Stairway detection using Gabor Filter and FFPG

In this paper, we consider the problem of vision based stairway recognition. This is a fundamental task for the navigation of robot's autonomous stairway climbing, as well as stairway modeling in mapping and building reconstruction. A serious challenge of this task is the vibration of the robot and the poor illuminating conditions. Under the influence of these factors, blurred images can be produced. Therefore, a robust and efficient stairway detecting algorithm is essential. By using Gabor Filter and introducing a Fuzzy Fusion Phase-grouping (FFPG) method, the blurred stair edges in image are extracted efficiently. Then the offset parameters are estimated for the robot navigation and stairway localization. Experiment results prove the validity and effectiveness of the proposed approaches.