目的 针对可见光-近红外多模态探测背景下,传统迷彩设计在光谱特性与真实背景匹配度方面的不足,本研究提出一种将人工智能生成图像与材料光谱特性深度融合的“图谱合一”设计方法,旨在显著提升多模态伪装效能。方法 首先利用Stable Diffusion（SD）模型生成三色迷彩图案,并通过综合相似度评估筛选出最优方案。随后,在纹理斑块尺寸受限的条件下,对材料光谱特性进行迭代优化设计,并据此制备1 m×1 m三色迷彩标准样板。实验采用无人机搭载400⁓1 000 nm波段高光谱成像仪,在50 m高度采集数据;从中提取400⁓700 nm可见光波段的反射率光谱曲线,并反演生成相应的光学图像。光谱相似性采用相关系数进行定量评估,同时结合YOLOv12目标检测模型对伪装效果进行客观检验。结果 将SD模型生成的纹理图案与经迭代优化的材料光谱特性相结合的“图谱合一”策略,可使迷彩标准样板在400⁓700 nm波段的反射率曲线与草地背景实现高度吻合。光谱相似度指标与YOLOv12光学检测结果呈现出高度一致性,充分验证了该方法的有效性。结论 提出的“图谱合一”协同设计方法能够有效对抗可见光+高光谱复合探测,为图谱多模探测条件下的伪装一体化设计提供了新思路与技术路径。
Abstract
The work aims to propose a "spectrum-pattern integration" camouflage design method that fuses AI-generated imagery with material spectral characteristics to enhance multimodal concealment against visible-NIR detection. Stable Diffusion (SD) was employed to generate tricolor camouflage patterns, with optimal designs selected via comprehensive similarity assessment. Material spectral properties were iteratively optimized under fixed patch-size constraints, yielding 1 m × 1 m camouflage standard panels. Hyperspectral data (400-1 000 nm) were acquired via UAV at 50 m altitude; 400-700 nm reflectance curves were extracted and inverted to optical images. Spectral similarity was quantified using correlation coefficients, and camouflage effectiveness was objectively validated through YOLOv12 target detection. Results demonstrated that the integrated strategy achieved high spectral conformity with grass backgrounds in the 400-700 nm range, with strong consistency between spectral metrics and YOLOv12 detection performance, fully verifying the effectiveness of this method. This approach effectively counters combined visible-hyperspectral threats, offering a novel ideas and technical pathway for integrated spectrum-pattern camouflage design under the condition of multi-mode detection of spectra.
关键词
稳定扩散模型（Stable Diffusion） /
迭代材料光谱设计 /
YOLOv12检测模型 /
综合相似度评估
Key words
stable diffusion (SD) /
iterative material spectral design /
YOLOv12 detection model /
comprehensive similarity evaluation
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] LIU Y, WANG C Q, ZHOU Y J.Camouflaged People Detection Based on a Semi-Supervised Search Identification Network[J]. Defence Technology, 2023, 21:176-183.
[2] CHENG C Y, LIU J C, WANG F Q, et al.Photonic Structures in Multispectral Camouflage:From Static to Dynamic Technologies[J]. Materials Today, 2025, 85:253-281.
[3] 余松林, 陈玉华, 何鹄, 等. 基于颜色聚类的光学伪装效果评估背景选取方法[J]. 兵工学报, 2021, 42(3):617-624.
YU S L, CHEN Y H, HE H, et al.Background Selection Method of Optical Camouflage Effect Evaluation Based on Color Clustering[J]. Acta Armamentarii, 2021, 42(3):617-624.
[4] HE Z X, GAN Y Y, MA S X, et al.Evaluation Method for the Hyperspectral Image Camouflage Effect Based on Multifeature Description and Grayscale Clustering[J]. EURASIP Journal on Advances in Signal Processing, 2023, 2023(1):11.
[5] 喻钧, 双晓. 仿造数码迷彩的设计方法[J]. 应用科学学报, 2012, 30(4):331-334.
YU J, SHUANG X.Design of Imitation Digital Camouflage[J]. Journal of Applied Sciences, 2012, 30(4):331-334.
[6] XUE F, XU S, LUO Y T, et al.Design of Digital Camouflage by Recursive Overlapping of Pattern Templates[J]. Neurocomputing, 2016, 172:262-270.
[7] GONG Y M, WANG H B, LUO J X, et al.Research Progress of Bioinspired Structural Color in Camouflage[J]. Materials, 2024, 17(11):2564.
[8] LU H P, BAI X Z, WANG Z X, et al.Hyperspectral Camouflage Coating Using Palygorskite to Simulate Water Absorption of Healthy Green Leaves[J]. Materials Science in Semiconductor Processing, 2023, 156:107293.
