Lane detection for intelligent vehicle employing omni-directional camera

This paper describes a new robust lane detection algorithm which is applicable for outdoor environments. In order to detect lanes robustly, we use a sequence of omni-directional color images. The color feature of lanes is extracted by using RGB eigenspace method. The extracted lane color images are transformed to Hough space to detect lanes as lane candidate points. This algorithm can solve problems in lane detection often occur in outdoor events due to sunshine and/or shadow effects.