Changes between Version 53 and Version 54 of GSoC/2014


Ignore:
Timestamp:
02/13/14 23:58:29 (11 years ago)
Author:
mlennert
Comment:

--

Legend:

Unmodified
Added
Removed
Modified
  • GSoC/2014

    v53 v54  
    107107=== Extracting the medial axis of polygons ===
    108108
    109 The medial axis, sometimes also called topological skeleton, is a line version of a polygon where all points on the lines are equidistant to the polygon's boundaries. The medial axis can be used to extract the center line of a polygon, useful e.g. to convert a river represented as a polygon to a line. While the theory for extracting a medial axis is well known, actual implementations often use an approximation instead of the exact medial axis. Extracting the exact medial axis poses problems with regard to floating point accuracy around concave vertices where the medial axis has the form of a parabola. A collection of algorithms exist to define parabola equations around concave vertices, intersect parabolas, and approximate parabolas with straight line segments. A working implementation is not yet available for GRASS.
     109The medial axis, sometimes also called topological skeleton, is a line version of a polygon where all points on the lines are equidistant to the polygon's boundaries. The medial axis can be used to extract the center line of a polygon, useful e.g. to convert a river represented as a polygon to a line, but also for characterising a polygon's shape, e.g. for object-based classification of images. While the theory for extracting a medial axis is well known, actual implementations often use an approximation instead of the exact medial axis. Extracting the exact medial axis poses problems with regard to floating point accuracy around concave vertices where the medial axis has the form of a parabola. A collection of algorithms exist to define parabola equations around concave vertices, intersect parabolas, and approximate parabolas with straight line segments. A working implementation is not yet available for GRASS.
    110110
    111111Mentor: Markus Metz