Leonard, Ryan E听1听;听Heckman, Christoffer听2听;听Minear, J. Toby听3听;听Anderson, Robert S听4
1听乐播传媒 at Boulder, Department of Computer Science听
2听乐播传媒 at Boulder, Department of Computer Science
3听乐播传媒, Boulder and Cooperative Institute for Research in Environmental Sciences
4听乐播传媒 at Boulder, Department of Geological Sciences and INSTAAR
Hydrologic surveys often consist of key measurements relating to the discharge, stage, and velocity of a river. Unfortunately, traditional measurement tools face several major problems, including slow measurement times, prohibitive cost, and single point measurements. In addition, several of these measurements require personnel to enter potentially hazardous situations to make contact with the water. It is during flood flows that measurements are of the greatest importance to the measurement record. However, it is during these periods of high flow that the limitations imposed by traditional methods are particularly restrictive.
We are testing a prototype low-cost computer-vision based surveying approach using a stereo camera configuration. While our approach has yet to be verified against traditional methods, preliminary results suggest this setup is capable of near real-time measurements of a river, including water level and surface velocity. Our approach also iterates upon LSPIV (Large Scale Particle Imaging Velocimetry) by allowing the user to take contactless surface velocimetry measurements from perspectives oblique to the river鈥檚 surface.
Because our method allows the user to take non-contact measurements, it eliminates many of the hazards faced by surveyors and equipment associated with traditional methods. Our configuration is highly portable, requires only a single operator, and is orders of magnitude less expensive than existing hardware. It is our hope that the use of the use of computer vision systems will lower the barrier to entry for surveying in remote regions and ultimately increase the volume of data available to hydrologists.