Resolving position after homography projection

The videos from three caeras, the foreground regions and the feet positions are extracted. The feet positions are correctly calculated despite the ambiguities of the projection.