Passenger Flow Direction Detection for Public Transportation Based on Video

This paper aims to analyze the direction of the passengers' movement according to the video shot in the public locations. The analysis result will help related departments to manage the transport situation and make a decision when confronting emergency. First, the paper realizes the detection of people's head based on Haar feature and Adaboost algorithm through OpenCV, Second, the paper investigates the head recognition technology based on color histogram, and proposes an improved algorithm which adds the spatial information into the color histogram, At last, the paper proposes to set up a human tracking chain to realize the human tracking and analyze the human moving direction.