li Peiyao. Infrared Image Generation Method Based on Improved CycleGAN[J]. Journal of System Simulation, 2026, 38(8): 2425-2435.

Figures/Tables 12

Fig. 1

Fig. 2

Fig. 3

Fig. 4

Fig. 5

Fig. 6

Fig. 7

Fig. 8

Fig. 9

Fig. 10

Table 1

Table 2

References 25

[1]	蔡伟, 姜波, 蒋昕昊, 等. 非成对训练样本条件下的红外图像生成[J]. 光学精密工程, 2023, 31(24): 3651-3661.
	Cai Wei, Jiang Bo, Jiang Xinhao, et al. Infrared Image Generation with Unpaired Training Samples[J]. Optics and Precision Engineering, 2023, 31(24): 3651-3661.
[2]	冯云飞. 基于实拍图像的红外图像波段拓展方法研究[D]. 西安: 西安电子科技大学, 2019.
	Feng Yunfei. Research on Band Expansion Method for Infrared Image Based on Real Image[D]. Xi'an: Xidian University, 2019.
[3]	Zhu Junyan, Park T, Isola P, et al. Unpaired Image-to-image Translation Using Cycle-consistent Adversarial Networks[C]//2017 IEEE International Conference on Computer Vision (ICCV). Piscataway: IEEE, 2017: 2242-2251.
[4]	化彦伶. 基于生成对抗网络的红外图像生成方法研究[D]. 武汉: 华中科技大学, 2020.
	Hua Yanling. Research on Infrared Image Generation Method Based on Generative Adversarial Network[D]. Wuhan: Huazhong University of Science & Technology, 2020.
[5]	李彤. 可见光与红外图像的特征转换方法研究[D]. 北京: 北方工业大学, 2021.
	Li Tong. Research on Feature Conversion Between Visible Light and Infrared Images[D]. Beijing: North China University of Technology, 2021.
[6]	Woo Sanghyun, Park Jongchan, Lee J Y, et al. CBAM: Convolutional Block Attention Module[C]//Computer Vision-ECCV 2018. Cham: Springer International Publishing, 2018: 3-19.
[7]	He Kaiming, Zhang Xiangyu, Ren Shaoqing, et al. Deep Residual Learning for Image Recognition[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway: IEEE, 2016: 770-778.
[8]	Wang Xiaolong, Girshick R, Gupta A, et al. Non-local Neural Networks[C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2018: 7794-7803.
[9]	石宝, 周昊, 武文红. 基于改进CycleGAN的花粉灰度图像着色方法[J]. 北京工业大学学报, 2025, 51(9): 1063-1070.
	Shi Bao, Zhou Hao, Wu Wenhong. Pollen Gray-scale Image Colorizing Method Based on Improved CycleGAN[J]. Journal of Beijing University of Technology, 2025, 51(9): 1063-1070.
[10]	李艳玲, 张博翔, 李飞涛, 等. 基于改进CycleGAN的小样本玉米病害图像扩充方法[J]. 华中农业大学学报, 2025, 44(5): 198-207.
	Li Yanling, Zhang Boxiang, Li Feitao, et al. Improved CycleGAN-based Method for Augmenting Images of Few-shot Maize Diseases[J]. Journal of Huazhong Agricultural University, 2025, 44(5): 198-207.
[11]	刘晗. 基于图像风格迁移的红外视景仿真技术研究[D]. 武汉: 湖北大学, 2024.
	Liu Han. Research on Infrared View Simulation Technology Based on Image Style Transfer[D]. Wuhan: Hubei University, 2024.
[12]	王村松, 徐明宇, 张泉灵, 等. 基于注意力机制的轻量化刀具磨损监测方法[J]. 组合机床与自动化加工技术, 2025(6): 13-19, 25.
	Wang Cunsong, Xu Mingyu, Zhang Quanling, et al. A Lightweight Tool Wear Monitoring Method Based on Attention Mechanism[J]. Modular Machine Tool & Automatic Manufacturing Technique, 2025(6): 13-19, 25.
[13]	Guo Xiaojun, Wang Yifei, Du Tianqi, et al. ContraNorm: A Contrastive Learning Perspective on Oversmoothing and Beyond[EB/OL]. (2023-05-02) [2025-10-14]. .
[14]	石丽芬, 张鹏, 景亚蔓, 等. 改进CycleGAN实现可见光红外图像的迁移(特邀)[J]. 红外与激光工程, 2025, 54(3): 352-366.
	Shi Lifen, Zhang Peng, Jing Yaman, et al. Improved CycleGAN Algorithm to Transfer Visible Images to Infrared Images(Invited)[J]. Infrared and Laser Engineering, 2025, 54(3): 352-366.
[15]	李廷轩. 红外图像仿真和生成方法研究[D]. 西安: 西安电子科技大学, 2024.
	Li Tingxuan. Research on Simulation and Generation Methods of Infrared Images[D]. Xi'an: Xidian University, 2024.
[16]	Han Zonghao, Zhang Ziye, Zhang Shun, et al. Aerial Visible-to-infrared Image Translation: Dataset, Evaluation, and Baseline[J]. Journal of Remote Sensing, 2023, 3: 0096.
[17]	Han Zonghao, Zhang Shun, Su Yuru, et al. DR-AVIT: Toward Diverse and Realistic Aerial Visible-to-infrared Image Translation[J]. IEEE Transactions on Geoscience and Remote Sensing, 2024, 62: 1-13.
[18]	牛玉贞, 陈铭铭, 李悦洲, 等. 基于任务解耦的低照度图像增强方法[J]. 电子学报, 2024, 52(1): 34-45.
	Niu Yuzhen, Chen Mingming, Li Yuezhou, et al. Task Decoupling Guided Low-light Image Enhancement[J]. Acta Electronica Sinica, 2024, 52(1): 34-45.
[19]	黄哲凯. 基于BSP算法的3D高斯泼溅渲染外观优化研究[D]. 南昌: 江西财经大学, 2025.
	Huang Zhekai. Research on 3D Gaussian Splatting Appearance Optimization Based on BSP Tree[D]. Nanchang: Jiangxi University of Finance and Economics, 2025.
[20]	郑雯青. 基于场景图理解的图像外推方法研究[D]. 北京: 北京邮电大学, 2024.
	Zheng Wenqing. Image Outpainting Based on Scene Graph Understanding[D]. Beijing: Beijing University of Posts and Telecommunications, 2024.
[21]	Isola P, Zhu Junyan, Zhou Tinghui, et al. Image-to-image Translation with Conditional Adversarial Networks[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway: IEEE, 2017: 5967-5976.
[22]	Han Junlin, Shoeiby Mehrdad, Petersson Lars, et al. Dual Contrastive Learning for Unsupervised Image-to-image Translation[C]//2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW). Piscataway: IEEE, 2021: 746-755.
[23]	Fu Huan, Gong Mingming, Wang Chaohui, et al. Geometry-consistent Generative Adversarial Networks for One-sided Unsupervised Domain Mapping[C]//2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway: IEEE, 2019: 2422-2431.
[24]	Park T, Efros A A, Zhang R, et al. Contrastive Learning for Unpaired Image-to-image Translation[C]//Computer Vision-ECCV 2020. Cham: Springer International Publishing, 2020: 319-345.
[25]	Liu M Y, Breuel T, Kautz J. Unsupervised Image-to-image Translation Networks[EB/OL]. (2017-03-08) [2025-10-08]..

