The fusion of both LIDAR and hyperspectral imagery for the creation of realistic urban models for simulation purposes was the focus of a presentation this morning at ILMF by Raul Campos-Marquetti, senior hyperspectral scientist at Merrick & Company. The ultimate purpose of this model was a simulation for military training purposes by the U.S. Army’s RDECOM.
Hyperspectal provided the means to classify features and to create a spectral library of road surface types, roof types, vegetation classifications, and an understanding of different building types. The hyperspectral classifications were then used to do a pixel by pixel, point to point fusion to create an informed 3D model with real world features and land cover/land use classifications.
The “physical morphology” model informed the synthetic creation of building exteriors and interiors based on real observation of different material types. The resulting large-scale city model was more of a true modeled reality than what can be accomplished with simply point clouds, because the classifications informed more realistic simulation that could take into account the physics of the different material types.
