Reversal+of+geometrical+distortion+through+block-matching+against+an+existing+video+reference

It would appear that if it were possible to exactly track the original TV scan lines on the film recording then it may be possible to reconstruct a viable PAL signal. Such a signal could then be PAL decoded to provide a solution to colour recovery.
 * Recreation of line-structure**

In the preparation of the PAL signal, one would also need to compensate for the stretching/squeezing of the original signal introduced by the horizontal element of geometric distortion inherent in the film recording process, and introduce new colour bursts at the appropriate positions within the reconstructed PAL signal. Trial-and-error could then be used to find the correct positioning of the recovered PAL signal within the PAL 8-field sequence.

Scanning of the film in high-definition would be necessary in order to capture sufficient detail from the chroma dots.


 * Reversal of geometric distortion through block-matching against an existing video reference**

One idea muted for PAL signal reconstruction would be to employ block-matching or phase correlation techniques on an HD scanned film recording in conjunction with a poor quality video reference of the same material. If the HD scan of the film was synchronised with the reference video then specially written software could be used to analyse each frame of material and create a spatial mapping between positions of objects on the film with those of the poor quality video. The spatial mapping between the two sources would make it possible to ‘DVE’ each area of a film recording frame back to the correct positions using a technique such as [|mesh-warping]. An exact match would be required in order to recover the exact line positionings for satisfactory PAL decoding.

A new spatial mapping for each frame would probably need to be created – as opposed to one ‘generic mapping’ for use throughout the programme. This would be necessary in order to counter the effects of film hop and weave, and the ‘EHT breathing’ of the CRT in the original film-telerecorder.

A side-effect of this process would also be the recovery of temporal information. The two video fields (which are ‘locked together’ in the film frame) could be separated and the ‘video-look’ restored to the picture.

Some experimental work on this idea has already been conducted and is written up within the Previous Work section.