helicopterlandmarkdetection.PDFVIP

256 Ren et al. / J Zhejiang Univ Sci A 2008 9(2):256-261 Journal of Zhejiang University SCIENCE A ISSN 1673-565X (Print); ISSN 1862-1775 (Online) /jzus; E-mail: jzus@ Boosting multi-features with prior knowledge for mini unmanned helicopter landmark detection* Qin-yuan REN?, Ping LI, Bo HAN (The State Key Laboratory of Industrial Control Technology, Zhejiang University, Hangzhou 310027, China) ?E-mail: qyren@ Received Mar. 14, 2007; revision accepted July 5, 2007; published online Jan. 11, 2008 Abstract: Without sufficient real training data, the data driven classification algorithms based on boosting method cannot solely be utilized to applications such as the mini unmanned helicopter landmark image detection. In this paper, we propose an approach which uses a boosting algorithm with the prior knowledge for the mini unmanned helicopter landmark image detection. The stage forward stagewise additive model of boosting is analyzed, and the approach how to combine it with the prior knowledge model is presented. The approach is then applied to landmark image detection, where the multi-features are boosted to solve a series of problems, such as rotation, noises affected, etc. Results of real flight experiments demonstrate that for small training examples the boosted learning system using prior knowledge is dramatically better than the one driven by data only. Key words: Boosting, Prior knowledge model, Landmark detection doi:10.1631/jzus.A071131 Document code: A CLC number: TP75 INTRODUCTION sented a real-time computer vision system based on the circumrotate-patt