基于声光可调谐滤波器的中波红外光谱成像探测系统可用于空中目标探测，获取红外光谱图像序列，增强抗诱饵干扰能力。但是，由于探测过程中目标特征随着观测角度变化，增加了抗干扰识别的难度。针对这一问题，提出了基于时空谱特征的智能双分支抗干扰识别模型，利用多光谱图像序列，提取时空谱特征，实现远距离探测时空中目标与干扰的有效区分。该模型依据空中目标的光谱特征随观测角度的变化规律设计，主要包括时空谱特征提取与识别模型分支和多特征融合识别模型分支。前者基于改进的ConvLSTM构建，利用红外光谱图像序列提取时空谱特征进行目标与干扰区分，用于尾后大角度观测条件下目标与干扰的光谱差异大的场景；后者基于运动、辐射、波段比等多特征提取模块与Transformer识别模块构建，融合多维度特征进行目标与干扰区分，用于尾后小角度观测条件下目标与干扰的光谱差异小的场景。利用模拟实验数据和仿真数据进行验证，在远距离小目标阶段，对于不同的观测角度下的干扰，识别准确率优于90%，可有效区分目标与干扰。
The mid-wave infrared spectral imaging detection system based on an acousto-optic tunable filter can be used for aerial target detection, acquiring infrared spectral image sequences to enhance anti-decoy interference capability. However, during the detection process, the target features vary with the observation angle, increasing the difficulty of anti-interference recognition. To address this issue, an intelligent dual-branch anti-interference recognition model based on spatial-temporal-spectral features is proposed. This model utilizes multispectral image sequences to extract spatial-temporal-spectral features, enabling effective discrimination between aerial targets and interference during long-range detection. The model is designed based on the variation pattern of spectral features of aerial targets with observation angles. It mainly consists of two branches: the spatial-temporal-spectral feature extraction and recognition model branch and the multi-feature fusion recognition model branch. The former is constructed based on an improved ConvLSTM, which extracts spatial-temporal-spectral features from infrared spectral image sequences to distinguish between targets and interference, suitable for scenarios with significant spectral differences between targets and interference under large-angle tail observation conditions. The latter is built upon multi-feature extraction modules and a Transformer-based recognition module, integrating multi-dimensional features for target-interference discrimination, suitable for scenarios with small spectral differences under small-angle tail observation conditions. Validated using simulated experimental data and synthetic data, the model achieves a recognition accuracy of over 90% for interference under different observation angles during the long-range small-target stage, effectively distinguishing between targets and interference.
[1] SUN W, YAO M J. Research on jamming strategy of surface-type infrared decoy against by infrared-guided simulation [C]//2020 20th International Conference on Control, Automation and Systems (ICCAS). Busan: IEEE, 2020.
[2] 牛得清，伍友利，徐洋，等。点源红外诱饵干扰下环境复杂度量化建模 [J]. 红外与激光工程，2020, 49 (2): 020400.
[3] 陈宁，王刚，王鹏，等。先进的机载红外诱饵对抗技术措施发展研究 [J]. 航天电子对抗，2017, 33 (5): 1-5.
[4] ZHAO H J, JI Z, ZHAO Y, et al. MWIR thermal imag-ing spectrometer based on the acousto-optic tunable fil-ter [J]. Applied Optics, 2017, 56 (26): 7269-7276.
[5] ZHAO H, LI Y, ZHANG Y, et al. Aerial target spatial-spectral discrimination for imaging spectrometer based on the acousto-optic tunable filter [J]. Optics Letters, 2022, 47 (9): 2366-2369.
[6] 赵军民，魏嘉艺，吴思捷，等。复杂战场对抗环境下空中目标抗干扰识别算法 [J]. 兵工学报，2022, 43 (10): 2576-2587.
[7] 张凯，王凯迪，杨曦，等。基于 DNET 的空中红外目标抗干扰识别算法 [J]. 航空学报，2021, 42 (S1): 128-137.
[8] LIU T, WANG Y. A Dual Stream Model for Anti-infrared Decoys: Exploiting Interaction and Contextual Relationship [J]. arXiv preprint arXiv:2203.01234, 2022.
