Automatic head detection for passenger flow analysis in bus surveillance videos

The passenger flow analysis in bus surveillance videos is of great importance. This paper addressed the detection of passenger heads in the image sequences captured by a monocular camera mounted on the bus ceiling. We exploited the combination features of consistent hair colour and solidity filling in head circles to tackle the main difficulties of cluttered background, lightning effects and non-rigid body movements in the videos. The experimental results demonstrate that our method is promising.