基于频率净化和交互融合的无人机可见光-红外目标检测

doi:10.7527/S1000-6893.2026.33818

Abstract

Abstract:

Visible-infrared object detection for Unmanned Aerial Vehicles （UAV） can exploit the texture details of visible images and the thermal radiation information of infrared images， thereby improving target perception capability in complex environments. However， existing methods mainly focus on cross-modal feature interaction and feature fusion， while ignoring high-frequency interference in single-modal features before fusion， such as thermal noise， pseudo-hotspots， texture redundancy， and background edges. As a result， high-frequency noise is further amplified during cross-modal propagation， reducing detection accuracy. To address these problems， this paper proposes a UAV visible-infrared object detection method based on frequency purification and interactive fusion. Through a “purification-before-fusion” strategy， the proposed method suppresses single-modal high-frequency interference and enhances cross-modal complementary information. Specifically， a Low-frequency Aggregation Frequency Purification （LAFP） module is designed to use cross-modal low-frequency structural priors to guide the selection and purification of high-frequency detail features in each modality， suppressing invalid high-frequency responses while preserving target edges and local discriminative features. In addition， a Bidirectional Interactive Guidance Fusion （BIGF） module is introduced. This module employs bidirectional cross-attention to mine the complementary relationships between purified visible and infrared features， and combines a dynamic gating mechanism to achieve adaptive cross-modal feature fusion. Experimental results on three visible-infrared object detection datasets， namely DroneVehicle， VEDAI， and LLVIP， show that the proposed method achieves good performance in both complex aerial vehicle detection and low-light pedestrian detection scenarios. Compared with existing state-of-the-art methods， the proposed method achieves mean Average Precision （mAP） values of 82.4% and 80.5% on the DroneVehicle and VEDAI datasets， respectively， outperforming the second-best methods by 3.0% and 2.7%. On the LLVIP dataset， it obtains an average precision of 96.1%， which is 0.7% and 0.4% lower than the best and second-best results， respectively， but still demonstrates strong competitiveness. The experimental results demonstrate that the proposed algorithm has high effectiveness and robustness in UAV object detection scenarios.

Key words: UAV, visible-infrared, frequency purification, cross-attention, object detection

CLC Number:

V249

Shuang SU, Zhiyong LI, Ruitao LU, Lixin LI, Xiaogang YANG. Frequency purification and interactive fusion for UAV-based VIS-IR object detection[J]. Acta Aeronautica et Astronautica Sinica, 2026, 47(19): 633818.

