A novel image-based water level measuring instrument for hydrological and limnological studies
Participants: Troy Gilmore, François Birgand (Bio&Ag), Ken Chapman and Andrew Brown (GaugeCam)
Timeline: Started 2009
Project Description
This project is made in collaboration with the GaugeCam company. Taking and sending images has become one of the simplest things in our digital and Internet connected world. Soon, live images will be expected to be available from just about anywhere in the world. This wealth of data ought to be exploited. This may come with the automatic extraction of objective information from the images. This project is an application of the machine vision techniques to the field of hydrology.
The GaugeCam system automatically recognizes water level in streams and rivers from images taken by rugged on-site webcams. Day- and night-time images
of the water level are captured at custom intervals and sent via solar powered wireless technology to a server. The server analyzes the images and calculates
water height and flow rates on a real-time basis. The technique involves calibrating the pixel coordinates to real world coordinates and find the water level
in the pixel coordinates. The server then stores the water level information and the images in a database. This database is accessible anywhere in the world
using a web-based data management tool for automatic accumulation of time lapse videos, graphical data displays, etc.
The initial cost is comparable to existing technologies, but this technique requires untrained maintenance (cleaning the lenses and the targeted background). The images are physically saved on a memory card on the camera for backup. Should the algorithm not manage to measure a water level, the data can be retrieved and reanalyzed by a human interpretation of the images. At the time when aging instrumentation will have to be replaced in many hydrological stations, this new technique has the potential to become a new standard.
links
This page was last updated on January 27, 2011.
