Hierarchical Optical Flow Estimation Architecture Using Color Cues

This work presents an FPGA implementation of a highly parallel architecture for color motion estimation. It implements the wellknown Lucas & Kanade algorithm with a multiscale extension for an accurate computation. Our system fulfills real-time requirements estimating 32 frames per second with 512×512 resolution. It presents our architecture based on fine pipelines and the benchmark of the alternatives analyzing the accuracy and the hardware resource requirements.

[1]  Polina Golland,et al.  Motion from Color , 1997, Comput. Vis. Image Underst..

[2]  Antonio Cañas,et al.  Hardware description of multi-layer perceptrons with different abstraction levels , 2006, Microprocess. Microsystems.

[3]  Javier Díaz,et al.  Superpipelined high-performance optical-flow computation architecture , 2008, Comput. Vis. Image Underst..

[4]  Vishal Markandey,et al.  Multispectral constraints for optical flow computation , 1990, [1990] Proceedings Third International Conference on Computer Vision.

[5]  David J. Fleet,et al.  Performance of optical flow techniques , 1994, International Journal of Computer Vision.

[6]  David J. Fleet,et al.  Performance of optical flow techniques , 1992, Proceedings 1992 IEEE Computer Society Conference on Computer Vision and Pattern Recognition.

[7]  D. B. Davis,et al.  Intel Corp. , 1993 .

[8]  Naoya Ohta,et al.  Optical flow detection by color images , 1990 .

[9]  Javier Díaz,et al.  Multi-port abstraction layer for FPGA intensive memory exploitation applications , 2010, J. Syst. Archit..

[10]  Steven S. Beauchemin,et al.  The computation of optical flow , 1995, CSUR.

[11]  Brian C. Lovell,et al.  Color Optical Flow , 2003 .

[12]  Reinhard Klette,et al.  Quantitative color optical flow , 2002, Object recognition supported by user interaction for service robots.

[13]  Edward H. Adelson,et al.  The Laplacian Pyramid as a Compact Image Code , 1983, IEEE Trans. Commun..