复杂场景下无水尺水位的影像水位反演智能检测方法

doi:10.11947/j.AGCS.2024.20220561

摘要/Abstract

摘要： 实现精细化水务管控和洪涝灾害预警,需要实时、准确感知水位突变事件。现有技术不能满足夜晚、雾霾、雨天、雪天、漂浮物遮挡及阴影等复杂恶劣环境下的水位识别需求。为此,本文提出一种融合改进YOLOv5与卡尔曼滤波原理的无水尺水位智能检测技术:①引入YOLOv5对水位线(水岸分界线)进行检测,并利用线性拟合方法获得实际水位线;②针对水位线在延伸方向无限大而在其法向无限小特点,提出强化中尺度特征的多层级特征融合方法改进原YOLOv5算法;③利用卡尔曼滤波引入水位历史信息作为先验知识,提高本技术对复杂恶劣环境的泛化性能;④将图像中事先标定的固定的标志物加入到深度学习网络中训练,根据标志位真实尺寸解算实际水位高程,实现无水尺检测方案。相关试验和实践表明,改进的YOLOv5更加轻量化;本文所述水位智能检测技术斜率准确性为97.3%,较原算法提高了2.4%;截距准确性为99.3%,较原算法提高了0.5%;在夜晚、雾霾、雨天、雪天、漂浮物遮挡及阴影等复杂恶劣环境下可以自动、准确识别出水位高程,误差小于0.1m。

关键词: 水位识别, 无水尺水位检测, 深度学习, YOLOv5, 卡尔曼滤波

Abstract: Realizing fine water control and flood warning requires real-time and accurate perception of sudden water level change events. The existing water level recognition technology cannot meet the needs of water level recognition in complex and harsh environments such as night, haze, rain, snow, floating object occlusion and shadow. To this end, this paper proposed an intelligent water level detection technique without water scale by integrating improved YOLOv5 and Kalman filtering principles: ① Introducing YOLOv5 to detect the water level line (water shore demarcation line) and using linear fitting methods to obtain the actual water level line; ② The water level is infinitely large in the extension direction and infinitely small in its normal direction., a multi-level feature fusion method was proposed to strengthen the mesoscale features to improve the original YOLOv5 algorithm; ③ Using Kalman filtering to introduce water level history information as a priori knowledge to improve the generalization performance of this technique to complex and harsh environments; ④ Adding a fixed marker pre-calibrated in the image to the deep learning network for training, and solving the actual water level elevation based on the real size of the marker to achieve a water-rule-free detection scheme. Relevant experiments and practice showed that the improved YOLOv5 was more lightweight; the slope accuracy of the water level intelligent detection method described in this paper was 97.3%, which was 2.4% higher than the original model; the intercept accuracy was 99.3%, which was 0.5% higher than the original model; the water level elevation could be automatically and accurately identified in complex and harsh environments such as night, haze, rain, snow, floating object occlusion, shadow, and the error was less than 0.1m.

Key words: water level identification, water level detection without water gauge, deep learning, YOLOv5, Kalman filtering

中图分类号:

TP391
P237

孙传猛, 魏宇, 李欣宇, 马铁华, 武志博. 复杂场景下无水尺水位的影像水位反演智能检测方法[J]. 测绘学报, 2024, 53(3): 558-568.

SUN Chuanmeng, WEI Yu, LI Xinyu, MA Tiehua, WU Zhibo. Intelligent detection method of image water level inversion for water level without water scale in complex scenes[J]. Acta Geodaetica et Cartographica Sinica, 2024, 53(3): 558-568.

