c++ - Improving camshift algorithm in open cv -


i using camshift algorithm of opencv object tracking. input being taken webcam , object tracked between successive frames. how can make tracking stronger? if move object @ rapid rate, tracking fails. when object not in frame there false detections. how improve ?

object tracking active research area in computer vision. there numerous algorithms it, , none of them work 100% of time.

if need track in real time, need simple , fast. assuming have way of segmenting moving object background. can compute representation of object, such color histogram, , compare the object find in next frame. should check object has not moved far between frames. if want try more advanced motion tracking, should kalman filter.

determining object not in frame big problem. first, kinds of objects trying track? people? cars? dogs? can build object classifier, tell whether or not moving object in frame object of interest, opposed noise or other kind of object. classifier can simple, such constraint on size, or can complicated. in latter case need learn features can computed, classification algorithms, such support vector machines, , need collect training images train it.

in short, reliable tracker not easy thing build.


Comments

Popular posts from this blog

python - Scipy curvefit RuntimeError:Optimal parameters not found: Number of calls to function has reached maxfev = 1000 -

c# - How to add a new treeview at the selected node? -

java - netbeans "Please wait - classpath scanning in progress..." -