作者：赵晨1,2, 谢谟文1,2, 黄正均1,2, 吴志祥1,2
作者单位：1. 北京科技大学土木与资源工程学院，北京 100083;
2. 北京科技大学 城市地下空间工程北京市重点实验室，北京 100083
关键词：水位识别;图像处理;模板匹配;二值化分割;透视变换
摘要：
针对库水位传统量测方法中水尺易锈蚀倾斜，精度低且成本高的问题，该文提出一种不依赖水尺，基于图像匹配的库水位变动识别方法。首先对监控相机拍摄的库坝上游面光学图像进行畸变消除和透视变换，消除相机误差。进一步选取包含水面的感兴趣区域(region of interest, ROI)图像进行自适应二值化分割、形态学处理等前处理操作，将水面和库坝特征分离，突出水位线位置。最后，利用归一化互相关匹配算法（normalised cross correlation, NCC）对水位变动前后的两幅图像进行匹配计算，识别水位线变动距离。通过室内试验与现场测试验证上述方法实用性。结果表明：基于图像匹配的库水位动态识别方法可准确识别水位变动，相对误差约为5%，此算法鲁棒性较高。该研究可为库水位自动化、低成本监测提供一种新思路。
Research on reservoir water level change identification based on image matching
ZHAO Chen1,2, XIE Mowen1,2, HUANG Zhengjun1,2, WU Zhixiang1,2
1. School of Civil and Environmental Engineering, University of Science & Technology Beijing, Beijing 100083, China;
2. Key Laboratory of Urban Underground Space Engineering, University of Science & Technology Beijing, Beijing 100083, China
Abstract: The traditional measurement methods of reservoir water level are limited by corrosion and tilting, low frequency and high cost. To address such problems, this paper proposes a method to identify the changes of reservoir water level based on image matching without relying on the water level ruler. Firstly, the optical image of the upstream surface of the dam taken by the monitoring camera is distortion-removed and perspective-transformed to eliminate camera errors. Subsequently, the image of the region of interest (ROI) containing the water surface is selected for pre-processing operations such as adaptive binarization and morphological processing to separate the water surface and reservoir dam features and highlight the water level line location. Finally, the normalised cross correlation (NCC) matching algorithm is used to match the two images before and after the water level change to identify the distance of water level change. The practicality of the above proposed method is verified through indoor experiments and field tests. The results show that the dynamic identification method based on image matching can accurately identify water level changes with a relative error of about 5%, and the algorithm has high robustness. This study can provide a new method for automated and low-cost monitoring of reservoir water levels.
Keywords: water level recognition;image processing;template matching;binarization;perspective transformation
2024, 50(7):10-16 收稿日期: 2022-09-08;收到修改稿日期: 2023-01-16
基金项目: 国家重点研发计划（2019YFC1509602）；国家自然科学基金（41572274）
作者简介: 赵晨（1998-），男，河北保定市人，硕士研究生，专业方向为工程安全监测。
参考文献
[1] 李蔚. 基于立体视觉与LSPIV的河流水动力过程近距遥感测量系统[D]. 杭州: 浙江大学, 2016.
LI W. A close range remote sensing measurement system for river hydrodynamic processes based on stereoscopic vision and LSPIV [D]. Hangzhou: Zhejiang University, 2016.
[2] 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 Y, WANG J P, ZHANG G, et al. Research progress on image based water level detection[J]. Electronic Measurement Technology, 2021, 44(13): 104-113.
[3] 王旺, 刘源, 胡鹤鸣, 等. 基于超声传播时间的水位精测技术[J]. 中国测试, 2022, 48(4): 35-41.
WANG W, LIU Y, HU H M, et al. Research on precision measurement method of water level based on ultrasonic propagation time[J]. China Measurement & Test, 2022, 48(4): 35-41.
[4] 廖爱民, 刘九夫, 张建云, 等. 实验小流域尺度下高精度水位计的比测分析[J]. 水科学进展, 2019, 30(3): 337-347.
LIAO A M, LIU J F, ZHANG J Y, et al. Comparative analysis of high-precision water level gauges at the scale of experimental small watersheds[J]. Advances in Water Science, 2019, 30(3): 337-347.
[5] ZHANG Z, ZHOU Y, LIU H, et al. In-situ water level measurement using NIR-imaging video camera[J]. Flow Measurement and Instrumentation, 2019, 67: 95-106.
[6] 刘铭辉, 车国霖, 张衎, 等. 一种不定长水尺图像水位测量方法[J]. 仪器仪表学报, 2021, 42(7): 250-258.
LIU M H, CHE G L, ZHANG Y, et al. A water level measurement method using an indefinite length water gauge image[J]. Chinese Journal of Scientific Instrument, 2021, 42(7): 250-258.
[7] LIN F, CHANG W, LEE L, et al. Applications of image recognition for real-time water level and surface velocity, 2013[C]. IEEE, 2013.
[8] MA J, JIANG X, FAN A, et al. Image matching from handcrafted to deep features: A survey[J]. International Journal of Computer Vision, 2021, 129(1): 23-79.
[9] 刘卫南, 谢谟文. 基于点云密度特征的滑坡位移监测方法[J]. 岩土力学, 2020, 41(11): 3748-3756.
LIU W N, XIE M W. A landslide displacement monitoring method based on point cloud density characteristics[J]. Rock and Soil Mechanics, 2020, 41(11): 3748-3756.
[10] RAN Q, LI W, LIAO Q, et al. Application of an automated LSPIV system in a mountainous stream for continuous flood flow measurements[J]. Hydrological Processes, 2016, 30(17): 3014-3029.
[11] 李子印, 许斌, 陈立峰. 基于自适应步长选择的NCC图像匹配算法[J]. 光电工程, 2013, 40(10): 77-84.
LI Z Y, XU B, CHEN L F. NCC image matching algorithm based on adaptive step size selection[J]. Opto-Electronic Engineering, 2013, 40(10): 77-84.
[12] 王谭, 王磊磊, 张卫国, 等. 基于张正友标定法的红外靶标系统[J]. 光学精密工程, 2019, 27(8): 1828-1835.
WANG T, WANG L L, ZHANG W G, et al. Infrared target system based on Zhang Zhengyou calibration method[J]. Optics and Precision Engineering, 2019, 27(8): 1828-1835.
[13] 杨通钰, 彭国华. 基于NCC的图像匹配快速算法[J]. 现代电子技术, 2010, 33(22): 107-109.
YANG T Y, PENG G H. A fast image matching algorithm based on NCC[J]. Modern Electronic Technology, 2010, 33(22): 107-109.
[14] 张伟, 葛耀君, 杨詠昕. 粒子图像测速技术互相关算法研究进展[J]. 力学进展, 2007(3): 443-452.
ZHANG W, GE Y J, YANG Y X. Research progress on cross-correlation algorithms in particle image velocimetry technology[J]. Advances in Mechanics, 2007(3): 443-452.
