Implementation of Improved Census Transform Stereo Matching on a Multicore Processor
Traditionally, sub-pixel interpolation in stereo-vision systems has been used for the block-matching algorithm. In this paper, Census transform algorithm which has been on area-based matching algorithm is improved and it’s compared with existing census transform algorithm. Two algorithms are compared using Tsukuba stereo images provided by Middlebury web site. As a result, disparity map error rate is decreased from 16.3 to 11.8 %.
KeywordsCensus transform Stereo matching Area-based matching Multi-core processing
This work was sponsored by Industrial Strategic Technology Development Program funded by the Ministry of Knowledge Economy (10039188, SoC platform development for smart vehicle info-tainment system) and Industrial Strategic Technology Development Program funded by the Ministry of Knowledge Economy (10039145, the development of system semiconductor technology for IT fusion revolution).
- 1.Humenberger M, Zinner C, Kubinger W (2009) Performance evaluation of a census-based stereo matching algorithm on embedded and multi-core hardware. In: Proceedings of the international symposium on image and signal processing and analysis, vol 6Google Scholar
- 2.Zabih R, Woodfill J (1994) Non-parametric local transforms for computing visual correspondence. In: Proceedings of the European conference on computer vision, pp 151–158 Google Scholar
- 3.Frὃba B, Ernst A (2004) Face detection with the modified census transform. In: Proceeding of IEEE Conference on automatic face and gesture recognition, pp 91–96Google Scholar
- 5.Schastein D, Szeliski R (2002) Ataxonomy and evaluation of dense two-frame stereo correspondence algorithm. Int J Comput Vis 47(1):7–42Google Scholar