A parallel root-finding method for omnidirectional image unwrapping

Nguan Soon Chong, M. L.Dennis Wong, Yau Hee Kho

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

The panoramic unwrapping of catadioptric omnidirectional view (COV) sensors have mostly relied on a precomputed mapping look-up table due to an expensive computational load that generally has its bottleneck occur at solving a sextic polynomial. However, this approach causes a limitation to the viewpoint dynamics as runtime modifications to the mapping values are not allowed in the implementation. In this paper, a parallel root-finding technique using Compute Unified Device Architecture (CUDA) platform is proposed. The proposed method enables on-the-fly computation of the mapping look-up table thus facilitate in a real-time viewpoint adjustable panoramic unwrapping. Experimental results showed that the proposed implementation incurred minimum computational load, and performed at 10.3 times and 2.3 times the speed of a current generation central processing unit (CPU) respectively on a single-core and multi-core environment.

Original languageEnglish
Title of host publicationIEEE VCIP 2013 - 2013 IEEE International Conference on Visual Communications and Image Processing
DOIs
Publication statusPublished - Dec 1 2013
Event2013 IEEE International Conference on Visual Communications and Image Processing, IEEE VCIP 2013 - Kuching, Sarawak, Malaysia
Duration: Nov 17 2013Nov 20 2013

Publication series

NameIEEE VCIP 2013 - 2013 IEEE International Conference on Visual Communications and Image Processing

Other

Other2013 IEEE International Conference on Visual Communications and Image Processing, IEEE VCIP 2013
CountryMalaysia
CityKuching, Sarawak
Period11/17/1311/20/13

Keywords

  • CPU
  • CUDA
  • GPU
  • Image unwrapping
  • Multi-core
  • Omnidirectional Sensor
  • Parallelisation

ASJC Scopus subject areas

  • Computer Graphics and Computer-Aided Design
  • Computer Vision and Pattern Recognition

Fingerprint Dive into the research topics of 'A parallel root-finding method for omnidirectional image unwrapping'. Together they form a unique fingerprint.

Cite this