参考文献

[1] 杜文英, 陈能成, 袁赛. 传感网支持下的洪涝事件过程监测与准实时服务方法及验证[J]. 测绘学报, 2020, 49(2):191-201.DOI:10.11947/j.AGCS.2020.20180378. DU Wenying, CHEN Nengcheng, YUAN Sai. Flood events process detection and near realtime service based on sensor web[J]. Acta Geodaetica et Cartographica Sinica, 2020, 49(2):191-201.DOI:10.11947/j.AGCS.2020.20180378.
[2] YAO Chaolong, LUO Zhicai, HU Yueming, et al. Detecting droughts in Southwest China from GPS vertical position displacements[J].Journal of Geodesy and Geoinformation Science,2020,3(2):114.
[3] YAN P, FANG Y, CHEN J, et al. Automated extraction for water bodies using new water index from Landsat 8 OLI images[J].Journal of Geodesy and Geoinformation Science,2023,6(1):59-75.
[4] ZHAO Qing, PEPE A, DEVLIN A, et al. Impact of sea-level-rise and human activities in coastal regions:an overview[J].Journal of Geodesy and Geoinformation Science,2021,4(1):124-143.
[5] IWAHASHI M, UDOMSIRI S. Water level detection from video with fir filtering[C]//Proceeding of the 16th International Conference on Computer Communications and Networks.Honolulu:IEEE, 2007:826-831.
[6] 鲍江, 陶青川, 张鹏. 基于图像处理的水位线检测算法[J]. 水电能源科学, 2015, 33(4):96-99, 210. BAO Jiang, TAO Qingchuan, ZHANG Peng. Image processing based water level detection algorithm[J]. Water Resources and Power, 2015, 33(4):96-99, 210.
[7] SUN Chuanmeng, XU Ruijia, WANG Chong, et al. Coal rock image recognition method based on improved CLBP and receptive field theory[J]. Deep Underground Science and Engineering, 2022, 1(2):165-173.
[8] 孙传猛, 王燕平, 王冲, 等. 融合改进YOLOv3与三次样条插值的煤岩界面识别方法[J]. 采矿与岩层控制工程学报, 2022, 4(1):81-90. SUN Chuanmeng, WANG Yanping, WANG Chong, et al. Coal-rock interface identification method based on improved YOLOv3 and cubic spline interpolation[J]. Journal of Mining and Strata Control Engineering, 2022, 4(1):81-90.
[9] SCHMIDHUBER J. Deep learning in neural networks:an overview[J]. Neural Networks, 2015, 61:85-117.
[10] 孙传猛,陈嘉欣,裴东兴,等.基于深度学习的恶劣战场环境图像恢复方法[J/OL].控制与决策,1-8[2024-02-23].https://doi.org/10.13195/j.kzyjc.2022.1684. SUN Chuanmeng, CHEN Jiaxin, PEI Dongxing, et al. A deep learning based approach for image recovery in harsh battlefield environments[J]. Control And Decision,1-8[2024-02-23].https://doi.org/10.13195/j.kzyjc.2022.1684.
[11] 刘媛媛, 刘洪伟, 霍风霖, 等. 基于机器学习短历时暴雨时空分布规律研究[J].水利学报, 2019, 50(6):773-779. LIU Yuanyuan, LIU Hongwei, HUO Fenglin, et al. An application of machine learning on examining spatial and temporal distribution of short duration rainstorm[J]. Journal of Hydraulic Engineering, 2019, 50(6):773-779.
[12] 鲍振鑫, 张建云, 王国庆, 等. 基于水文模型与机器学习集合模拟的水沙变异归因定量识别:以黄河中游窟野河流域为例[J]. 水科学进展, 2021, 32(4):485-496. BAO Zhenxin, ZHANG Jianyun, WANG Guoqing, et al. Quantitative assessment of the attribution of runoff and sediment changes based on hydrologic model and machine learning:a case study of the Kuye River in the Middle Yellow River Basin[J]. Advances in Water Science, 2021, 32(4):485-496.
[13] 程淑红, 赵考鹏, 张仕军, 等. 基于U-net的水位线检测[J]. 计量学报, 2019, 40(3):361-366. CHENG Shuhong, ZHAO Kaopeng, ZHANG Shijun, et al. Water level detection based on U-net[J]. Acta Metrologica Sinica, 2019, 40(3):361-366.
[14] 肖卓, 陶青川, 沈建军. 基于SSD目标检测的视频水位检测算法[J]. 现代计算机(专业版), 2019(9):60-64. XIAO Zhuo, TAO Qingchuan, SHEN Jianjun. A video water-level recognition based on SSD object detect network[J]. Modern Computer, 2019(9):60-64.
