在戰場上，無人機需要清晰的影像來執行各種任務，例如投擲炸彈或識別敵方目標。當戰場環境中存在光線強度變化時，影像可能會受到影響，需要進行影像恢復以確保資訊的可用性以符合任務需求。再者，鑑於戰場環境需要實時且清晰的影像，無人機不僅需要恢復受損影像，且必須在短時間內完成這項工作，以便更快地識別目標和制定作戰策略。本文採監督式學習(Supervised Learning)解決此問題，以U-Net作為網路整體架構，並在其編碼器(Encoder)與解碼器(Decoder)中加入類感受模塊(Inception-Like Block)、雙注意力機制(Dual Attention Unit)、選擇性核特徵融合(Selective Kernel Feature Fusion)和影像降噪模塊。另外，以現有高曝光光線落差測試資料集與其他網路模型進行影像光線恢復品質比較，並同時考慮模型整體可訓練的參數量以便建構輕巧的影像曝光落差修正之模型。最後，計畫並根據所設計的類神經網路進行影像辨識和偵測來驗證所設計的類神經網路性能，藉由讓無人載具在天候環境不佳的條件下，仍具任務執行能力。

In battlefields, drones need clear images to perform various tasks, including dropping bombs or identifying enemy targets. When the battlefield environment is affected by changes in light intensity, the image may be damaged so that image restoration is required to ensure the information is available to fulfill the mission requirements. Furthermore, due to the need for real-time and clear images of the battlefield environment, the drone not only needs to recover the damaged images, but also must be able to recover them in a short period of time that allows for faster target identification and tactical planning. This paper adopts a supervised learning approach to address this problem. We use U-Net as our main architecture of the network and we add suitable modules to its encoder and decoder, such as Inception-Like Blocks, Dual Attention Units, Selective Kernel Feature Fusion, and image denoising modules. Furthermore, an existing dataset of high exposure light disparity is used to compare the image light restoration quality with other network models, as well as considering the overall trainable parameters of the model to construct a lightweight and real-time image exposure disparity correction model. Finally, the designed neural network is employed for image recognition and detection to validate its performance. With the designed network, UAVs are allowed to restore images in a short t time with the help of our high exposure difference image restoration neural network under poor weather conditions and our UAVs can detect enemy before they do and we can formulate countermeasures in advance

[1] O. Ronneberger, P. Fischer, and T. Brox, "U-net: Convolutional networks for biomedical image segmentation," in Medical Image Computing and Computer-Assisted Intervention–MICCAI 2015: 18th International Conference, Munich, Germany, October 5-9, 2015, Proceedings, Part III 18, 2015: Springer, pp. 234-241.
[2] X. Lv, S. Zhang, Q. Liu, H. Xie, B. Zhong, and H. Zhou, "BacklitNet: A dataset and network for backlit image enhancement," Computer Vision and Image Understanding, vol. 218, p. 103403, 2022.
[3] Z. Zhang, "Study on the Influence of Driving Fatigue on Drivers' Effective Reaction Time," in 2022 10th International Conference on Traffic and Logistic Engineering (ICTLE), 2022: IEEE, pp. 89-94.
[4] Q. Wang, X. Fu, X.-P. Zhang, and X. Ding, "A fusion-based method for single backlit image enhancement," in 2016 IEEE International Conference on Image Processing (ICIP), 2016: IEEE, pp. 4077-4081.
[5] Z. Li and X. Wu, "Learning-based restoration of backlit images," IEEE Transactions on Image Processing, vol. 27, no. 2, pp. 976-986, 2017.
[6] L. Zhang, L. Zhang, X. Liu, Y. Shen, S. Zhang, and S. Zhao, "Zero-shot restoration of back-lit images using deep internal learning," in Proceedings of the 27th ACM international conference on multimedia, 2019, pp. 1623-1631.
[7] A. Bochkovskiy, C.-Y. Wang, and H.-Y. M. Liao, "Yolov4: Optimal speed and accuracy of object detection," arXiv preprint arXiv:2004.10934, 2020.
[8] B. Mishra, D. Garg, P. Narang, and V. Mishra, "Drone-surveillance for search and rescue in natural disaster," Computer Communications, vol. 156, pp. 1-10, 2020.
[9] U. R. Mogili and B. Deepak, "Review on application of drone systems in precision agriculture," Procedia computer science, vol. 133, pp. 502-509, 2018.
