Hello pd list. I would like to know if its possible or if there is a way to have something like the pix_multiblob object but not based on luminosity but instead on background substraction. I been seen that with opencv it is possible to track blobs based on backgound substraction instead of luminosity. How can i achieve this in puredata?
thanks in advance
Ronni Montoya Huamani
You want [pix_background]. Please see attached patch, originally by Ben Bogart.
D.
On 8/28/10 7:07 AM, ronni montoya wrote:
Hello pd list. I would like to know if its possible or if there is a way to have something like the pix_multiblob object but not based on luminosity but instead on background substraction. I been seen that with opencv it is possible to track blobs based on backgound substraction instead of luminosity. How can i achieve this in puredata?
well, errr, there is opencv for pure data ( pdp or gem ) : http://www.hangar.org/wikis/lab/doku.php?id=start:puredata_opencv
but i think it's not linked anywhere, hum, yes, i love democracy
ciao, sevy
ronni montoya wrote:
Hello pd list. I would like to know if its possible or if there is a way to have something like the pix_multiblob object but not based on luminosity but instead on background substraction. I been seen that with opencv it is possible to track blobs based on backgound substraction instead of luminosity. How can i achieve this in puredata?
thanks in advance
Ronni Montoya Huamani
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
On 08/28/2010 07:07 AM, ronni montoya wrote:
Hello pd list. I would like to know if its possible or if there is a way to have something like the pix_multiblob object but not based on luminosity but instead on background substraction. I been seen that with opencv it is possible to track blobs based on backgound substraction instead of luminosity. How can i achieve this in puredata?
[pix_multiblob] actually has nothing to do with "luminosity". it internally uses a single-channel image. if you pass it an RGB image. it will convert to the single-channel image based on luminosity. if you pass it a grey-scale image, it will simply accept that as the input.
it doesn't care about the semantics.
so you only have to create a single channel (aka greyscale) image that "is" what you need (e.g. difference between a background image and the current frame), then you can feed that to the object and be happy.
fgm,asdr IOhannes