My background and profession is as a Graphic Designer / Image Analyst and Image Restorer. I intend to take a slightly different route of an image based approach. As there are individuals within the group who are far more proficient in programming, I intend to examine image sectors both in original monochrome HD scans, and their resulting SD CR output. I hope to be able to identify some of the reasons why the software is falling down on certain sections, and perhaps avoid the need to adjust the quadrant.

As we know film stock and film recorder equipment may play a part in the difference between a reasonably accurate colour recovery (ie Dad's Army) and one with many errors (ie Planet 3), I hope I may be able to reverse the effect. I have access to many image analysis programs and it is my hope I may be able to utilise these and my knowledge to find out what's going wrong. If I'm sucessful, I will then experiment with compensating for this effect. Perhaps by boosting the chroma noise in specific areas where it is falling below levels the CR can pick up.
I will post more details of experiments as I begin them.

If a method could be devised to detect low chroma noise and boost it on a frame by frame basis - it is hoped this algorithm could be incorporated into the existing software to produce less errors.