by Liu, J, Chen, H, Xu, Y, Rong, W, Vaudrey, T and Klette, R
Abstract:
Real-time implementations of stereo algorithms are required for the latest driver assistance and robotic applications. Current speeds of dense stereo algorithms are no where near the required 0.1 seconds per image pair (on a convensional CPU) required for a “real-time” frame rate. This paper describes an efficient parallel implementation of dynamic programming and belief propagation algorithms on a cell processor that speeds up stereo image analysis. Dynamic programming processes image data by scanline optimization; thus it is easily implemented on a cell processor. Belief propagation differs from dynamic programming by having potentially the whole image area as an area of influence for every pixel; this potentially global optimization scheme produces improved results, but requires more running time than the dynamic programming method. Furthermore, we define limitations of the Cell architecture for these applications. For evaluation, we use synthetic and real-world image sequences. Real-world images are typically degraded by various types of noise, changes in lighting, differing exposures, and so on. Sobel edge and residual images can improve the stereo matching results compared to the use of original real-world images; our results show that a cell processor also reduces running time for these processes.
Reference:
Disparity map computation on a cell processor (Liu, J, Chen, H, Xu, Y, Rong, W, Vaudrey, T and Klette, R), In Proceedings of the IASTED International Conference on Modelling, Simulation, and Identification, MSI 2009, 2009.
Bibtex Entry:
@inproceedings{liu2009disparityprocessor,
author = "Liu, J and Chen, H and Xu, Y and Rong, W and Vaudrey, T and Klette, R",
booktitle = "Proceedings of the IASTED International Conference on Modelling, Simulation, and Identification, MSI 2009",
title = "Disparity map computation on a cell processor",
year = "2009",
abstract = "Real-time implementations of stereo algorithms are required for the latest driver assistance and robotic applications. Current speeds of dense stereo algorithms are no where near the required 0.1 seconds per image pair (on a convensional CPU) required for a "real-time" frame rate. This paper describes an efficient parallel implementation of dynamic programming and belief propagation algorithms on a cell processor that speeds up stereo image analysis. Dynamic programming processes image data by scanline optimization; thus it is easily implemented on a cell processor. Belief propagation differs from dynamic programming by having potentially the whole image area as an area of influence for every pixel; this potentially global optimization scheme produces improved results, but requires more running time than the dynamic programming method. Furthermore, we define limitations of the Cell architecture for these applications. For evaluation, we use synthetic and real-world image sequences. Real-world images are typically degraded by various types of noise, changes in lighting, differing exposures, and so on. Sobel edge and residual images can improve the stereo matching results compared to the use of original real-world images; our results show that a cell processor also reduces running time for these processes.",
isbn = "9780889868106",
keyword = "Cell processor",
keyword = "Computer vision",
keyword = "Disparity map",
keyword = "Driver assistance system",
keyword = "Parallel computation",
keyword = "Residual image",
language = "eng",
}