Abstract
Based on Global Positioning System (GPS), Inertial Measurement Unit (IMU) and altitude calculation to obtain the attitude and position information of Unmanned Aerial Vehicle (UAV) for image alignment, this method significantly improves the stitching speed and stitching quality compared with the traditional image matching. However, as UAVs can take thousands of images, the images can be blurred and distorted, making manual screening time consuming and difficult to ensure that the right reference image is selected. In order to be able to select the right reference image for the final stitching quality. This paper proposed a Deep Q-Network (DQN)-based image stitching algorithm for UAVs (IMDQN). The UAV captured images were preprocessed and fed into DQN, and the globally optimal reference image was decided by the action selection model, Root Mean Square Error (RMSE) metrics were also used as an incentive mechanism. The action selection model was used to select the optimal reference image to integrate the input to Position and Orientation System (POS) data for image alignment operation, while the final stitching was accomplished using the weighted average fusion algorithm for picture fusion. The study’s findings demonstrate the algorithm’s clear autonomy and usefulness above the conventional approach, particularly given the enhanced image stitching quality. The study’s findings aid in the creation of later picture stitching methods.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Liu, C., Zhang, S., Akbar, A.: Ground feature oriented path planning for unmanned aerial vehicle mapping. IEEE J. Sel. Top. Appl. Earth Obser. Remote Sens. 12(4), 1175–1187 (2019)
Cui, H., Li, Y., Zhang, K.: A fast UAV aerial image mosaic method based on improved KAZE. In: 2019 Chinese Automation Congress, pp. 2427–2432 (2019)
Yuanting, X., Yi, L., Kun, Y., Chun Xue, S.: Research on image mosaic of low altitude UAV based on Harris corner detection. In: 2019 14th IEEE International Conference on Electronic Measurement & Instruments, pp. 639–645, IEEE, Changsha (2019)
Fangjie, C., Jun, H., Zuwu, W., Guoqiang, Z., Jianlian, C.: Image registration algorithm based on improved GMS and weighted projection transform. Laser & Optoelectronics Progress 55(11), 180–186 (2018)
Zhenyu, L., Yuan, T., Fang-jie, C., Jun, H.: Uav aerial image Mosaic algorithm based on improved ORB and PROSAC. Adv. Laser Optoelectronics 56(23), 91–99 (2019)
Lan, X., Guo, B., Huang, Z., Zhang, S.: An improved UAV aerial image mosaic algorithm based on GMS-PROSAC. In: 2020 IEEE 5th International Conference on Signal and Image Processing, pp.148–152. IEEE, Nanjing (2020)
Liu, Y.M., Wang, H.Y., Sun, Y., Gao, X.: Farmland Aerial Images Fast-Stitching Method and Application Based on Improved SIFT Algorithm. IEEE Access 10, 95411–95424 (2022)
Leng, J., Wang, S.: UAV remote sensing image mosaic technology combined with improved SPHP algorithm. In: 2020 IEEE International Conference on Mechatronics and Automation, pp. 1155–1160. IEEE, Beijing (2020)
Xiangdong, F., Chunying, Wei.: UAV Aerial image stitching method based on adaptive optimization of coverage area. In: Radio Engineering, pp.1–8 (2022)
Kim, J., Kim, T. Shin, D.: Fast and robust geometric correction for mosaicking UAV images with narrow overlaps. Remote Sens, pp. 2557–2576 (2017)
Wu, X.: Research on UAV image positioning optimization technology. PLA Information Engineering University (2017)
Yu, G., Ha, C., Shi, C., Gong, L., Yu, L.: A fast and robust UAV images mosaic method. In: Wang, L., Wu, Y., Gong, J. (eds.) Proceedings of the 7th China High Resolution Earth Observation Conference (CHREOC 2020). CHREOC 2020. LNEE, vol. 757, pp. 229–245. Springer, Singapore (2022). https://doi.org/10.1007/978-981-16-5735-1_17
Ren, M., Li, J., Song, L., Li, H., Xu, T.: MLP-Based Efficient Stitching Method for UAV Images. IEEE Geosci. Remote Sens. Lett. 19, 1–5 (2022)
Quan, L., et al.: A survey on deep reinforcement learning. J. Comput. Sci. 1–28 (2017)
Morales, E.F., Murrieta-Cid, R., Becerra, I., et al.: A survey on deep learning and deep reinforcement learning in robotics with a tutorial on deep reinforcement learning. Intel. Serv. Robotics. 14, 773–805 (2021)
Le, N., Rathour, V.S., Yamazaki, K., et al.: Deep reinforcement learning in computer vision: a comprehensive survey. Artif. Intell. Rev. 55, 2733–2819 (2022)
Chen, S.A., Tangkaratt, V., Lin, H.T., et al.: Active deep Q-learning with demonstration. Mach. Learn. 109, 1699–1725 (2020)
Mishra, S., Arora, A.: A Huber reward function-driven deep reinforcement learning solution for cart-pole balancing problem. Neural Comput. Appl. 35, 16705–16722 (2022)
Bellver, M., Giroinieto, X., Marques, F.et al.: Hierarchical object detection with deep reinforcement learning. ZarXivpreprintarXiv:1611, p. 03718 (2016)
Hongya, T., Yun-cai, L.: A two-step rectification algorithm for airborne linear images with POS data. Zheijang Univ. 6, 492–496 (2005)
Pei, H-X., Liu, J-D., Ge, J-L., et al.: A review of image stitching techniques. J. Zhengzhou Univ. (Science Edition) 51(4), 1–10 (2019)
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2023 The Author(s), under exclusive license to Springer Nature Switzerland AG
About this paper
Cite this paper
Ma, J., Liu, W., Chen, T. (2023). DQN-Based Stitching Algorithm for Unmanned Aerial Vehicle Images. In: Yang, X., et al. Advanced Data Mining and Applications. ADMA 2023. Lecture Notes in Computer Science(), vol 14177. Springer, Cham. https://doi.org/10.1007/978-3-031-46664-9_9
Download citation
DOI: https://doi.org/10.1007/978-3-031-46664-9_9
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-031-46663-2
Online ISBN: 978-3-031-46664-9
eBook Packages: Computer ScienceComputer Science (R0)