[15] 廖赟, 段清, 刘俊晖, 等. 基于深度学习的水位线检测算法[J]. 计算机应用, 2020, 40(S1):274-278. LIAO Yun, DUAN Qing, LIU Junhui, et al.Water line detection algorithm based on deep learning[J].Journal of Computer Applications, 2020, 40(S1):274-278.
[16] 王磊, 陈明恩, 孟凯凯, 等. 基于深度学习算法的水位识别方法研究[J]. 水利信息化, 2020(3):39-43, 56. WANG Lei,CHEN Mingen, MENG Kaikai, et al. Research on water level recognition method based on deep learning algorithms[J]. Water Resources Informatization, 2020,(3):39-43,56.
[17] 夏平, 王峰, 雷帮军, 等. 基于超像素和图割算法的智能视觉水位识别[J]. 计算机仿真, 2021, 38(3):430-436, 441. XIA Ping, WANG Feng, LEI Bangjun, et al. Intelligent visual water level recognition algorithm based on super-pixel and graph cut segmentation[J]. Computer Simulation, 2021, 38(3):430-436, 441.
[18] 李欣宇, 孙传猛, 魏宇, 等. 融合Transformer与残差通道注意力的恶劣场景水位智能检测方法[J]. 电子测量与仪器学报, 2023, 37(1):59-69. LI Xinyu, SUN Chuanmeng, WEI Yu,et al. Water level intelligent detection method based on fuse Transformer residual channel attention mechanism in harsh environments[J]. Journal of Electronic Measurement and Instrumentation, 2023, 37(1):59-69.
[19] 张衎, 王剑平, 张果, 等. 图像法水位检测研究进展[J]. 电子测量技术, 2021, 44(13):104-113. ZHANG Kan, WANG Jianping, ZHANG Guo,et al. Review of image water level detection[J]. Electronic Measurement Technology, 2021, 44(13):104-113.
[20] CHEN C, LIU M Y, TUZEL O, et al. RCNN for small object detection[C]//Proceedings of 2016 Asian Conference on Computer Vision. Cham:Springer, 2016:214230.
[21] 崔晓宁, 王起才, 李盛, 等. 基于YOLOv5的双块式轨枕裂缝智能识别[J]. 铁道学报, 2022, 44(4):104-111. CUI Xiaoning, WANG Qicai, LI Sheng, et al. Intelligent recognition of cracks in double block sleeper based on YOLOv5[J]. Journal of the China Railway Society, 2022, 44(4):104-111.
[22] 张娜, 戚旭磊, 包晓安, 等. 基于优化预测定位的单阶段目标检测算法[J]. 浙江大学学报(工学版), 2022, 56(4):783-794. ZHANG Na, QI Xulei, BAO Xiaoan, et al. Single-stage object detection algorithm based on optimizing position prediction[J]. Journal of Zhejiang University (Engineering Science), 2022, 56(4):783-794.
[23] HE K, ZHANG X, REN S, et al. Spatial pyramid pooling in deep convolutional networks for visual recognition[J]. IEEE Trans Pattern Anal Mach Intell, 2015, 37(9):1904-1916.
[24] LIN T Y, DOLLAR P, GIRSHICK R, et al. Feature pyramid networks for object detection[C]//Proceedings of 2017 IEEE Conference on Computer Vision and Pattern Recognition.Honolulu:IEEE, 2017.
[25] LIU Shu, QI Lu, QIN Haifang, et al. Path aggregation network for instance segmentation[C]//Proceedings of 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition.Salt Lake City:IEEE, 2018.
[26] REN Shaoqing, HE Kaiming, GIRSHICK R, et al. Faster R-CNN:towards real-time object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(6):1137-1149.
[27] LIU W, ANGUELOV D, ERHAN D, et al.SSD:single shot MultiBox detector[C]//Proceedings of 2016 European Conference on Computer Vision. Cham:Springer, 2016.
[28] REDMON J, FARHADI A. YOLOv3:an incremental improvement[C]//Proceedings of 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Salt Lake City:IEEE, 2018.
[29] DU Xuwei, CHEN Dong, LIU Huajiang, et al. Real-time hand tracking based on YOLOv4 model and Kalman filter[J]. The Journal of China Universities of Posts and Telecommunications, 2021, 28(3):86-94.
[30] HE Nanjun, FANG Leyuan, PLAZA A. Hybrid first and second order attention Unet for building segmentation in remote sensing images[J]. Science China Information Sciences, 2020, 63(4):140305.