Application of computer vision to earth science problems: An initial application using 3D scene reconstruction and image velocimetry to estimate surface water velocities in rivers

Velocity is one of the more difficult measurements to collect in the Earth Sciences, particularly for processes that are fast-moving or dangerous, such as debris flows, snow avalanches, rock falls, and steep streams. Dr. J. Toby Minear of the CIRES Earth Science and Observation Center, along with Assistant Professor Christoffer Heckman (Autonomous Robotics and Perception Group, Department of Engineering) and Professor Robert Anderson (Department of Geological Sciences and INSTAAR), are teaming up to develop a new measurement technique. The team will use robotics-based computer vision algorithms and hardware to develop an inexpensive stereoscopic non-contact system to measure surface elevations and velocities that cannot be measured with traditional techniques. This project has the potential to greatly improve the ability to measure and monitor difficult and dangerous natural hazard phenomena. Initial testing and development of the system will investigate river surface water elevations and velocities on Boulder Creek and the South Platte River, near Denver, Colorado.