Abstract: In order to improve the accuracy of people counting in video surveillance, an
approximate pedestrian size in the vertical view is proposed as a feature when the camera is
mounted vertically. In this method, a virtual detection line is set up and the Lucas-Kanade optical
flow algorithm is used to calculate the velocity vector of pixels on the line. The method constructs
the approximate pedestrian size in the vertical view to estimate the number of people passing
through the detection line. Experimental results show that this method can reduce interference
caused by shadows and image preprocessing, and can accurately count the number of people
passing through the entrance.

Key words: people counting, pedestrian size, Lucas-Kanade optical flow algorithm