3D Object Detection and Modeling from Images and Lidar

a. Satellite images

• 3D reconstruction from satellite images
• Extracting and reconstructing building models automatically with an end-to-end workflow

Figure. From satellite images and image-derived point clouds, our approach segments and models buildings (grey), bridges (cyan), and terrain (green) before texturing from the imagery.

b. Airborne Lidar

• Generating 3D models of buildings from airborne lidar data

Figure. (a) color coded planar roof points and (b) boundary model extracted from airborne Lidar.

c. Ground Doppler Lidar

• Building a model-free detection-based tracking approach to detect and track moving objects in Doppler LiDAR scans

Figure. Detection results on static (sequences (a,b)) and mobile datasets (sequences (c,d)).

d. Smartphone Images

• Point clouds of pavement from smartphone images
• Mean Profile Depth (MPD) calculation from point clouds

Figure. 4 examples of 3D point clouds rendered over images respectively from the INDOT test site (upper left), SR 32 (upper right), SR 446 (lower left), and SR 205 (lower right).