Smart Surface Reconstruction

June 1, 2026Open Access

View Full Paper

Ask AI

Bookmark

Share

Key Points

This research aims to address the challenges of reconstructing accurate 3D objects from faulty point clouds, focusing on enhancing reconstruction quality.
Developed Points2Surf model combining local and global priors for generalization across object classes.
Utilized Point-Convolutions and Attention in the PPSurf model for improved reconstruction quality.
Created LidarScout for real-time processing and visualization of large aerial LIDAR scans.
Achieved high-quality, generalizable reconstructions by balancing local details with global context.
Demonstrated the importance of model design and efficient representations in handling noise and missing data.
Provided practical solutions applicable for reconstructing both individual objects and entire landscapes.

The dissertation introduces new methods for transforming noisy point clouds into accurate 3D surfaces, enhancing geometrical applications.

No takes yet. Share an insight, caveat, or question.

Philipp Erler (2026) studied this question.

View Full Paper

Ask AI

Bookmark

Share