Botts' Dot Marker Detection Method Using a Cumulative Hough Transform

Lane marker detection is a key function in such intelligent transportation systems as lane departure warning (LDW), driver navigation, and autonomous vehicle systems. The study of lane marker detection using computer vision has a long history. Botts’ dot marker detection is more difficult than that for white lines because small, sparse, low contrast regions are mixed in with road-texture noise. The authors propose a Botts’ dot marker detection method in which a Botts’ dot feature filter and a cumulative Hough transform is introduced. The Botts’ dot feature filter was designed to better locate Botts’ dot candidates within a noisy road texture image, both in direct sunlight and in shadow, by calculating the brightness within a triplicate ROI (region of interest) in which Botts’ dot shapes are adjusted, and by using results of a learned brightness-characteristic model. A cumulative Hough transform was designed to robustly estimate lane boundary lines formed by Botts’ dots, using sparse and noisy feature points and extending an ordinary Hough transform for use in spatiotemporal filtering. Experimental results confirmed that the method offers robust detection. They also indicate that the cumulative Hough transform achieves good spatiotemporal interpolation of sparse image features, significantly improving robustness in the detection and tracking of lane boundaries formed by Botts’ dot markers. In particular, a false detection rate with this method was roughly 1/5 that with the reference method using an ordinary Hough transform (at a false rejection rate level of 0.2).


  • English

Media Info

  • Media Type: Digital/other
  • Features: Figures; Photos; References;
  • Pagination: 12p
  • Monograph Title: 18th ITS World Congress, Orlando, 2011. Proceedings

Subject/Index Terms

Filing Info

  • Accession Number: 01484818
  • Record Type: Publication
  • Files: TRIS
  • Created Date: Jun 25 2013 8:55AM