Classification Moving Vehicle Based on Multisensor Data Using Fusion of Multi-Class SVMs Methods

Abstract:

In this paper, we propose a special Multi-class SVMs (MSVM) data fusion strategy which is applied to classify vehicle based on multiple pavement structural strain time histories. The centralized and distributed fusion strategies are applied to combine information from several data sources. In the centralized strategy, all information from several data sources is centralized and combined to construct an input space. Then a MSVM classifier is trained. In distributed schemes, the individual data sources are processed separately and modeled by using the MSVM. Then new data fusion strategies are used to combine the information from the individual MSVM to acquire the final classification outputs. Two popular Multi-class SVMs algorithms (One-against-all OAA, One-against-one OAO) are used to construct classifier based on aforementioned two fusion strategies, respectively. The results are compared between SVM-based fusion approach and single data source SVM using two MSVM algorithms, respectively. The result shows this SVM-based fusion approach significantly improves the results of classification accuracy and robustness. The proposed Multisensor data fusion methods can also be applied in other fields.

• Periodical: Advanced Materials Research (Volumes 945-949)
• Pages: 1978-1981
• DOI: https://doi.org/10.4028/www.scientific.net/AMR.945-949.1978
• Online since: June 2014
• Authors: Xiao Yun Chen*, Xian Fu Chen, Shao Quan Zhang, Wen Bin Zhang
• Keywords: Multisensor Data Fusion, Multi-Source Data Classification, Vehicle Classification

References

[1] Sun, C. Ritchie, S. G., Oh., S. 2003 Inductive classifying artificial network for vehicle type categorization. Comput. Aided Civ. Infrastruct. Eng., 18 (2003), p.161–172.

DOI: 10.1111/1467-8667.00307

[2] Stefano Messelodi; Carla Maria Modena; Gianni Cattoni. Vision-based bicycle/motorcycle classification. Pattern Recognition Letters, 28 (2007), pp.1719-1726.

DOI: 10.1016/j.patrec.2007.04.014

[3] Goyal, A.; Verma, B. A Neural Network based Approach for the Vehicle Classification. Computational Intelligence in Image and Signal Processing, 2007, pp.226-231.

DOI: 10.1109/ciisp.2007.369173

[4] J. A. Benediktsson and I. Kanellopoulos. Classification of multisource and hyperspectral data based on decision fusion. IEEE Trans. Geosci. Remote Sens. 37 (1999), p.1367–1377.

DOI: 10.1109/36.763301

[5] M. Fauvel, J. Chanussot, and J. A. Benediktsson. Decision fusion for the classification of urban remote sensing images, " IEEE Trans. Geosci. Remote Sens. 44 (2006), p.2828–2838.

DOI: 10.1109/tgrs.2006.876708

[6] B. Schölkopf and A. Smola. Learning With Kernels. MIT Press: Cambridge, MA, (2002).