Generalize entities

Edit elements

Generalize a drawing document's graphical entities by removing points that contribute no relevant geometric information at a given scale.

The dialog requests:

  • Scale: denominator used to determine generalization thresholds. A point important at 1:500 may be irrelevant at 1:1000. Larger denominators remove more points and produce coarser entities. Change thresholds in the dialog opened with Advance Settings....
  • Linear generalization: filters points contributing no information to an entity's geometry.
  • Surface generalization: filters closed entities below a specified area threshold.
  • Delete peaks: removes spikes caused by a displaced point in a line, resembling a triangle with a very acute angle.

Contours before generalization

After generalization

  • 3 dimensions: considers point elevations, retaining points whose elevations matter even if their plan positions do not.
  • Entities from LiDAR data: enables different considerations for files generated from LiDAR, such as extracted boundaries. Requires:
    • Linear tolerance: LiDAR measurement tolerance.
    • Angular tolerance: angle tolerance between segments for rejecting spikes.
  • Typology: entity origin, affecting processing decisions:
    • General: unspecified general entities.
    • Buildings: building boundaries, where artificial construction commonly produces alignment, perpendicularity and parallelism relationships to detect between segments.
    • Treat entities as contours: uses additional algorithms to join nearby surfaces or generate new surfaces where the original entity narrows. Can be applied to any entity set representing surfaces.

Contours before generalization

Contour split after generalization

Contours before generalization

Contours joined after generalization

  • Algorithm: generalization methods:
    • Douglas-Peucker: widely used for vector generalization. Optional angle checking retains significant points.
    • One point in every n: random; removes one point for each specified number of points.
    • Lang
    • Wang-Muller
    • Reumann-Witkam