Detecting False Matches

False matches are corresponding points in the two images which are not the same point in the real object (they are not true matches).

This matches may appear if you make a mistake in the Manual Matching process (a pair of points are matched incorrectly) or can be produced by the Automatic Matching process (remember that a 100% success is not guaranteed).

False matches can spoil the resemblance of the 3D model, so it is necessary to detect and remove them.

As you can see in the figure, disparity vectors can help in the detection of false matches. The disparity vectors of a group of matches usually have the same shape, so if you detect that a disparity vector cross another vector of a near point there should be a false match in that location.