Figures/Tables 14

Fig.1

Fig.2

Fig.3

Fig.4

Fig.5

Fig.6

Table 1

Table 2

Fig.7

Table 3

Fig.8

Table 4

Fig.9

Table 5

References 41

[1]	WAN Y T， YAO H Y， LIU J W， et al. Low-light and infrared multimodal remote sensing in nighttime rescue mission： A review of anomaly detection methods［J］. IEEE Transactions on Geoscience and Remote Sensing， 2025， 63： 5403518.
[2]	吴一全，童康. 基于深度学习的无人机航拍图像小目标检测研究进展［J］. 航空学报， 2025， 46（3）： 030848.
	WU Y Q， TONG K. Research advances on deep learning-based small object detection in UAV aerial images［J］. Acta Aeronautica et Astronautica Sinica， 2025， 46（3）： 030848 （in Chinese）.
[3]	HOANH N， PHAM T VU. A multi-task framework for car detection from high-resolution UAV imagery focusing on road regions［J］. IEEE Transactions on Intelligent Transportation Systems， 2024， 25（11）： 17160-17173.
[4]	WANG R P， LIN C， LI Y J. RPLFDet： A lightweight small object detection network for UAV aerial images with rational preservation of low-level features［J］. IEEE Transactions on Instrumentation and Measurement， 2025， 74： 5013514.
[5]	HOU T， LENG C C， WANG J X， et al. MFEL-YOLO for small object detection in UAV aerial images［J］. Expert Systems with Applications， 2025， 291： 128459.
[6]	HOU Y L， SONG Y Y， HAO X L， et al. Multispectral pedestrian detection based on deep convolutional neural networks［C］∥2017 IEEE International Conference on Signal Processing， Communications and Computing （ICSPCC）. Piscataway： IEEE Press， 2018： 1-4.
[7]	PENG R H， LAI J， YANG X T， et al. Camouflaged target detection based on multimodal image input pixel-level fusion［J］. Frontiers of Information Technology & Electronic Engineering， 2024， 25（9）： 1226-1239.
[8]	YUAN M X， WEI X X. C²Former： Calibrated and complementary transformer for RGB-infrared object detection［J］. IEEE Transactions on Geoscience and Remote Sensing， 2024， 62： 5403712.
[9]	SUN X， YU Y H， CHENG Q. Low-rank multimodal remote sensing object detection with frequency filtering experts［J］. IEEE Transactions on Geoscience and Remote Sensing， 2024， 62： 5637114.
[10]	CHAOXIA C， SHANG W W， ZHANG F， et al. Weakly aligned multimodal flame detection for fire-fighting robots［J］. IEEE Transactions on Industrial Informatics， 2023， 19（3）： 2866-2875.
[11]	LI Q， ZHANG C Q， HU Q H， et al. Confidence-aware fusion using dempster-shafer theory for multispectral pedestrian detection［J］. IEEE Transactions on Multimedia， 2023， 25： 3420-3431.
[12]	HU Z H， JING Y G， WU G Q. Decision-level fusion detection method of visible and infrared images under low light conditions［J］. EURASIP Journal on Advances in Signal Processing， 2023， 2023（1）： 38.
[13]	李峻宇，刘乾坤，付莹. 融合注意力机制的红外小目标检测［J］. 航空学报， 2024， 45（14）： 628959.
	LI J Y， LIU Q K， FU Y. Infrared small object detection based on attention mechanism［J］. Acta Aeronautica et Astronautica Sinica， 2024， 45（14）： 628959 （in Chinese）.
[14]	HU S J， BONARDI F， BOUCHAFA S， et al. Rethinking self-attention for multispectral object detection［J］. IEEE Transactions on Intelligent Transportation Systems， 2024， 25（11）： 16300-16311.
[15]	DONG S H， XIE W Y， YANG D N， et al. SeaDATE： Remedy dual-attention transformer with semantic alignment via contrast learning for multimodal object detection［J］. IEEE Transactions on Circuits and Systems for Video Technology， 2025， 35（5）： 4713-4726.
[16]	LIU Y Y， DIAN R W， LI S T. Low-rank transformer for high-resolution hyperspectral computational imaging［J］. International Journal of Computer Vision， 2025， 133（2）： 809-824.
[17]	YUAN M X， MENG D N， XI Z T， et al. Seeing through the noise： Improving infrared small target detection and segmentation from noise suppression perspective［C］∥Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE Press； 2026： 27783-27792.
[18]	FANG Q Y， HAN D P， WANG Z K. Cross-modality fusion transformer for multispectral object detection［DB/OL］. arXiv preprint： 2111.00273， 2021.
[19]	ZHANG L， LIU Z Y， ZHANG S F， et al. Cross-modality interactive attention network for multispectral pedestrian detection［J］. Information Fusion， 2019， 50： 20-29.
[20]	ZHANG L， LIU Z Y， ZHU X Y， et al. Weakly aligned feature fusion for multimodal object detection［J］. IEEE Transactions on Neural Networks and Learning Systems， 2025， 36（3）： 4145-4159.
