Abstract
In the research of crowd analysis in a multi-camera environment, the key problem is how to get target correspondence between cameras. Two main popular methods are epipolar geometric constraint and homography matrix constraint. For large view-angle and wide baseline, these two methods exist obvious disadvantages and have a low performance. The paper utilizes a new correspondence algorithm based-on the constraint of line-of-sight for the crowd positioning. Since the target area is discrete, the paper proposes to use blocking policy: dividing the target regions into blocks with certain size. The approach may provide appropriate redundancy information for each object and decrease the risk of objects missing which is caused by large view-angle and wide baseline between different perspective images. The experimental results show that the method has a high accuracy and a lower computational complexity.
Chapter PDF
Similar content being viewed by others
References
Eshel, R., Moses, Y.: Homography based multiple camera detection and tracking of people in a dense crowd. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, June 24-26, pp. 1–8 (2008)
Eshel, R., Moses, Y.: Tracking in a Dense Crowd Using Multiple Cameras. International Journal of Computer Vision 88(1), 129–143 (2010)
Ge, W., Collins, R.T.: Crowd detection with a multi-view sampler. In: Daniilidis, K., Maragos, P., Paragios, N. (eds.) ECCV 2010, Part V. LNCS, vol. 6315, pp. 324–337. Springer, Heidelberg (2010)
Castle, R.O., Murray, D.W.: Key-frame-based recognition and localization during video-rate parallel tracking and mapping. Image and Vision Computing 29(8), 524–532 (2011)
Mazzon, R., Cavallaro, A.: Multi-camera tracking using a Multi-Goal Social Force Model. Neuro-Computing 100, 41–50 (2013)
Ma, H., Zeng, C., Ling, C.: A Reliable People Counting System via Multiple Cameras. ACM Transactions on Intelligent Systems and Technology 3(2), 1–22 (2012)
Lloyd, S.A.: An optimization approach to relaxation labelling algorithms. Image and Vision Computing 1(2), 85–92 (1983)
Kittlera, J., Illingworth, J.: Relaxation labelling algorithms — A review, Image and vision computing. Image and Vision Computing 3(4), 206–210 (1985)
Rosenfeld, A., Hummel, R.A., Zucker, S.W.: Scene labeling by relaxation operations. IEEE Trans. SMC. 6(6), 420–433 (1976)
Cui, L., Dongqing, F.: Improved image segmentation algorithm based on K-means clustering. Journal of Zhengzhou University (Natural Science) 43(1), 109–113 (2011)
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2014 Springer International Publishing Switzerland
About this paper
Cite this paper
Zhu, Q., Yuan, S., Chen, B., Wang, G., Xu, J., Zhang, L. (2014). Crowd Target Positioning under Multiple Cameras Based on Block Correspondence. In: Streitz, N., Markopoulos, P. (eds) Distributed, Ambient, and Pervasive Interactions. DAPI 2014. Lecture Notes in Computer Science, vol 8530. Springer, Cham. https://doi.org/10.1007/978-3-319-07788-8_47
Download citation
DOI: https://doi.org/10.1007/978-3-319-07788-8_47
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-07787-1
Online ISBN: 978-3-319-07788-8
eBook Packages: Computer ScienceComputer Science (R0)