Extraction and Reconstruction of Zebra Crossings from High Resolution Aerial Images

Zebra Crossings have played an important role in public traffic safety, so the reconstruction of zebra crossings is very helpful for reducing the occurrence of traffic accidents. An automatic approach using high-resolution aerial images for zebra crossing extraction and reconstruction is proposed in this paper. In the approach, zebra crossings are extracted by JointBoost classifier based on GLCM (Gray Level Co-occurrence Matrix) features and 2D Gabor Features. A geometric parameter model based on spatial repeatability relationships is globally fitted to reconstruct the geometric shapes of zebra crossings. Representative experiments under interfered conditions such as zebra crossings covered by pedestrians, shadows and color fading were conducted to verify the validity of the proposed method in both the extraction and reconstruction of zebra crossings.