An Improved Algorithm for Real-Time Moving Target Detection Based on Gaussian Mixture Model

Abstract:

Gaussian Mixture Model is a popular method to detect moving targets for static cameras. Since the traditional Gaussian Mixture Model has a poor adaptability when the illumination is changing in the scene and has passive learning rate, this paper describes a method that can detect illumination variation and update the learning rate adaptively. It proposes an approach which uses the color histogram matching algorithm and adjusts the learning rate automatically after introducing illumination variation factor and model parameters. Furthermore, the proposed method can select the number of describing model component adaptively, so this method reduced the computation complexity and improved the real-time performance. The experiment results indicate that the detection system gets better robustness, adaptability and stability.

• Periodical: Applied Mechanics and Materials (Volumes 599-601)
• Pages: 814-818
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.599-601.814
• Online since: August 2014
• Authors: Xue Yuan Chen*, Xia Fu Lv, Jie Liu
• Keywords: Background Subtraction, Gaussian Mixture Model, illumination variation, Moving Target Detection

References

[1] Ruben H E, Michael P, Thomas S. Splitting Gaussian in Mixture Models. Pro of IEEE International Conference on Advanced Video and Signal Based Surveillance. Berlin, Germany, 2012: 300-305.

[2] Wenzhe Zhao, Shiyin Qin. Comparative Study on Detection Methods for Video Motion Targets, J. Science and Technology Review. 2009, 27(10): 64-70.

[3] Stauffer C, Grimson W E L. Adaptive background mixture models for real-time tracking. Pro of IEEE Computer Society Conference on Computer Vision and Pattern Recognition. MA, USA, 1999, 2(6), 23-25.

DOI: 10.1109/cvpr.1999.784637

[4] Kaewtrakulpong P, Bowden R. An improved adaptive background mixture model for real-time tracking with shadow detection. Pro of the Second European Workshop on Advanced Video Based Surveillance Systems. Kingston, UK, 2001: 149-153.

DOI: 10.1007/978-1-4615-0913-4_11

[5] Lee D S. Effective Gaussian mixture learning for video background subtraction. Pro of IE-EE Tran on Pattern Analysis and Machine Intelligence. CA, USA, 2005, 27(5): 827-832.

DOI: 10.1109/tpami.2005.102

[6] Bouttefroy P L M, Bouzerdoum A, Phung S L, et al. On the Analysis of Background Subtraction Techniques Using Gaussian Mixture Models. Proc of IEEE International Conference on Acoustics Speech and Signal Processing. Dallas, USA, 2010: 4042-4045.

DOI: 10.1109/icassp.2010.5495760

[7] Gorur P, Amrutur B. Speed up Gaussian Mixture Model algorithm for background subtraction. Proc of IEEE International Conference on Advanced Video and Signal Based Surveillance. Bangalore, India, 2011: 386-391.

DOI: 10.1109/avss.2011.6027356

[8] Mingzhi Li, Zhiqiang Ma, Yong Shan. Adaptive background update based on Gaussian mixture model under complex condition, J. Journal of Computer Applications, 2011, 31(7): 1831-(1935).

[9] Zivkovic Z, Improved adaptive Gaussian mixture model for background subtraction. Proc of International Conference on Pattern Recognition. Amsterdam Univ., Netherlands, 2004: 28-31.

DOI: 10.1109/icpr.2004.1333992

[10] WallflowerDataset[EB/OL]. http: /research. microsoft. com/en-us/um/people/jckrumm/WallFlo-wer/TestImage. htm.

[11] ChangeDetectionDataset[EB/OL]. http: /www. changedetection. net.

[12] ATONDataset[EB/OL]. http: /cvrr. ucsd. edu/aton/testbed.