ReKlaSat 3D Spatial Reconstuction and Classification of high-resolution optic Satellite Data
Project Leader: Ronald Hochreiter
Scientific Staff: To be announced (Summer 2017)
Partners: TU Vienna, Austrian Federal Ministry of Defence and Sports, Siemens, VS Vermessung Schmid GmbH
Planning, acquisition and monitoring of infrastructure is a task in which the different requirements are substantially depending on the concerned stakeholder. Area-wise acquisition is necessary for the planning and implementation of road constructions, where in addition to the geometric position also surface condition and use of the surrounding objects are relevant. Linear structures, such as pipelines, need geometric, semantic and temporal high frequency information for the construction and operation in order to assess the correct use and environmental influences. Local 3D-information including specific use by (masses of) people is for example essential for the protection of critical infrastructure.
The Austrian armed forces have to protect Austrian embassies worldwide. In case of an evacuation 3D-information is essential and cannot be obtained by local surveying. Siemens has been active in equipment and operation of pipelines which are often situated in difficult territory. The access to orthophotos in high quantity let alone precise height information is even in European countries not guaranteed. For Vermessung Schmid planning information embedded in a digital landscape model is essential to participate in construction projects. Still, existing terrestrial and airborne acquisition methods do not meet the requirements for independent, fast, complete and cost-efficient acquisition.
High-resolution satellite data as provided by e.g. Pléiades and Worldview theoretically cover these requirements. The rapid, regular and automatic access to information is often the decisive factor. The information therefore has to go beyond a satellite image itself as the requirements request an automatic interpretation of the scene and, if necessary, also change detection. A manual analysis is not feasible for time-critical and large-scale tasks.
ReKlaSat 3D therefore aims at an innovative, automatic interpretation of individual scenes as well as spatial and temporal "Image Stacks" for planning, acquisition and operation of infrastructure. A review of geometric, semantic and temporal quality is carried out as well as an estimation of the costs for the operational implementation. The project partners bring together complementary skills such as requirements engineering and evaluation of solutions with minimum one partner having a particularly high level of competence regarding each processing step. Three different infrastructure scenarios will therefore be fully covered by the proposed processing chain.
FFG project number: 859792
Duration: 01.06.2017 - 31.05.2019