A Robust Width/Pitch Model for Visual Lane Detection and Classification

カテゴリ: 部門大会

論文No: MC4-2

グループ名: 【C】平成22年電気学会電子・情報・システム部門大会講演論文集

発行日: 2010/09/02

タイトル(英語): A Robust Width/Pitch Model for Visual Lane Detection and Classification

著者名: 王 臣豪(熊本大学),長岡 孝太郎(熊本大学),胡 振程(熊本大学)

著者名(英語): Chenhao Wang(Kumamoto University),kotaro nagaoka(Kumamoto University),Zhencheng Hu(Kumamoto University)

キーワード: Lane detection|PW/YL Model|Kalman filter|Front camera|Rear camera

要約(日本語): This paper presents a robust road perspective projection model by utilizing of monocular camera for road lane detection and classification. Width/Pitch model (WP model) is introduced with consideration of visual perception and road shape. And it correlates to difference results between left and right pavement markings. WP model with pair of width and pitch angle seeks the difference results of feature points. Depending on maptching probability map, probabilities higher than certain threshold are selected for candidate as road lanes. Furthermore, some special kinds of pavement markings could be classified which helps for further lane detection. Experiment results gives not only good evaluation of this approach, but also suggests utilizing rear camera for lane detection assistance.

PDFファイルサイズ: 5,716 Kバイト