Pedestrian-Flow Statistics Algorithm Based on Image Processing

Abstract:

The automatic statistics to pedestrian flow can greatly improve the duty staff efficiency and thus becomes the hot research field. This paper presents a pedestrian flow statistics algorithm based on image-processing. Our research is on the basis of color histogram and mean-shift algorithm, and introduces the SVM model through generating confidence function to judge the actual attribute of background pixels. Such way improves the accuracy of target detection. Experimental results prove our algorithm better handles to pedestrian's coverage problem and meets the requirement of real-time and accuracy.

• Periodical: Applied Mechanics and Materials (Volumes 347-350)
• Pages: 3381-3384
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.347-350.3381
• Online since: August 2013
• Authors: Hong Mei Jian, Zheng Xi Wei
• Keywords: Color Histogram, Computer Version, Pedestrian Detection, SVM Model

References

[1] J. Hou and Y. Chen: Television Technology. Vol. 2(2009), p.63.

[2] S. Yaser and S. Mubarak: IEEE Transactions on pattern analysis and machine intelligence. Vol. 11 (2005), p.1778.

[3] Jeffrey Ng and S. G. Gong: Image and Vision Computing. Vol. 5 (2002), p.359.

[4] P. H. Li: Signal Processing: Image Communication. Vol. 8(2006), p.676.

[5] Y. Z. Wang and J. Yang: Pattern Recognition. Vol. 1(2006), p.386.