[21]	YANG F， LIANG B B， LI W， et al. Multidimensional fusion network for multispectral object detection［J］. IEEE Transactions on Circuits and Systems for Video Technology， 2025， 35（1）： 547-560.
[22]	SUN Y M， CAO B， ZHU P F， et al. Drone-based RGB-infrared cross-modality vehicle detection via uncertainty-aware learning［J］. IEEE Transactions on Circuits and Systems for Video Technology， 2022， 32（10）： 6700-6713.
[23]	YUAN M X， WANG Y Y， WEI X X. Translation， scale andRotation： Cross-modal alignment meets RGB-infrared vehicle detection［C］∥Computer Vision-ECCV 2022. Cham： Springer， 2022： 509-525.
[24]	ZHOU K L， CHEN L S， CAO X. Improving multispectral pedestrian detection by addressing modality imbalance problems［C］∥Computer Vision-ECCV 2020. Cham： Springer， 2020： 787-803.
[25]	ZHANG Y， YU H， HE Y J， et al. Illumination-guided RGBT object detection with inter- and intra-modality fusion［J］. IEEE Transactions on Instrumentation and Measurement， 2023， 72： 2508013.
[26]	WANG H Y， WANG C P， FU Q， et al. Cross-modal oriented object detection of UAV aerial images based on image feature［J］. IEEE Transactions on Geoscience and Remote Sensing， 2024， 62： 5403021.
[27]	刘奎，孙浩，伍瀚，等. 动态亮度重建的无人机可见光-红外融合目标检测［J］. 航空学报， 2025， 46（23）： 631968.
	LIU K， SUN H， WU H， et al. Dynamic brightness reconstruction for UAV visible-infrared fusion object detection［J］. Acta Aeronautica et Astronautica Sinica， 2025， 46（23）： 631968 （in Chinese）.
[28]	ZENG Y Q， LIANG T F， JIN Y， et al. MMI-det： Exploring multi-modal integration for visible and infrared object detection［J］. IEEE Transactions on Circuits and Systems for Video Technology， 2024， 34（11）： 11198-11213.
[29]	ZHU H D， DONG W H， YANG L L， et al. WaveMamba： Wavelet-driven mamba fusion for RGB-infrared object detection［C］∥2025 IEEE/CVF International Conference on Computer Vision （ICCV）. Piscataway： IEEE Press， 2026： 11219-11229.
[30]	WU W C， ZHANG X W， YIN H L， et al. FreDFT： Frequency domain fusion transformer for visible-infrared object detection［J］. IEEE Transactions on Multimedia， 2026： 1-12.
[31]	HU Y， CHEN X B， WANG S， et al. Deformable cross-attention transformer for weakly aligned RGB-T pedestrian detection［J］. IEEE Transactions on Multimedia， 2025， 27： 4400-4411.
[32]	SHEN J F， CHEN Y F， LIU Y， et al. ICAFusion： Iterative cross-attention guided feature fusion for multispectral object detection［J］. Pattern Recognition， 2024， 145： 109913.
[33]	HU K， HE Y D， LI Y， et al. EI²Det： Edge-guided illumination-aware interactive learning for visible-infrared object detection［J］. IEEE Transactions on Circuits and Systems for Video Technology， 2025， 35（7）： 7101-7115.
[34]	RAZAKARIVONY S， JURIE F. Vehicle detection in aerial imagery： A small target detection benchmark［J］. Journal of Visual Communication and Image Representation， 2016， 34： 187-203.
[35]	JIA X Y， ZHU C， LI M Z， et al. LLVIP： A visible-infrared paired dataset for low-light vision［C］∥2021 IEEE/CVF International Conference on Computer Vision Workshops （ICCVW）. Piscataway： IEEE Press， 2021： 3489-3497.
[36]	HE X， TANG C， ZOU X， et al. Multispectral object detection via cross-modal conflict-aware learning［C］∥Proceedings of the 31st ACM International Conference on Multimedia. New York： ACM， 2023： 1465-1474.
[37]	ZHANG J Q， LEI J， XIE W Y， et al. SuperYOLO： Super resolution assisted object detection in multimodal remote sensing imagery［J］. IEEE Transactions on Geoscience and Remote Sensing， 2023， 61： 5605415.
[38]	ZHANG J Q， CAO M X， XIE W Y， et al. E2E-MFD： Towards end-to-end synchronous multimodal fusion detection［C］∥Advances in Neural Information Processing Systems 37. Neural Information Processing Systems Foundation， Inc. （NeurIPS）， 2024： 52296-52322.
[39]	BAO W， HUANG M Y， HU J J， et al. Dual-dynamic cross-modal interaction network for multimodal remote sensing object detection［J］. IEEE Transactions on Geoscience and Remote Sensing， 2025， 63： 5401013.
[40]	ALTHOUPETY A， WANG L Y， FENG W C， et al. DaFF： Dual attentive feature fusion for multispectral pedestrian detection［C］∥2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops （CVPRW）. Piscataway： IEEE Press， 2024： 2997-3006.
[41]	ZHANG T， LU R T， YANG X G， et al. CCSFuse： Collaborative compensation and selective fusion for UAV-based RGB-IR object detection［J］. IEEE Transactions on Geoscience and Remote Sensing， 2026， 64： 5400314.

