Description

Buildings and other infrastructure provide a limited, yet challenging, domain for 3D computer vision techniques. Terrestrial laser scanners, RGB-D cameras, mobile scanning robots, and micro air vehicles (MAVs) are just some of the platforms that are being used to capture and model the built environment. Such 3D models have enormous potential for aiding practitioners in a wide variety of fields, ranging from architecture, engineering, and construction (AEC) to post-disaster search and rescue. This workshop will explore the state of the art in techniques that use 3D imaging for modeling, analyzing, and understanding the built environment.

Call for paper

Topics of submission

Abstract: Long-range laser scanners can capture dense point-clouds from indoor and outdoor scenes. In this keynote, we discuss point processing and shape reconstruction techniques for large-scale point-clouds. We first discuss shape reconstruction of manufacturing plants using terrestrial laser scanning. 3D models of engineering plants are useful for simulating reorganization of production lines. However, it is not easy to reconstruct object shapes from huge, noisy, and incomplete point-clouds. To solve these problems, we introduce efficient surface detection and template-based shape modeling techniques. Then we discuss point processing for mobile mapping systems (MMS). A MMS can capture point-clouds of road surfaces, buildings, and roadside objects while running on the road. Since roadside objects, such as utility poles, traffic signs, streetlights, guardrails, have to be maintained periodically, their 3D models are useful for planning maintenance tasks without on-site survey. In this keynote, we introduce our point processing techniques for segmentation, mesh generation, object classification, and shape reconstruction from MMS data.

Message

Leave a message

Refresh

Contact information

  • dhuber@cs.cmu.edu
  • -

Sponsored By

  • University of Tokyo