MRI/RUI: Acquisition of Range-Scanning Equipment and of Data Servers for the Reconstruction of Large-Scale Scenes from 3D Range and 2D Color Data.

Cuny Hunter College, New York NY

Investigators

Abstract

EIA-0215962 Ioannis Stamos Charles Ehlschlaeger CUNY Hunter College MRI/RUI: Acquisition of Range-Scanning Equipment and of Data Servers for the Reconstruction of Large-Scale Scenes from 3D Range and 2D Color Data This proposal, addressing the problem of 3D-photorealistic reconstruction of large-scale scenes in a tightly integrated system that features novel registration, segmentation, modeling, and visualization algorithms, tackles the following issues: 1) Nonplanar segmentation, 2) Automatic registration between 3D and 2D data sets, 3) Integration of the data sets, and 4) Simplification of the complex geometry. The project requests the acquisition of a highly accurate range-scanner that can provide dense 3D data of outdoor scenes. Proposed is a system that operates on an input sequence of calibrated and uncalibrated 3D range and 2D image data. The system will segment the range and color images, extract their features, establish topological relationships between features, and derive global topological relationships between features in the final 3D geometric model. The whole process will be fully automated except for the range-to-image registration process that requires minimal user interaction. The expected output consists of a complete photorealistic model of the scene. The resulting 3D models will be utilized in geospatial; the acquired data will also be used for teaching purposes in CS and Geography departments. Aided by the Center of the Analysis and Research of Spatial Information (CARSI), urban areas of downtown New York and the interior of Grand Central Station in Manhattan are being used as testbeds.