ADAPTIVE MOTION ESTIMATION ALGORITHM FOR CROWDED SCENES

The motion estimation in video pictures has attracted a great deal of
interest in computer vision. For crowd analytics and surveillance systems, motion
estimation is an indispensable first-step. Lots of crowd motion estimation algorithms
have been presented in the last years comprising pedestrian motion. Most of this
method depends on optical flow and background subtraction. Several areas, such as
evacuation movements, pedestrian movements at transit stations and local interaction
behavior, were studied. However, algorithms based on optical flow and background
subtraction cannot estimate themotion in real timeespecially in crowded scenes.