Method	AVIID				DroneVehicle
Method	SSIM	PRSN/dB	FID	LPIPS	SSIM	PRSN/dB	FID	LPIPS
Pix2pix	0.659	21.21	72.09	0.197	0.446	11.88	212.80	0.596
CycleGAN	0.634	19.20	58.89	0.231	0.486	11.81	121.96	0.482
DCLGAN	0.631	18.19	81.71	0.283	0.466	11.95	189.21	0.574
GcGAN	0.538	14.13	76.50	0.398	0.468	12.81	57.62	0.617
CUT	0.553	19.09	57.67	0.233	0.460	11.78	157.36	0.520
UNIT	0.653	20.84	61.27	0.218 4	0.470	11.75	218.76	0.587
本文	0.669	21.38	53.03	0.175	0.519 2	13.76	65.89	0.366

Method	CBAM	NLB	Perceptual Loss	AVIID
Method	CBAM	NLB	Perceptual Loss	SSIM↑	PRSN/dB↑	FID↓	LPIPS↓
Baseline				0.634	19.20	58.89	0.231
Model-Ⅰ	√		√	0.645	20.42	56.44	0.207
Model-Ⅱ	√	√		0.650	20.69	54.73	0.187
Model-Ⅲ		√	√	0.659	21.21	53.36	0.181
本文方法	√	√	√	0.669	21.38	53.03	0.175