A Feature Selection Algorithm Based on Boosting for Road Detection

Feature selection is very important for road detection. Generally, optimal feature set is very hard to be determined manually by prior-knowledge. In this paper, a feature selection algorithm based on boosting is proposed. To fully utilize potential feature correlations, the features are combined. The feature vector is enlarged by the combined features, and the new feature vector is called raw feature vector. In this paper, the classify power of each feature is evaluated by the error rate and converge speed of boosting classifier which is based on single feature. After that, the features are selected according to itpsilas classify power. The selected features are reassembled to B-feature vector. Then features are weighted according to its power in classification. The weighted B-feature vector is called B-W-Feature Vector. Three classifiers are used to evaluate the raw feature vector, the B-Feature and the B-W-Feature. The experiment results show selected and weighted feature vector can improve the classification performance.

[1]  Paul A. Viola,et al.  Robust Real-time Object Detection , 2001 .

[2]  Ran El-Yaniv,et al.  Distributional Word Clusters vs. Words for Text Categorization , 2003, J. Mach. Learn. Res..

[3]  Paul A. Viola,et al.  Robust Real-Time Face Detection , 2001, International Journal of Computer Vision.

[4]  Gunnar Rätsch,et al.  An Introduction to Boosting and Leveraging , 2002, Machine Learning Summer School.

[5]  Leo Breiman,et al.  Random Forests , 2001, Machine Learning.

[6]  M. Foedisch,et al.  Adaptive real-time road detection using neural networks , 2004, Proceedings. The 7th International IEEE Conference on Intelligent Transportation Systems (IEEE Cat. No.04TH8749).

[7]  Craig Schlenoff,et al.  Towards an approach for knowledge-based road detection , 2005, KRAS '05.

[8]  Lei Zhang,et al.  Feature Selection on Combinations for Efficient Learning from Images , 2004 .

[9]  Bernhard Schölkopf,et al.  Use of the Zero-Norm with Linear Models and Kernel Methods , 2003, J. Mach. Learn. Res..

[10]  Massimo Bertozzi,et al.  Vision-based intelligent vehicles: State of the art and perspectives , 2000, Robotics Auton. Syst..

[11]  Edoardo Amaldi,et al.  On the Approximability of Minimizing Nonzero Variables or Unsatisfied Relations in Linear Systems , 1998, Theor. Comput. Sci..

[12]  Nizar Bouguila,et al.  A Graphical Model for Content Based Image Suggestion and Feature Selection , 2007, PKDD.