Color Image Processing for Navigation: Two Road Trackers

D. Aubert and Chuck Thorpe
tech. report CMU-RI-TR-90-09, Robotics Institute, Carnegie Mellon University, April, 1990


Download
  • Adobe portable document format (pdf) (595KB)
Copyright notice: This material is presented to ensure timely dissemination of scholarly and technical work. Copyright and all rights therein are retained by authors or by other copyright holders. All persons copying this information are expected to adhere to the terms and constraints invoked by each author's copyright. These works may not be reposted without the explicit permission of the copyright holder.

Abstract
To build an autonomous vehicle capable of road following, it is necessary to compute the position and the orientation of the vehicle relative to the road. One way of doing so is to acquire images from a camera on the vehicle and extract some road features from these images. Knowing the geometric transformation between the camera and the ground plane, it is easy to compute the position and the orientation of the vehicle relative to these road features. In our case, we are building a system to track "structured roads", roads that have lane markings, shoulders, and other structure. For these roads, the easiest visual features to track are white lines and the yellow lines. To deal with a variety of roads, light and weather conditions, we need to build robust extractors of these features. In the case of the white stripe, we use our knowledge about its shape to extract it. We create a mask with a shape similar to the current white stripe and convolve a search area in the image with it. The maximum correlation gives us the location of the stripe. To reduce the signaVnoise ratio, and to be robust in the cases of light and weather variations, we use a large mask. By taking into account the current running total during the convolution phases, we reduce the computational time drastically and so get a fast operator to uack the white stripe. In the case of the yellow stripes, the color is a smng caractenstic of this kind of stripe, and the hue is only slightly affected by lighting and weather variations (we can get in the image a dark or a light yellow, but it is always a yellow color). We build the yellow stripe tracker based on the hue infonnation.

Notes
Grant ID: #F33615-86-C-5-38
Associated Center(s) / Consortia: Vision and Autonomous Systems Center
Associated Lab(s) / Group(s): NavLab

Text Reference
D. Aubert and Chuck Thorpe, "Color Image Processing for Navigation: Two Road Trackers," tech. report CMU-RI-TR-90-09, Robotics Institute, Carnegie Mellon University, April, 1990

BibTeX Reference
@techreport{Thorpe_1990_221,
   author = "D. Aubert and Chuck Thorpe",
   title = "Color Image Processing for Navigation: Two Road Trackers",
   booktitle = "",
   institution = "Robotics Institute",
   month = "April",
   year = "1990",
   number= "CMU-RI-TR-90-09",
   address= "Pittsburgh, PA",
}