206 David Geronimo, Angel D. Sappa and Antonio M. Lopez´

tion of about 75 − 90% with respect to the sliding window, depending on the stride of this latter. Then, we apply the filtering stage with a cell size of σ = 0.2 and χ = 2000, reducing again a 90% the number of candidates. This represents a reduction of 97 − 99% compared to the sliding window. Figure 9 illustrates the results in six of the frames used to test the algorithm. As can be seen, the pedestrians in the scenario are correctly selected as candidates, while other freespace areas are discarded to be classified. In addition, attending to the re sults, the number of false negatives is marginal, which is a key factor for the whole sytem performance.

7.Conclusions

We have presented a three-stage candidate generation algorithm to be used in the foreground segmentation module of a PPS. The stages consist of road surface estimation, road scanning and candidate filtering. Experimental re sults demonstrate that the number of candidates to be sent to the classifier can be r e- duced by a 97 − 99% compared to the typical sliding window approach, while minimizing the number of false negatives to around 0%. Future work will be focused on the research of algorithms to fuse the cues used to select the candidates, which can potentially improve the proposed pipeline process.

Acknowledgements

The authors would like to thank Mohammad Rouhani for his ideas with the road scanning section. This work was supported by the Spanish Ministry of Education and Science under project TRA2007-62526/AUT and research programme Consolider Ingenio 2010: MIPRCV (CSD200700018); and Catalan Government under project CTP 2008 ITT 00001. David Geronimo´ was supported by Spanish Ministry of Education and Science and European Social Fund grant BES-2005-8864.

References

[1] M. Bertozzi, A. Broggi, R. Chapuis, F. Chausse, A. Fascioli, and A. Tibaldi. Shape–based pedestrian detection and localization. In Proc. of

the IEEE International Conference on Intelligent Transportation Systems pages 328–333, Shangai, China, 2003.

[2]C.-Y. Chan and F. Bu. Literature review of pedestrian detection technologies and sensor survey. Technical report, Institute of Transportation Studies, Uni. of California at Berkeley, 2005.

[3]N. Dalal and B. Triggs. Histograms of oriented gradients for human detection. In Proc. of the IEEE Conference on Computer Vision and Pattern Recognition, volume 1, pages 886–893, San Diego, CA, USA, 2005.

[4]Y. Fang, K. Yamada, Y. Ninomiya, B. Horn, and I. Masaki. A shapeindependent method for pedestrian detection with far-infrared images.

IEEE Trans. on Vehicular Technology , 53(6):1679–1697, 2004.

[5]M. Fischler and R. Bolles. Random sample consensus: A paradigm for model fitting with applications to image analysis and automated cartography. Graphics and Image Processing, 24(6):381–395, June 1981.

[6]United Nations Economic Commission for Europe. Statistics of road traffic accidents in Europe and North America, 2005.

[7]D.M. Gavrila, J. Giebel, and S. Munder. Vision–based pedestrian de tection: The PROTECTOR system. In Proc. of the IEEE Intelligent Vehicles Symposium, pages 13–18, Parma, Italy, 2004.

[8]D. Geronimo,´ A. Lopez,´ A.D. Sappa, and T. Graf. Survey of pedestrian detection for advanced driver assistance systems. In IEEE Transactions on Pattern Analysis and Machine Intelligence (in press), 2009.

[9]R. Labayrade and D. Aubert. A single framework for vehicle roll, pitch, yaw estimation and obstacles detection by stereovision. In Proc. of the IEEE Intelligent Vehicles Symposium, pages 31–36, Columbus, OH, USA, June 2003.

[10]C. Papageorgiou and T. Poggio. A trainable system for object detection.

International Journal on Computer Vision , 38(1):15–33, 2000.

[11]M. Peden, R. Scurfield, D. Sleet, D. Mohan, A.A. Hyder, E. Jara wan, and C. Mathers. World Report on road traffic injury prevention . World Health Organization, Geneva, Switzerland, 2004.

[12]P. Rousseeuw and A. Leroy. Robust Regression and Outlier Detection. John Wiley & Sons, New York, 1987.

208 David Geronimo, Angel D. Sappa and Antonio M. Lopez´

[13]A.D. Sappa, F. Dornaika, D. Ponsa, D. Geronimo,´ and A. Lopez´. An efficient approach to onboard stereo vision system pose estimation. IEEE Trans. on Intelligent Transportation Systems, 9(3):476–490, 2008.

[14]P. Viola and M. Jones. Rapid object detection using a boosted cascade of simple features. In Proc. of the IEEE Conference on Computer Vision and Pattern Recognition, pages 511–518, Kauai, HI, USA, 2001.

[15]C. Wang, H. Tanahashi, H. Hirayu, Y. Niwa, and K. Yamamoto. Comparison of local plane fitting methods for range data. In Proc. of the IEEE Conference on Computer Vision and Pattern Recognition, pages 663–669, Kauai Marriot, HW, USA, December 2001.