__/ [derek] on Monday 07 November 2005 05:43 \__
> Dave: I'm not sure if your question is directed at me or not... but I
> can tell you what the voids are, if you're interested. The fracture
> surface that I mentioned above is of an adhesive of which we're trying
> to ascertain properties under dynamic load. For whatever reason, one
> batch of the adhesive that we got does not perform in the same manner
> as a different batch, and the easiest explanation seems to be that
> somewhere in the packaging process (of the adhesive), there was a fair
> amount of air trapped inside of the stuff. This causes, of course,
> 'voids' where there should have been adhesive when you look at the
> fracture surface. Does that make any sense?
Not only does it makes sense, but it makes the disucssion a more interesting
You might also wish to try sci.image.processing where people can suggest
algorithms and recommend software packages.
> Roy: Thanks much for the idea... sorry though, I'm not too familiar
> with available graphics programs... do you have any suggestions for
> something I might use? It might take some fiddling to get it to work
> right, as there's a bit of black in the pictures that is not caused by
> the voids. Of course, I'm not sure what 'shade' of black it is, so it
> may be that, like you said, I can pick out certain parts of the
> histogram that correspond to the particular colors that I find in the
> Anyways, thanks for the responses! I'll keep at it.
If you have some 'interference' due to other smaller dark elements, try to
classify them. If you are going to deal with many such images, I also
suggest that you automate this. To classify based on size, try checking the
pixel neighbourhood and put a threshold on the number of neighbouring
pixels. If you save the image as a 24-bit (1 byte red, 1 byte green, 1 byte
blue) bitmap file, then you can easily scan the file and check the
thresholds while accumulating a count of dark pixels.