Single-Shot Accurate 3D Reconstruction Using Structured Light Systems Based on Local Optimization

  • Author / Creator
    Aslsabbaghpourhokmabadi, Neda
  • Creating 3D models of objects has been studied extensively for several decades. These models have different applications in different fields. As a result, creating 3D models is an interesting area in computer vision. There are many proposed methods for extracting 3D information of 2D images. One of the most common methods for 3D reconstruction is structured light methods. Although structured light methods can get valuable results of 3D reconstruction, they have limitations. For example, the structured light methods can get dense results on static scenes or get sparse results on dynamic scenes. In static scenes, the structured light method projects several patterns, and it results in dense models. However, in dynamic scenes, the structured light method projects just one pattern since the object is moving, and it results in sparse models. The limitation of the structured light methods in dynamic scenes is the most important motivation for this thesis. In this thesis, a single-shot structured light method is developed to overcome the sparse results in dynamic scenes. In particular, the proposed method can obtain more accurate reconstruction with just one image of dynamic scenes than that of existing methods. The new method applies global and local optimizations to establish dense correspondences. The result of simulated experiments comparison with the ground truth demonstrates that the proposed method in this thesis achieves more accurate results than that of previous methods. Lastly, the technique developed in this thesis is applied to real data in order to obtain high quality 3D reconstruction results. The results of the new method are more accurate compared to previous methods.

  • Subjects / Keywords
  • Graduation date
  • Type of Item
  • Degree
    Master of Science
  • DOI
  • License
    This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.
  • Language
  • Institution
    University of Alberta
  • Degree level
  • Department
    • Department of Computing Science
  • Supervisor / co-supervisor and their department(s)
    • Herbert Yang (Computing Science)
  • Examining committee members and their departments
    • Pierre Boulanger (Computing Science)
    • Eleni Stroulia (Computing Science)