On Thu, 13 Aug 2009, sisil mehta wrote:
I found a patch called pix-multiblob-help.pd in PD-extended. is it similar to the multiblob-for-gridflow patch that u had written?
Well... yes and no. If you compared the help file with what I've told you about [#labeling], you would see that several features don't match: with [#labeling] you're supposed to do the threshold yourself (with [# >=] or anything else); [#labeling] does not give you the min/max (bounding box) but [pix_multiblob] does not give you the variance. In my opinion the variance is a very useful measurement, but then that's why I wrote [#labeling] the way I did.
Also, [#labeling] gives you the data in little pieces instead of a big table like [pix_multiblob] does. You'd think it would be more in GridFlow's style to put data in big chunks (which is the normal way to do things in GridFlow) but I made this an exception to make it easy to filter out unwanted blobs using [spigot]. For example, in multiblob.png you can see that it only keeps blobs that have at least 25 pixels.
You could make any other filters you want: for example, you could keep only blobs that spread on a large amount of space (regardless of actual area of the blob); or you could keep directional blobs (long and narrow), which would keep narrow ovals and narrow rectangles, to reject circles and squares (which are as long as they are wide).
I don't know what the iemmatrix offers for processing [pix_multiblob]'s output. Perhaps Johannes can give some examples.
_ _ __ ___ _____ ________ _____________ _____________________ ... | Mathieu Bouchard - tél:+1.514.383.3801, Montréal, Québec