Gap interpolation algorithm

The aim of the gap interpolation algorithm is to provide a means to fill in any remaining gaps in the 3D data, once all approaches for tracking marker data through the take have been exhausted.

Gap interpolation is performed using a non-linear interpolation technique, which predicts the location of a missing marker point in a frame, based on its last known location, and also how its neighboring marker points move between the frame where the point was last present and the current frame. The final predicted point position will be (optionally) a blend of the following two results:

❙       The predicted point position calculated by predicting forwards based upon its last known location just before the gap being filled.

❙       The predicted point position calculated by predicting backwards based upon its last known location just after the gap being filled.

The gap interpolation algorithm also includes an option to use single sensor tracking gap-filling. When this option is enabled, for any cases where a 3D point is attached to a blob detection in a single camera view, the interpolation algorithm constrains its 3D result to be consistent with the attached detection, ie, the 3D interpolated point must re-project exactly onto the attached 2D blob center.