madjax

I am working on a more robust tracking program in max using a similar method as contained in the cv.jit.features2track patch: I select a point in the image and search for features (areas of high contrast) in a defined area around that point, and use these features as tracking points in cv.jit.track.

Is there a way to intentionally drop a tracking point while maintaining the rest of the points you are tracking? The object has a limited set of message commands, and the only thing I can think of is changing the "npoints" value. I can't see how to specifically affect a particular point that the object is tracking. Sending an "npoints" message will reset the position of all currently tracked points.

I want to use the average position of a number of features to track a person, but detect (using standard deviation) if one of those points is not traveling with the others (stuck) and drop that point, and perhaps re-target it.

I suppose I will have to code around the limitations of this object. Any ideas?

Is there a way to intentionally drop a tracking point while maintaining the rest of the points you are tracking?  The object has a limited set of message commands, and the only thing I can think of is changing the "npoints" value.  I can't see how to specifically affect a particular point that the object is tracking.  Sending an "npoints" message will reset the position of all currently tracked points.  

I suppose I will have to code around the limitations of this object. Any ideas?


cv.jit.track - change number of points and keep current locations