Towards automatic Digital Surface Model generation using a Graphics Processing Unit

Abstract

Digital Surface Models (DSM) are widely used in the earth sciences. It provides information for various geological studies and other applications. There are a number of methods for automatic DSM generation, each of which has its own strengths and weaknesses, none of which are perfect. Even though there are plenty of algorithms to date which can generate the DSMs, it is a computationally complex calculation and does tend to take some time to complete. In order to achieve faster DSMs, an algorithm was implemented on a Graphics Processing Unit. This platform greatly improves the processing speed but does have some drawbacks; namely, no support for double precision floating point values, limited memory and algorithms has to be adapted to be usable in a stream processor environment. A speed performance increase of 900% was obtained while having a negligble decrease in accuracy. The GPU does offer a more efficient platform for computationally extensive digital surface modelling applications.