|
基于改进长短焦图像融合技术的轨道交通障碍物检测系统研究
|
Abstract:
近年来,随着轨道交通网络的快速扩展和列车速度的不断提升,传统轨交障碍物检测系统在视觉感知能力上面临目标模糊、范围受限等挑战,难以满足日益严格的铁路运输安全需求。针对这一问题,本文提出了一种基于改进长短焦图像融合技术的轨道交通障碍物检测系统。该系统通过结合长焦相机的高分辨率细节捕捉能力与短焦相机的广角场景感知优势,生成清晰且信息丰富的融合图像用于障碍物检测。同时,设计了一种结合SIFT特征提取与相位相关法的创新配准算法,将边缘对齐误差从2.16像素显著降低至0.27像素,显著提升了铁轨区域的对齐精度和全局图像一致性。在此基础上,利用YOLO11目标检测模型的多尺度特征提取能力与高效推理性能开发障碍物检测系统。检测结果表明,相比单一长焦或短焦图像,融合图像可有效探测近距离障碍物并兼顾远距离细节,目标检测mAP50-95指标由64.19%提升至81.32%,增幅达26.69%。这展现了融合图像在远近目标识别、铁轨细节捕捉和全局视野感知方面的显著优势。本研究为轨道交通障碍物检测系统的视觉感知能力提升提供了重要的技术参考。
In recent years, with the rapid expansion of rail transit networks and the continuous increase in train speeds, traditional rail transit obstacle detection systems face challenges such as target blurriness and limited detection range, making it difficult to meet the increasingly stringent requirements of railway transportation safety. To address these issues, this paper proposes a rail transit obstacle detection system based on improved long-focal and short-focal image fusion technology. By combining the high-resolution detail capture capability of long-focal cameras with the wide-angle scene perception advantage of short-focal cameras, the system generates clear and information-rich fused images for obstacle detection. An innovative registration algorithm that integrates SIFT feature extraction and phase correlation is designed, reducing edge alignment error from 2.16 pixels to 0.27 pixels, significantly enhancing alignment accuracy in rail regions and improving overall image consistency. On this basis, the YOLO11 object detection model’s multi-scale feature extraction capability and efficient inference performance are leveraged to develop the obstacle detection system. Experimental results show that, compared to individual long-focal or short-focal images, the fused image effectively detects nearby obstacles while preserving distant details, with the mAP50-95 metric for target detection improving from 64.19% to 81.32%, an increase of 26.69%. This demonstrates the significant advantages of fused images in recognizing both near and distant targets, capturing rail track details, and perceiving the global scene. This study provides a valuable technical reference for enhancing the visual perception capabilities of rail transit obstacle detection systems.
[1] | Chellaswamy, C., Dhanalakshmi, A., Chinnammal, V. and Malarvizhi, C. (2017) An IoT-Based Frontal Collision Avoidance System for Railways. 2017 IEEE International Conference on Power, Control, Signals and Instrumentation Engineering (ICPCSI), Chennai, 21-22 September 2017, 1082-1087. https://doi.org/10.1109/icpcsi.2017.8391877 |
[2] | Bhavsar, S.S. and Kulkarni, A.N. (2016) Train Collision Avoidance System by Using RFID. 2016 International Conference on Computing, Analytics and Security Trends (CAST), Pune, 19-21 December 2016, 30-34. https://doi.org/10.1109/cast.2016.7914935 |
[3] | 姚巍巍, 田野, 李晨. 基于改进SOLO的列车主动避障视觉算法研究[J]. 电子测量技术, 2022, 45(9): 133-139. |
[4] | Sahba, F. and Sahba, R. (2018) Prevention of Metro Rail Accidents and Incidents in Stations Using RFID Technology. 2018 World Automation Congress (WAC), Stevenson, 3-6 Jun 2018, 1-5. https://doi.org/10.23919/wac.2018.8430408 |
[5] | Li, X., Li, X., Cheng, X., Wang, M. and Tan, H. (2023) MCDFD: Multifocus Image Fusion Based on Multiscale Cross-Difference and Focus Detection. IEEE Sensors Journal, 23, 30913-30926. https://doi.org/10.1109/jsen.2023.3330871 |