[9] LIU X, LI X, LI L, et al. Dim and Small Target Detec-tion in Multi-Frame Sequence Using Bi-Conv-LSTM and 3D-Conv Structure [J]. IEEE Access. 2021, 9: 123456-123466.
[10] 杨开，李少毅，张凯，等。基于朴素贝叶斯分类器的空中红外目标抗干扰识别方法研究 [J]. 飞控与探测，2019, 2 (4): 1-9.
[11] 张良，田晓倩，李少毅，等。基于时空推理网络的空中红外目标抗干扰识别算法 [J]. 红外与激光工程，2022, 51 (7): 473-482.
[12] 陈海永，张岩，晏行伟。基于 IRLT-YOLO 的红外图像无人机目标实时检测研究 [J]. 红外技术，2026, 48 (1): 1-9.
[13] 马滔，熊英灼，潘庆娜，等。基于 VSA-YOLO v11n 的无人机红外小目标检测 [J]. 红外技术，2025, 47 (12): 1510-1517.
[14] XU W, DING Z, WANG Z, et al. Think Locally and Act Globally: A Frequency–Spatial Fusion Network for Infrared Small Target Detection [J]. IEEE Transactions on Geoscience and Remote Sensing. 2025, 63: 1-17.
[15] ZHANG G, XU G, CHEN S, et al. It’s Not the Target, It’s the Background: Rethinking Infrared Small-Target Detection via Deep Patch-Free Low-Rank Rep-resentations [J]. IEEE Transactions on Geoscience and Remote Sensing. 2025, 63: 1-13. doi: 10.1109/TGRS.2025.3608239.
[16] LIU X, XIANG X, TAN Y. A Global Spatial–Temporal Detection Framework for Infrared Small Targets in Complex Ground Scenes [J]. IEEE Transac-tions on Geoscience and Remote Sensing. 2025, 63: 1-14.
[17] FANG H Z, HOU Z, LIU Y, et al. Spatio-Temporal Context Learning with Temporal Difference Convolu-tion for Moving Infrared Small Target Detection [J]. arXiv preprint arXiv:2511.09352, 2025.
[18] QIN H, XU T, TANG Y, et al. OSFormer: One-Step Transformer for Infrared Video Small Object Detection [J]. IEEE Transactions on Image Processing. 2025, 34: 5725-5736.
[19] 苑智玮，黄树彩，熊志刚，等。尾焰特征光谱在主动段弹道目标识别中的应用 [J]. 光学学报，2017, 37 (2): 306-313.
[20] 郭文凤，焦志刚。基于包围盒约束光谱聚类的红外目标识别算法 [J]. 红外与激光工程，2021, 50 (11): 377-382.
[21] 卢晓，梁晓庚，贾晓洪。红外成像空空导弹智能化抗干扰研究 [J]. 红外与激光工程，2021, 50 (4): 21-29.
[22] 邵晓光。红外抗诱饵干扰技术研究 [J]. 制导与引信，2019, 40 (2): 12-16, 35.
[23] VIAU C R. Expendable Countermeasure Effectiveness against Imaging Infrared Guided Threats [R]. Otta-wa: Defence Research and Development Canada, 2005.
[24] 王怡，王浩，卫子毓，等。基于光谱辐射计的航空发动机红外辐射特性测试方法 [J]. 红外技术，2023, 45 (3): 221-226.
[25] SHI X, CHEN Z, WANG H, et al. Convolutional LSTM Network: A Machine Learning Approach for Precipitation Nowcasting [C]//Advances in Neural In-formation Processing Systems. Montreal: MIT Press, 2015: 802-810.
[26] WANG Y, LONG M, WANG J, et al. PredRNN: Recur-rent neural networks for predictive learning using spa-tio-temporal LSTMs [C]//Advances in Neural Infor-mation Processing Systems. Long Beach: NIPS, 2017.
[27] WANG Y, WU H, ZHANG J, et al. PredRNN: A Re-current Neural Network for Spatio-temporal Predictive Learning [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2023, 45 (2): 2208-2225.
[28] IOFFE S, SZEGEDY C. Batch Normalization: Accel-erating Deep Network Training by Reducing Internal Covariate Shift [C]//International Conference on Ma-chine Learning. Lille: JMLR, 2015: 448-456.
CJA