Road network extraction specifications

From OTBWiki
Jump to: navigation, search

This application allows a user to extract road networks from an image. Simplified use case:

  1. Open image
  2. Select a reference pixel for a type of road
  3. Detect roads
  4. Add more reference pixels for other types of roads
  5. Keep or erase previous detections
  6. For the future:
    1. Concept of network (T junctions, X junctions, etc.)
    2. Multiresolution processing for extraction roads with different widths
    3. Manual edition of the nodes of the extracted polylines

The following lists the existing and needed items for this application.

  1. Open an image [Ok]
  2. Choose the input data for the extraction algorithm in one of:
    1. Image intensity [Ok]
    2. Image channel. In this case, select a channel [Ok]
    3. Spectral angle. In this case, select a reference pixel [Ok]
    4. Water index
  3. Tune some parameters: thresholds, resolution, alpha
  4. Run extraction on the full resolution area [Ok]
  5. If area change, run the extraction again and keep previous results
  6. Clear previous results [Ok]
  7. Display thick roads, with thickness related to the alpha and resolution parameter
  8. Display roads in differents color according to the input data they were extracted from (red for spectral angle, blue for water index ...)
  9. Save the results in vector data format [Ok]
  10. Import vector data