[6] | 唐文彦. 图像拼接关键技术研究与实现[D]: [硕士学位论文]. 上海: 上海交通大学, 2014. |
[7] | 杜俊康. 场景自适应的图像拼接[D]: [硕士学位论文]. 南京: 东南大学, 2020. |
[8] | 关腾腾, 刘兵, 曾建波, 等. 一种长短焦相机图像融合的方法[P]. 中国, CN116740522A. 2023-09-12. |
[9] | Liu, Y., Liu, S. and Wang, Z. (2015) Multi-Focus Image Fusion with Dense Sift. Information Fusion, 23, 139-155. https://doi.org/10.1016/j.inffus.2014.05.004 |
[10] | Liu, Y. and Yu, F. (2015) An Automatic Image Fusion Algorithm for Unregistered Multiply Multi-Focus Images. Optics Communications, 341, 101-113. https://doi.org/10.1016/j.optcom.2014.12.015 |
[11] | Zhang, Q. and Guo, B. (2009) Multifocus Image Fusion Using the Nonsubsampled Contourlet Transform. Signal Processing, 89, 1334-1346. https://doi.org/10.1016/j.sigpro.2009.01.012 |
[12] | Miao, Q., Shi, C., Xu, P., Yang, M. and Shi, Y. (2011) A Novel Algorithm of Image Fusion Using Shearlets. Optics Communications, 284, 1540-1547. https://doi.org/10.1016/j.optcom.2010.11.048 |
[13] | Singh, R. and Khare, A. (2014) Fusion of Multimodal Medical Images Using Daubechies Complex Wavelet Transform—A Multiresolution Approach. Information Fusion, 19, 49-60. https://doi.org/10.1016/j.inffus.2012.09.005 |
[14] | Tang, G., Wei, Z. and Zhuang, L. (2024) SAR Image Registration with Sift Features and Edge Points. IGARSS 2024-2024 IEEE International Geoscience and Remote Sensing Symposium, Athens, 7-12 July 2024, 9711-9715. https://doi.org/10.1109/igarss53475.2024.10642824 |
[15] | Song, K., Zhu, F. and Song, L. (2022) Moving Target Detection Algorithm Based on SIFT Feature Matching. 2022 International Conference on Frontiers of Artificial Intelligence and Machine Learning (FAIML), Hangzhou, 19-21 June 2022, 196-199. https://doi.org/10.1109/faiml57028.2022.00045 |
[16] | Renjith, R., Reshma, R. and Arun, K.V. (2017) Design and Implementation of Traffic Sign and Obstacle Detection in a Self-Driving Car Using SURF Detector and Brute Force Matcher. 2017 IEEE International Conference on Power, Control, Signals and Instrumentation Engineering (ICPCSI), Chennai, 21-22 September 2017, 1985-1989. https://doi.org/10.1109/icpcsi.2017.8392062 |
[17] | Luo, Y., Li, Z. and Luo, S. (2023) Research on Camera Calibration Method Based on Homography Matrix. 2023 IEEE 6th Information Technology, Networking, Electronic and Automation Control Conference (ITNEC), Chongqing, 24-26 February 2023, 464-468. https://doi.org/10.1109/itnec56291.2023.10082094 |
[18] | Gerganov, G., Papucharov, A., Kawrkow, I. and Mitev, K. (2013) Portal Image Registration Using the Phase Correlation Method. Proceedings of the 2013 IEEE Nuclear Science Symposium and Medical Imaging Conference (2013 NSS/MIC), Seoul, 26 October-2 November 2013, 1-3. |
[19] | Liu, A., Liu, Y. and Kifah, S. (2024) Deep Convolutional Neural Network for Enhancing Traffic Sign Recognition Developed on Yolo V5. 2024 International Conference on Advancements in Smart, Secure and Intelligent Computing (ASSIC), Bhubaneswar, 27-29 January 2024, 1-6. https://doi.org/10.1109/assic60049.2024.10508025 |
[20] | Her, L. and Yang, X. (2019) Research of Image Sharpness Assessment Algorithm for Autofocus. 2019 IEEE 4th International Conference on Image, Vision and Computing (ICIVC), Xiamen, 5-7 July 2019, 93-98. https://doi.org/10.1109/icivc47709.2019.8980980 |