Regards the Alignment Algorithm?

[whitedavidp]

The default for this is to use the red channel only. Using high accuracy adds checking of both blue and green channels.

On my meter, the last two digits on the right are in red printing and I have had to resort to passing the built-in LED light through a green filter to get anything close to decent recognition. I take this to mean that in doing so, the contents of the red channel will be significantly diminished. So I wonder if this is the reason I seem to need to use high accuracy instead to get decent alignment results?

If this is true, might it be possible to add a sub-option when the default algorithm is used that permits the user to choose either red, green, or blue channel to be used? It might help my performance significantly as this algorithm is documented to take 3 times longer and it seems that the choice of using red in the default is somewhat arbitrary.

Best!

[caco3]

You could try to use the cam effects, eg. to set it to gray scale or negative.

[whitedavidp]

Thanks again. I have actually already tried all of this. Gray scale and especially negative seem to work better. I also turn on zoom. I have fiddled with basically all the settings I can find and still have inconsistencies I cannot explain (for example, 2 consecutive readings with no water being used giving 2 different - and sometimes crazily different - results).

No matter what I have tried, I note that the photo returned is 640x480. I note here in section 1.2 a screen shot showing an image size setting but cannot find it in v 16.0.0. I wonder if increasing resolution might help. But no idea how to do this. I tried the expert Image Quality setting to 8. I do not seem to experience stability issues that are warned against. But it also does not seem to improve recognition.

Best!

[whitedavidp]

I just realized that while I enabled zoom, I did NOT (like a block-head) set a non-zero zoom level. I have set that to 4 - the highest I could get while still keeping all of the ROIs in the image and will see if I do any better with the recognition. Thanks