Class FeatureSelectUniformBest<Point>

All Implemented Interfaces:

public class FeatureSelectUniformBest<Point> extends Object implements FeatureSelectLimitIntensity<Point>
Attempts to select features uniformly across the image with a preference for locally more intense features. This is done by breaking the image up into a grid. Then features are added by selecting the most intense feature from each grid. If a cell has a prior feature in it then it is skipped for that iteration and the prior counter is decremented. This is repeated until the limit has been reached or there are no more features to add.
  • Field Details

    • configUniform

      public ConfigGridUniform configUniform
      Configuration for uniformly selecting a grid
  • Constructor Details

    • FeatureSelectUniformBest

      public FeatureSelectUniformBest(SampleIntensity<Point> sampler)
    • FeatureSelectUniformBest

      public FeatureSelectUniformBest()
  • Method Details

    • select

      public void select(@Nullable @Nullable GrayF32 intensity, int width, int height, boolean positive, @Nullable @Nullable FastAccess<Point> prior, FastAccess<Point> detected, int limit, FastArray<Point> selected)
      Description copied from interface: FeatureSelectLimitIntensity

      Selects features using a rule given the limit on detection objects. If the limit is higher than the number of detected features and prior is null then the detected features should be copied into selected. How prior features are used is dependent upon the implementation and their affect isn't specified in general.

      NOTE: The reason a null input image is supported is for the situation where the intensity image is no longer available and intensity has been copied into points. See SIFT

      Specified by:
      select in interface FeatureSelectLimitIntensity<Point>
      intensity - (Input) Intensity image. If null that means the point will provide intensity information. Nullable.
      width - (Input) if image is null then this must be the image's width
      height - (Input) if image is null then this must be the image's height
      positive - (Input) true if better features have more positive values, false if it's more negative values
      prior - (Input) Locations of previously detected features. Nullable
      detected - (Input) Locations of newly detected features
      limit - (Input) The maximum number of new features detected
      selected - (Output) Selected features. Element count not exceed the limit. Reset on every call.
    • setSampler

      public void setSampler(SampleIntensity<Point> sampler)
      Description copied from interface: FeatureSelectLimitIntensity
      Specifies how the intensity will be sampled.
      Specified by:
      setSampler in interface FeatureSelectLimitIntensity<Point>
    • getGridCell

      protected FeatureSelectUniformBest.Info<Point> getGridCell(Point p)
      Returns the grid cell that contains the point