[9] 徐晨, 苗珊珊, 张伟刚, 等. 基于绿色植被同色同谱的高光谱填料及性能研究[J]. 包装工程, 2025, 46(17):51-57.
XU C, MIAO S S, ZHANG W G, et al.Research on Hyperspectral Filler and Its Performance Based on Green Vegetation with the Same Color and Spectrum[J]. Packaging Engineering, 2025, 46(17):51-57.
[10] LI Y R, WANG F Q, ZHANG A Y, et al.Performance of the Multilayer Film for Infrared Stealth Based on VO2 Thermochromism[J]. Journal of Thermal Science, 2024, 33(4):1312-1324.
[11] DENG Z C, SU Y R, QIN W, et al.Nanostructured Ge/ZnS Films for Multispectral Camouflage with Low Visibility and Low Thermal Emission[J]. ACS Applied Nano Materials, 2022, 5(4):5119-5127.
[12] WU Q F, ZHANG R, ZHANG J T, et al.Sprayed AgNWs Interfacial Engineering Enabling Hyperspectral-Infrared Compatible Camouflage in Biomimetic Bilayer Coatings[J]. Journal of Colloid and Interface Science, 2026, 702:138813.
[13] QING X L, WENG X L, YUAN L, et al.Hyperspectral Camouflage with an Inorganic Coating Mimicking Vegetation in the 1.5-1.8 Μm Spectral Window[J]. Infrared Physics & Technology, 2026, 152:106246.
[14] HUPEL T, STÜTZ P. Adopting Hyperspectral Anomaly Detection for near Real-Time Camouflage Detection in Multispectral Imagery[J]. Remote Sensing, 2022, 14(15):3755.
[15] WANG X H, WANG Y H, MU Z H, et al.UFBSM:Unmixing Fusion and Background Sparse Dictionary Model for Hyperspectral Anomaly Detection[J]. International Journal of Remote Sensing, 2024, 45(11):3541-3559.
[16] LI Z, WANG L, LIU X, et al. Brochosome-Inspired Binary Metastructures for Pixel-by-Pixel Thermal Signature Control[J]. Science Advances, 2024, 10(9):eadl4027.
[17] ZHU R X, ZHU H Z, QIN B, et al.Digital Camouflage Encompassing Optical Hyperspectra and Thermal Infrared-Terahertz-Microwave Tri-Bands[J]. Nature Communications, 2025, 16:8112.
[18] HUANG F, YANG G H, CHEN J, et al.Semantic Segmentation of Camouflage Objects via Fusing Reconstructed Multispectral and RGB Images[J]. Defence Technology, 2025, 50:324-337.
[19] LYU X, REN X.Inverse Design of Composite Materials based on Latent Space and Bayesian Optimization[J]. Computer Modeling in Engineering & Sciences, 2026, 146(1):1-11.
[20] 徐晨. 融入视觉感知特性的迷彩伪装设计和评价方法研究[D]. 南京:南京航空航天大学, 2023.
XU C.Research on Camouflage Design and Evaluation Methods Based on Human Visual Perception[D]. Nanjing:Nanjing University of Aeronautics and Astronautics, 2023.
[21] 钱淇. 基于绿色植被背景的多频谱兼容隐身方法及材料制备研究[D]. 南京:南京航空航天大学, 2023.
QIAN Q.Research on Multi Spectrum Compatible Stealth Method and Material Preparation Based on Green Vegetation Background[D]. Nanjing:Nanjing University of Aeronautics and Astronautics, 2023.
[22] INAMDAR D, LEBLANC G, SOFFER R J, et al.The Correlation Coefficient as a Simple Tool for the Localization of Errors in Spectroscopic Imaging Data[J]. Remote Sensing, 2018, 10(2):231.
[23] BINGYAN C, LIU Z, YANG Q.UAV-YOLO12:A Multi-scale Road Segmentation Model for UAV Remote Sensing Imagery[J]. Drones, 2025, 9(8):533.
[24] AL RABBANI ALIF M, HUSSAIN M. YOLOv12:A Breakdown of the Key Architectural Features[EB/OL].2025:arXiv:2502.14740. https://arxiv.org/abs/2502.14740
[25] SHA H, CERPENTIER J, et al.Spectral Reflectance Imaging with Dual-Illumination and RGB Camera via Regularized End-to-End Learning[J]. Optics Express, 2025, 33(21):44191.
[26] WEI C N, LI J F, LIU S W.Applications of Visible Spectral Imaging Technology for Pigment Identification of Colored Relics[J]. Heritage Science, 2024, 12:321.
基金
“十四五”ZB发展共用技术预研（509080402XX）
PDF(4248 KB)
渝公网安备 50010702501716号 渝ICP备15012534号-2
