Read out

ReKlaSat 3D Spatial Reconstruction and Classification of high-resolution optic Satellite Data

Pro­ject Staff

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.

Fun­ded by:

FFG (Aus­trian Re­search Pro­mo­tion Agency)

FFG pro­ject num­ber: 859792

Dur­a­tion: 01.06.2017 - 31.05.2019