[10] L. Angrisani et al., "An innovative air quality monitoring system based on drone and IoT enabling technologies," in 2019 IEEE international workshop on metrology for agriculture and forestry (MetroAgriFor), 2019: IEEE, pp. 207-211.
[11] J. Howard, V. Murashov, and C. M. Branche, "Unmanned aerial vehicles in construction and worker safety," American journal of industrial medicine, vol. 61, no. 1, pp. 3-10, 2018.
[12] M. A. Ma'Sum et al., "Simulation of intelligent unmanned aerial vehicle (uav) for military surveillance," in 2013 international conference on advanced computer science and information systems (ICACSIS), 2013: IEEE, pp. 161-166.
[13] T. Salimans, I. Goodfellow, W. Zaremba, V. Cheung, A. Radford, and X. Chen, "Improved techniques for training gans," Advances in neural information processing systems, vol. 29, 2016.
[14] C. Wei, W. Wang, W. Yang, and J. Liu, "Deep retinex decomposition for low-light enhancement," arXiv preprint arXiv:1808.04560, 2018.
[15] S. W. Zamir et al., "Learning enriched features for real image restoration and enhancement," in Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part XXV 16, 2020: Springer, pp. 492-511.
[16] A. Buades, J. L. Lisani, A. B. Petro, and C. Sbert, "Backlit images enhancement using global tone mappings and image fusion," IET Image Processing, vol. 14, no. 2, pp. 211-219, 2020.
[17] T. Mertens, J. Kautz, and F. Van Reeth, "Exposure fusion," in 15th Pacific Conference on Computer Graphics and Applications (PG'07), 2007: IEEE, pp. 382-390.
[18] C.-Y. Tsai and C.-L. Chen, "Attention-Gate-Based Model with Inception-like Block for Single-Image Dehazing," Applied Sciences, vol. 12, no. 13, p. 6725, 2022. [Online]. Available: https://www.mdpi.com/2076-3417/12/13/6725.
[19] V. Dumoulin and F. Visin, "A guide to convolution arithmetic for deep learning," arXiv preprint arXiv:1603.07285, 2016.
[20] A. Odena, V. Dumoulin, and C. Olah, "Deconvolution and checkerboard artifacts," Distill, vol. 1, no. 10, p. e3, 2016.
[21] S. Ai and J. Kwon, "Extreme Low-Light Image Enhancement for Surveillance Cameras Using Attention U-Net," Sensors, vol. 20, no. 2, p. 495, 2020. [Online]. Available: https://www.mdpi.com/1424-8220/20/2/495.
[22] N. Ibtehaz and M. S. Rahman, "MultiResUNet: Rethinking the U-Net architecture for multimodal biomedical image segmentation," Neural networks, vol. 121, pp. 74-87, 2020.
[23] C. Szegedy, V. Vanhoucke, S. Ioffe, J. Shlens, and Z. Wojna, "Rethinking the inception architecture for computer vision," in Proceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 2818-2826.
[24] J. Gurrola-Ramos, O. Dalmau, and T. E. Alarcón, "A residual dense u-net neural network for image denoising," IEEE Access, vol. 9, pp. 31742-31754, 2021.
[25] O. Oktay et al., "Attention u-net: Learning where to look for the pancreas," arXiv preprint arXiv:1804.03999, 2018.
[26] X. Li, W. Wang, X. Hu, and J. Yang, "Selective kernel networks," in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2019, pp. 510-519.
[27] H. Zhao, O. Gallo, I. Frosio, and J. Kautz, "Loss functions for image restoration with neural networks," IEEE Transactions on computational imaging, vol. 3, no. 1, pp. 47-57, 2016.
[28] Y. Liu and G. Zhao, "Pad-net: A perception-aided single image dehazing network," arXiv preprint arXiv:1805.03146, 2018.
[29] D. P. Kingma and J. Ba, "Adam: A method for stochastic optimization," arXiv preprint arXiv:1412.6980, 2014.
[30] A. Mittal, R. Soundararajan, and A. C. Bovik, "Making a “completely blind” image quality analyzer," IEEE Signal processing letters, vol. 20, no. 3, pp. 209-212, 2012.
[31] A. Mittal, A. K. Moorthy, and A. C. Bovik, "Blind/referenceless image spatial quality evaluator," in 2011 conference record of the forty fifth asilomar conference on signals, systems and computers (ASILOMAR), 2011: IEEE, pp. 723-727.
[32] B. Li et al., "Benchmarking single-image dehazing and beyond," IEEE Transactions on Image Processing, vol. 28, no. 1, pp. 492-505, 2018.