Asphalt Pavement Crack Detection Based on SegNet Network

A novel scheme combining image processing with deep learning is proposed to solve the problems of illumination non-uniformity and impurities in asphalt pavement images. The method first used illumination non-uniformity correction, contrast enhancement, and image denoising methods to highlight the cracks. The SegNet network is then used to achieve effective crack segmentation. Finally, through morphological methods and regional connections, interference is effectively removed, and the final crack skeleton is obtained. The results show that the Mean Intersection over union can reach 70.08%, which is about 3% points higher than that of the method without pre-processing, and achieves a good detection effect.


  • English

Media Info

  • Media Type: Web
  • Features: References;
  • Pagination: pp 1930-1942
  • Monograph Title: CICTP 2020: Transportation Evolution Impacting Future Mobility

Subject/Index Terms

Filing Info

  • Accession Number: 01768116
  • Record Type: Publication
  • ISBN: 9780784483053
  • Files: TRIS, ASCE
  • Created Date: Dec 9 2020 3:03PM