Thanks KE5FX, I had thought about this a bit in general terms, but had not worked out the specifics. Do you mean actually deriving the logo in real-time from the video stream, or simply finding the spatial position of a predetermined logo? The former might be feasible, though i think it would depend on the intensity distribution of each pixel in the image being uniform. You might end up turning all the letterbox bars to grey unless some fairly advanced heuristics were used. Perhaps I will try some further experiments sometime.