Max
How about measuring the change in the pix_blob output? If if changes more than a certain threshold amount that would normally be a sign of some new action in the frame. You could also detect the edges of the frame for objects entering view. A table with the past few frames might help smooth things out and improve accuracy as well.
cgc
On May 30, 2005, at 6:22 PM, Max Neupert wrote:
hi IOhannes, thomas, list
sorry for beeing unclear, pix_blob seems already too intelligent for what i need. i do NOT want to TRACK anything. i just want to know if there is ANY motion in the image and not where. so again: it's not about tracking it is about movement. my approch was to use pix_movement because it provides the first step in the analysis: removing everything what stays. but how to “measure” the white areas after without tracking them? i am not interested in the coordinates of blobs.
max
Am 30.05.2005 um 17:09 schrieb IOhannes m zmoelnig:
Max Neupert wrote:
hi list, how do i turn a camera in a motion sensor? i can extract the movement with pix_movement, that's clear. but how to get a numeric value out of the whole picture corresponding to the pictures grayscale value? (not from a single pixel like pix_data supplies) am i completely overseeing something?
yes: [pix_blob] does blob-detection (only 1 single blob though) there should be an example-patch on how to do movement-detection with this: 04.video/03.movement_detection.pd
if using the CVS is an option for you, there is [pix_multiblob] which is far better (detects multiple blobs, gives more information on them)
there is even an abstraction (with example-patch) for blob-tracking.
mfg.ad.sr IOhannes
max _______________________________________________ PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list