# A Rapid Deployment Method for Real-Time Water Surface Elevation Measurement

## TL;DR

This paper presents a new non-contact method for measuring water levels using stereo vision cameras, which can work efficiently in hard-to-reach areas.

## Contribution

The novel approach uses stereo vision and point cloud processing without requiring physical water references.

## Key findings

- A ZED 2i binocular camera was used to measure water surface elevation with less than 2 cm error within a 5 m range.
- The method uses RANSAC to calculate the horizontal plane equation for accurate elevation estimation.
- The system is non-contact and suitable for remote or inaccessible locations.

## Abstract

In this research, I introduce a water surface elevation measurement method that combines point cloud processing techniques and stereo vision cameras. While current vision-based water level measurement techniques focus on laboratory measurements or are based on auxiliary devices such as water rulers, I investigated the feasibility of measuring elevation based on images of the water surface. This research implements a monitoring system on-site, comprising a ZED 2i binocular camera (Stereolabs, San Francisco, CA, USA). First, the uncertainty of the camera is evaluated in a real measurement scenario. Then, the water surface images captured by the binocular camera are stereo matched to obtain parallax maps. Subsequently, the results of the binocular camera calibration are utilized to obtain the 3D point cloud coordinate values of the water surface image. Finally, the horizontal plane equation is solved by the RANSAC algorithm to finalize the height of the camera on the water surface. This approach is particularly significant as it offers a non-contact, shore-based solution that eliminates the need for physical water references, thereby enhancing the adaptability and efficiency of water level monitoring in challenging environments, such as remote or inaccessible areas. Within a measured elevation of 5 m, the water level measurement error is less than 2 cm.