Background Substraction for ROI to use OpticalFlow tracking

asked 2019-06-12 07:34:52 -0500

Is this a feasible way for real time Object detection and tracking? Im using video data of surgical instruments on a static test background. The instruments are the only moving objects but simply tracking them with Lucas-Canade Optical flow gets those points lost when the instruments move to fast or move by black regions in the background.

It seems like the parameter of moving should be a strong feature in this environment.

I would appreciate any tips to further look into. Thanks in advance!

edit retag flag offensive close merge delete