Accurate DTM generation in sand beaches using mobile mapping

A Mobile Mapping System (MMS) is composed by a set of sensors placed on a mobile platform whose main objective is to obtain attitude and position over time of a referential in the platform together with the image of the objects surrounding the platform. The acquired data allows to obtain georeferenced positions of object points captured by the object sensors. This paper describes a methodology to automatically extract accurate Digital Terrain Models (DTM) of sand beach areas using a Terrestrial Mobile Mapping System (TMMS). The main goal of the presented TMMS is to obtain accurate 3D models of sand beaches. A very important consequence of this will be the abbility to track spatial and temporal changes in coastal geomorphology. The presented methodology has several advantages over more classical approaches: the overall cost, flexibility for regular surveys or at the most convenient epoch, such as after storms. Besides one of the major constraints in Mobile Mapping, that is poor GNSS (Global Navigation Satellite Systems) observation conditions, is usually not present in these areas. The TMMS uses two progressive colour video cameras and can work with any type of direct georeferencing system, which obvoiusly has impact in the final accuracy of the derived DTM. For the results herein reported, a dual frequency GNSS receiver and a low grade type of IMU (Inertial Measurement System) were used. The moving platform where the system was installed is a moto quad. The developed methodology for the DTM generation relies on an automatic choice of terrain points whose three-dimensional coordinates are calculated using a presented image matching technique. A cloud of points obtained in a nearly regular grid pattern is the base for the DTM. The proposed methodology was applied on a sector of the Vila Nova de Gaia Coast, in northern Portugal, and the results presented here show that it is possible with this methodology to produce DTMs with a precision of a few centimetres.