基于图像处理的小型水库水位测量方法  

Water Level Measurement Method of Small Reservoir Based on Image Processing

作　　者：聂鼎[1] 张龑[1] 刘毅[1] 商峰[1] 黄涛[1] NIE Ding;ZHANG Yan;LIU Yi;SHANG Feng;HUANG Tao(Research Center for Sustainable Hydropower,China Institute of Water Resources and Hydropower Research,Beijing 100038,China)

机构地区：[1]中国水利水电科学研究院水电可持续发展研究中心

出　　处：《水电能源科学》2019年第11期28-32,共5页Water Resources and Power

基　　金：国家重点研发计划项目(2018YFC0406905);中国水利水电科学院科研专项(SS0145C192017,SS0145B612017,SS0145B392016);国家自然科学基金项目(51779277)

摘　　要：针对目前水位测量方法中传统水尺需人工读数,雷达、超声波等水位自动化测量仪器需持续投入较高维护费用的问题,提出了一种基于图像处理技术的水位检测方法。首先人工采集并处理布设有标识物的大坝上游面图像,得到标识物在图像中的坐标位置及空间几何尺寸等参数;然后基于透视原理对坝坡图像进行透视变换,采用图像增强、带标记的分水岭分割等图像处理方法,自动检测识别大坝上游面水位线;最后根据几何相似原理推导出水位计算公式。试验结果表明,与人工检测的水位线结果对比,采用图像处理方法得到的水位线误差较小,可为多数小型水库水位的自动化测量提供一种新手段,在水库日常安全管理中具有应用推广价值。In view that among the current water level measurement method,the traditional water gauge requires manual measurement,and the radar,ultrasonic and other water level automatic measuring instruments need to continuously invest with higher maintenance costs,a water level detection method based image processing technology was proposed.Firstly,the upstream image of the dam with the predefined marker was manually collected and processed,and the coordinates of the marker in the image and its geometrical dimensions were obtained.Then based on the perspective principle,the dam slope image was transformed into perspective,and image processing methods such as image enhancement and marked watershed segmentation were used to automatically detect and identify the water level line on the upstream surface of the dam.Finally,the water level calculation formula was derived based on the geometric similarity principle.Compared with the manually detected water level line results,the error obtained by image processing method was small.A new means for automatic measurement of water level of most small reservoirs is provided and it has application promotion value in daily safety management of reservoirs.

关 键 词：水位测量 图像处理 分水岭分割 透视变换 

分 类 号：TV123[水利工程—水文学及水资源] TV39