Monitoring Landslide Phenomena along Li-Shing Estate Road in Nantou County of Central Taiwan by Applying an Object-Oriented Segmentation Approach

Li-Shing Estate Road is a mountain road that plays an important role in communication between mountain villages and allows the transport of agricultural products from mountain villages to urban areas. However, Li-Shing Estate Road is located in a fracture zone, and landslides occur regularly along the road. Identifying the possible areas of landslide occurrence is an important issue in the prevention of disasters associated with landslides. This study employs the modified multiphase segmentation method to segment the normalized difference vegetation index (NDVI) generated using infrared and red-band information provided from SPOT satellite images. Low NDVI values can be grouped together to form a series of sub-regions. These sub-regions indicate regions that are either villages or landslide areas. The proposed approach imposes edge information in the multiphase segmentation method proposed by Vese and Chan, such that a series of NDVI data over different years can be efficiently segmented. By eliminating village areas, the landslide areas can be located. The located areas are compared with suspected areas provided by Moh And Associates Group Consulting Engineers to verify the segmented results. The MAA Group carried out geological exploration and drilling along Li-Shing Estate Road, such that suspected areas where landslide phenomena could occur have been located and reported. In addition to the modified multiphase approach, another segmentation approach called statistical region merging is employed to demonstrate that the landslide phenomena can be identified by employing an objected-oriented segmentation approach.


  • English

Media Info

  • Media Type: Web
  • Features: References;
  • Pagination: pp 57-64
  • Monograph Title: Geo-China 2016: Geosynthetic Civil Infrastructure, Disaster Monitoring, and Environmental Geotechnics

Subject/Index Terms

Filing Info

  • Accession Number: 01608361
  • Record Type: Publication
  • ISBN: 9780784480045
  • Files: TRIS, ASCE
  • Created Date: Jul 21 2016 3:05PM