I was just trying to process a multi line graph and I can't seem to discretize a red trace. It seems like the white background registers as the same colour as red so I can't isolate that colour. Everything else worked fine. Can anyone help me out?
are you sure you also tried discretizing based on hue?
When two colors are quite close intensity wise, you should
be able to separate them by one of the other options.
If it does not work out, an image demonstrating the problem
would help. Just email me.
Yes, Tobias is correct.
The various discretizing options (intensity, foreground, hue, saturation, value) can be surprisingly powerful. In the samples directory, the pointplot.bmp file is black and white, yet discretizing by hue lets you select the diamond points by specifying a range of 10 to 50, or the triangle points by specifying a range of 90 to 99.
Personally, I find it easier to just abandon theory and simply try out every discretizing option, and slide the range bars back and forth between histogram "mountains" until I get what I want.
In the Discretize settings dialog you can try various discretizing methods, and adjust the lower and upper limits until you've extracted just the pixels you want.
Tobias has sent me a patch regarding this issue, but I'll finish some ongoing work on Point Match before I look at that (probably tomorrow). I'll also try to capture some of these suggestions in the documentation.
Oops, my mistake. The previous message (2006-12-13 00:52) says "discretizing by hue", but I later realized that I should have said "discretizing by intensity". The rest of the message still holds true however.