LAFP	BIGF	mAP50/%	AP/%
LAFP	BIGF	mAP50/%	Car	Truck	Bus	Van	Freight-car
		73.2	94.1	60.5	89.8	52.8	56.6
√		80.7	97.2	77.5	94.9	63.3	74.7
	√	78.6	95.7	75.2	93.1	61.4	71.5
√	√	82.4	98.4	79	96.4	64.4	76.5

算法	AP/%					类型	mAP50↑/%
算法	Car	Truck	Bus	Van	Freight-car	类型	mAP50↑/%
AR-CNN^［20］	90.1	64.8	89.4	51.5	62.1	两阶段	71.6
TSFADet^［23］	89.2	72.0	88.1	48.8	54.2	两阶段	70.4
UA-CMDet^［22］	87.51	60.7	87.1	37.9	46.8	两阶段	64.1
CFT^［18］	81.3	71.5	86.2	53.8	57.9	一阶段	70.1
CALNet^［36］	90.3	73.7	89.7	59.7	67.7	一阶段	76.4
SuperYOLO^［37］	81.5	64.1	80.6	51.5	56.3	一阶段	66.8
LF-MDet^［9］	82.2	73.6	86.6	57.0	59.6	一阶段	71.8
ICAFusion^［32］	81.6	56.0	85.7	31.8	33.3	一阶段	57.7
C2Former^［8］	87.6	58.1	80.3	53.8	55.2	一阶段	67.0
DaFF^［40］	89.6	73.9	89.9	58.8	62.6	一阶段	76.1
E2E-MFD^［38］	90.3	79.3	89.8	63.1	57.3	一阶段	77.4
DDCINet^［39］	91.0	78.9	90.7	65.5	66.1	两阶段	78.4
WaveMamba^［29］	95.0	75.9	91.4	64.5	67.7	一阶段	78.9
FreDFT^［30］	94.2	73.6	88.7	63.6	65.2	一阶段	77.6
CCSFuse^［41］	97.1	72.1	94.3	64.1	67.8	一阶段	79.4
本文算法	98.4	79.0	96.4	64.4	76.5	一阶段	82.4

算法	AP/%								mAP50↑/%
算法	Car	Pickup	Camping	Truck	Other	Tractor	Boat	Van	mAP50↑/%
CFT^［18］	93.6	83.3	73.9	74.8	51.0	75.9	64.7	72.1	73.6
CALNet^［36］	82.3	78.2	75.4	66.7	47.0	73.0	56.6	65.5	68.1
SuperYOLO^［37］	91.1	85.7	79.3	70.2	57.3	80.4	60.2	76.5	75.1
LF-MDet^［9］	90.8	84.9	81.6	74.3	57.8	80.7	62.2	78.0	76.9
ICAFusion^［32］	83.7	77.6	69.9	48.6	45.7	71.3	45.9	67.0	63.7
C2Former^［8］	87.2	80.7	82.7	77.4	58.4	72.9	71.4	75.2	75.7
DaFF^［40］	90.4	83.5	79.6	76.8	59.4	78.2	64.6	74.7	75.9
E2E-MFD^［38］	91.2	84.4	80.7	78.1	61.5	79.4	65.9	76.0	77.2
WaveMamba^［29］	89.7	80.6	81.5	80.5	57.8	78.0	58.7	75.4	73.2
FreDFT^［30］	85.3	75.0	68.8	80.5	57.9	76.1	64.3	60.0	71.5
CCSFuse^［41］	94.2	87.5	81.4	82.3	75.8	81.2	67.4	79.5	77.8
本文算法	95.1	89.7	82.1	81.9	74.2	78.8	71.6	79.3	80.5

算法	mAP50↑/%	算法	mAP50↑/%
CFT	96.5	DaFF	94.3
CALNet	94.8	E2E-MFD	95.8
SuperYOLO	93.8	WaveMamba	95.9
LF-MDet	95.5	FreDFT	95.7
ICAFusion	92.8	CCSFuse	96.8
C2Former	94.9	本文算法	96.1

算法	Params/10⁶	DroneVehicle				VEDAI				LLVIP
算法	Params/10⁶	mAP75/%	mAP/%	FPS	GFLOPs	mAP75/%	mAP/%	FPS	GFLOPs	mAP75/%	mAP/%	FPS	GFLOPs
CFT	206.0	69.3	56.6	13.9	116.7	43.0	35.0	15.2	74.3	70.7	62.9	14.7	116.6
SuperYOLO	4.8	70.9	57.6	67.3	28.1	45.1	38.3	73.3	18.0	70.2	61.7	35.6	28.1
LF-MDet	38.7	63.5	51.3		77.7	46.9	39.7		145.0	72.1	63.5		39.6
ICAFusion	120.2	70.2	58.0	39.6	85.3	45.0	37.0	43.7	54.6	71.3	62.4	40.4	85.3
C2Former	100.8	71.0	58.2	37.0	89.9	47.1	40.0	40.3	57.7	71.8	63.1	37.8	88.9
WaveMamba	69.1	73.4	60.5	25.0		46.5	39.8	27.3		72.4	63.7	25.7
FreDFT	152.6	73.0	60.4		234.7	45.8	39.0		148.5	72.7	64.3		232.1
CCSFuse	33.5	74.3	61.8	47.5	174.1	49.1	41.8	51.8	111.4	73.4	65.1	48.5	174.1
本文算法	8.9	75.0	62.2	125	16.7	50.5	43.2	175.4	16.7	73.3	64.0	130.2	16.7

