Journal of Algorithms & Computational Technology (Mar 2008)

An Improved Morphology Connective Criterion Based on Complete Lattice for Lane Detection

  • Hui Cai,
  • Jinhui Cai,
  • Guangxin Zhang,
  • Zekui Zhou

DOI
https://doi.org/10.1260/174830108784300394
Journal volume & issue
Vol. 2

Abstract

Read online

In order to make lane detection robust to noise in complex circumstance, a novel mathematical morphology connective criterion based on Complete Lattice was proposed. Firstly, histogram of connective element was calculated based on Jump-connective criterion. Secondly, the valley of the image histogram of connective element was computed as a minimal threshold. Only above the threshold, the smooth-connective criterion was used for binary segmentation. Finally, binary morphological erosion and dilation operator were employed to solve background disturb problem. It was demonstrated by series of experiments that the proposed approach is effective in suppressing noise, and the recognition rate is about 99.0∼99.6% even when the road appearance is disturbed by shadows, stain and varying illumination.