Class AssociateGreedyBase<D>

java.lang.Object
boofcv.alg.feature.associate.AssociateGreedyBase<D>
Type Parameters:
D - Feature description type.
Direct Known Subclasses:
AssociateGreedyBase2D, AssociateGreedyDescBase

public abstract class AssociateGreedyBase<D> extends Object

Performs association by greedily assigning matches to the src list from the dst list if they minimize a score function. Optional additional checks can be done. Backwards Validation sees of the dst feature would also select the src feature as its best match. Ratio Test sees if the second best match has a similar score to the first. If it does it's likely that the best match is ambiguous. Matches can also be rejected if they exceed a maximum fit score limit. In practice, forward-backwards and ratio test pruning are very effective.

Internally it compare a score matrix while performing the greedy src to dst assignments. This score matrix is then used to quickly perform a look up when doing forwards-backwards validation.

  • Method Details

    • setupForAssociate

      protected void setupForAssociate(int sizeSrc, int sizeDst)
      Clears and allocates memory before association starts.
      Parameters:
      sizeSrc - size of src list
      sizeDst - size of dst list
    • forwardsBackwards

      public final void forwardsBackwards(int indexSrc, int sizeSrc, int sizeDst)
      Uses score matrix to validate the assignment of src feature `indexSrc`
      Parameters:
      indexSrc - Index of source feature being validated
      sizeSrc - size of src list
      sizeDst - size of dst list
    • setMaxFitError

      public void setMaxFitError(double maxFitError)