image processing - Why don't we use motion vector data from video in optical flow? -
all implementation saw in optical flow in opencv uses video array of frames , implement optical flow on each image. involves slicing image nxn block , searching velocity vector.
although motion vector in video codec misleading , not contain motion information, why don't use check block has motion , run optical flow on blocks ? shouldn't fasten process ?
opencv universal image processing framework. takes in frames, not compressed video, algorithms.
you can write video decoder hands out info displacement codec opencv – codec-specific , isn't in scope of opencv itself.
Comments
Post a Comment