3D Plume
Reconstructing Volcanic Plumes using Multi-Camera Computer Vision Techniques
Reconstructing Volcanic Plumes using Multi-Camera Computer Vision TechniquesThis study presents a method and a proof of principle system for the direct measurement of volcanic ash plume 3D spatial properties. The shape of a plume is reconstructed in three dimensions using multi-view imagery collected from static ground-based cameras. The method was developed using data collected during an expedition to Volcan de Fuego in Guatemala, where four thermal cameras were deployed to capture simultaneous images of the regular ash-rich eruptions. A space carving method has been applied to the problem to estimate the volume of the plume at any moment in time. By successively applying the method to sequential sets of images, other quantitative measurements such as the drift direction, ascent rate, and dispersion rate can be deduced. The complete method work-flow is presented including data capture, calibration processes, image processing, the space-carving method, and practical implementation issues. The method is sensitive to the camera alignment, hence a novel technique for estimating the camera orientation angles, making use if a high-accuracy terrain model, is described. Other sources of error relating to the number, synchronisation and resolution of the cameras are also discussed. Preliminary results are presented using data collected at Volcan de Fuego in November 2017 over a period of 1.25h including three distinct eruptions.
|
Collaborators
Andrew Calway Helen Thomas Mathew Watson Lucy Berthoud Thomas Richardson |