An improved real-time miniaturized embedded stereo vision system (MESVS-II)

In this paper we describe a fully integrated, real-time, miniaturized embedded stereo vision system (MESVS-II), which fits within 5times5cm and consumes very low power. This is a significant improvement over the original MESVS-I system in terms of performance, quality and accuracy of results. MESVS-II running at 600MHz per core, is capable of operating at up to 20 fps, which is twice as fast as MESVS-I, due to the efficient implementation of stereo-vision algorithms, improved memory and data management, in-place processing scheme, code optimization, and the pipelined-programming model that takes advantage of the dual-core architecture of the embedded processor. The firmware incorporates sub-sampling, rectification, pre-processing, matching, LRC (Left/Right Consistency) check and post-processing. As demonstrated by our experimental results, we have also enhanced the robustness of the stereo-matching engine to radiometric variations by choosing census transform over rank transform.

[1]  Peter I. Corke,et al.  Quantitative Evaluation of Matching Methods and Validity Measures for Stereo Vision , 2001, Int. J. Robotics Res..

[2]  M. Marchionni,et al.  A PC-based real-time stereo vision system , 2004 .

[3]  Ramin Zabih,et al.  Non-parametric Local Transforms for Computing Visual Correspondence , 1994, ECCV.

[4]  Heiko Hirschmüller,et al.  Evaluation of Cost Functions for Stereo Matching , 2007, 2007 IEEE Conference on Computer Vision and Pattern Recognition.

[5]  Daniel P. Huttenlocher,et al.  Efficient Belief Propagation for Early Vision , 2004, Proceedings of the 2004 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2004. CVPR 2004..

[6]  D. Scharstein,et al.  A Taxonomy and Evaluation of Dense Two-Frame Stereo Correspondence Algorithms , 2001, Proceedings IEEE Workshop on Stereo and Multi-Baseline Vision (SMBV 2001).

[7]  Olga Veksler,et al.  Fast approximate energy minimization via graph cuts , 2001, Proceedings of the Seventh IEEE International Conference on Computer Vision.

[8]  Boguslaw Cyganek.  Comparison of Nonparametric Transformations and Bit Vector Matching for Stereo Correlation , 2004, IWCIA.

[9]  Q. M. Jonathan Wu,et al.  A New Miniaturized Embedded Stereo-Vision System (MESVS-I) , 2008, 2008 Canadian Conference on Computer and Robot Vision.

[10]  Apostolos Dollas,et al.  Methodology and Experimental Setup for the Determination of System-level Dynamic Reconfiguration Overhead , 2007 .

[11]  Ruigang Yang,et al.  Improved Real-Time Stereo on Commodity Graphics Hardware , 2004, 2004 Conference on Computer Vision and Pattern Recognition Workshop.

[12]  Gooitzen S. van der Wal,et al.  The Acadia vision processor , 2000, Proceedings Fifth IEEE International Workshop on Computer Architectures for Machine Perception.

[13]  Rick Gentile,et al.  Embedded media processing , 2005 .

[14]  Miao Liao,et al.  High-Quality Real-Time Stereo Using Adaptive Cost Aggregation and Dynamic Programming , 2006, Third International Symposium on 3D Data Processing, Visualization, and Transmission (3DPVT'06).

[16]  Chris Murphy,et al.  Low-Cost Stereo Vision on an FPGA , 2007, 15th Annual IEEE Symposium on Field-Programmable Custom Computing Machines (FCCM 2007).