Frequency purification and interactive fusion for UAV-based VIS-IR object detection

RichHTML

PDF (PC)

Knowledge

Abstract

Cite this article

share this article

Figures/Tables 14

References 41

Related Articles 15

Recommended Articles

Metrics

Comments

[1]	Xiaolong HE, Qifeng CHEN, Wenkang HAO. Distributed control of UAVs formation based on passivity-based sliding mode method [J]. Acta Aeronautica et Astronautica Sinica, 2026, 47(S1): 732965-732965.
[2]	Zijun ZHAO, Shitao CHEN, Weiyan HE, Longhao LIU, Zhenghao ZHANG. Intelligent assessment method for MAV/UAV collaborative combat effectiveness [J]. Acta Aeronautica et Astronautica Sinica, 2026, 47(S1): 732887-732887.
[3]	Bin YANG, Jun BIAN, Jianchun ZHANG, Kexin GUO, Yi YANG. Distributed safety cooperative control for multiple UAVs under multi-source risks [J]. Acta Aeronautica et Astronautica Sinica, 2026, 47(S1): 733010-733010.
[4]	Ruiping ZHENG, Jingping SHI, Tianyu LI, Yongxi LYU. Prescribed-time coordinated control for fixed-wing UAV close formation [J]. Acta Aeronautica et Astronautica Sinica, 2026, 47(9): 532496-532496.
[5]	Ying AO, Hao CHEN, Huiming LI, Kun XIAO, Xiangke WANG. Hierarchical control method for affine formation of fixed-wing UAV swarm [J]. Acta Aeronautica et Astronautica Sinica, 2026, 47(9): 532494-532494.
[6]	Xiaojuan WANG, Guangtai TIAN, Bin LI, Zhenqing HE. Fully actuated anti-disturbance control for quadrotor UAVs based on dynamic compensator [J]. Acta Aeronautica et Astronautica Sinica, 2026, 47(9): 123-135.
[7]	Wenyu ZHOU, Yingjie WANG, Chenhao OUYANG, Zixuan ZHENG, Xiaokui YUE. Bearing-based collision-avoidance formation control of quadrotor UAVs under internal and external disturbances [J]. Acta Aeronautica et Astronautica Sinica, 2026, 47(9): 533224-533224.
[8]	Mou CHEN, Zihan ZHEN, Bin JIANG. Overview of intelligent obstacle avoidance decision-making and safety control technologies for low-altitude UAV [J]. Acta Aeronautica et Astronautica Sinica, 2026, 47(9): 533388-533388.
[9]	Shufang XU, Wenxuan FEI, Heng LI, Hongmin GAO. UAV-UGV collaborative air-ground path planning method based on three-stage optimization [J]. Acta Aeronautica et Astronautica Sinica, 2026, 47(7): 332649-332649.
[10]	Yunxiao LIAN, Ni LI, Feng XIE, Pan ZHOU, Changyin DONG. A multi-UAV cooperative air combat decision-making method based on spatial-temporal information fusion [J]. Acta Aeronautica et Astronautica Sinica, 2026, 47(6): 332633-332633.
[11]	Peizhao WANG, Ming HE, Haihua CHEN, Hongpeng WANG. Real-time task scheduling algorithm for FANET considering communication topology control [J]. Acta Aeronautica et Astronautica Sinica, 2026, 47(6): 332636-332636.
[12]	Peng GUO, Tianlai XU, Anqi LANG, Hutao CUI, Zidi LI. A complex network-based information fusion approach for cooperative navigation of multi-UAV systems [J]. Acta Aeronautica et Astronautica Sinica, 2026, 47(5): 332428-332428.
[13]	Han WU, Hao SUN, Kui LIU, Kefeng JI, Gangyao KUANG. Multi-object feature association in UAV videos: Recent progress and perspectives [J]. Acta Aeronautica et Astronautica Sinica, 2026, 47(4): 331967-331967.
[14]	Zicheng FENG, Wenlong ZHANG, Donghui LIU, Qifeng YU. Robust infrared target tracking algorithm for anti-UAV in complexbackgrounds [J]. Acta Aeronautica et Astronautica Sinica, 2026, 47(4): 332264-332264.
[15]	Xiaolu LI, Changqing WANG, Yong GUO, Aijun LI. Prescribed-time fault-tolerant control for small tandem-rotor UAV formation with input saturation [J]. Acta Aeronautica et Astronautica Sinica, 2026, 47(19): 333222-333222.