Klaus Karcher wrote:Interesting. If I got it right, you consider a low slope as indication for channel saturation? It's exciting to see how sensitive this indicator becomes at the end of the scale:<http://digitalproof.info/argyll/inklimits-DeltaEslope.pdf>
Graeme Gill wrote:
That was the idea. Of course some smoothing would help in picking a robust stopping point.
I tested some smoothing algorithms with my data (see <http://digitalproof.info/argyll/inklimits-DeltaEslope-sm.pdf>).
Your method is definitely worth to be evaluated in practice. Moreover I think it is applicable to several similar problems (e.g. sensor calibration) -- I'll certainly recall it next time when I have to deal with issues like that.
Thank you :-) Klaus