Fast Surface Reconstruction and Segmentation with Terrestrial LiDAR Range Data

Abstract

Recent advances in range measurement devices have opened up new opportunities and challenges for fast 3D modeling of large scale outdoor environments. Applications of such technologies include virtual walk through, urban planning, disaster management, object recognition training, and simulations. In this thesis, we present general methods for surface reconstruction and segmentation of 3D colored point clouds, which are composed of partially ordered terrestrial range data. Our algorithms can be applied to a large class of LiDAR data acquisition systems, where terrestrial data is obtained as a series of scan lines. We develop an efficient and scalable algorithm that simultaneously reconstructs surfaces and segments the data. For surface reconstruction, we introduce a technique for setting local, data-dependent distance thresholds, and we present post-processing methods that fill holes and remove redundant surfaces in the generated meshes. We demonstrate the effectiveness of our results on datasets obtained by two different terrestrial acquisition systems. The first dataset contains 94 million points obtained by a vehicle-borne acquisition system during a 20 km drive. The second dataset contains 17 million points obtained by a stationary LiDAR sensor in a stop-and-go fashion over a 0.2 km2 area.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
May 18, 2009
Accession Number
ADA538884

Entities

People

  • Matthew A. Carlberg

Organizations

  • University of California, Berkeley

Tags

Communities of Interest

  • Sensors

DTIC Thesaurus Topics

  • Acquisition
  • Algorithms
  • Boundaries
  • Computational Complexity
  • Computer Science
  • Computer Vision
  • Data Acquisition
  • Data Sets
  • Electrical Engineering
  • Engineering
  • Identification
  • Image Processing
  • Image Recognition
  • Lists (Data Structures)
  • Object Recognition
  • Point Clouds
  • Recognition

Readers

  • Image Processing and Computer Vision.
  • Neural Network Machine Learning.
  • Robotics and Automation.