Automatic Counting of People Entering and Leaving Based on Dominant Colors and People Silhouettes

Abstract

Counting of people crowd is an important process for video surveillance, anomaly warning, public security control, as well as ongoing protection of people and public facilities. People counting methods are also applied for controlling the numbers of people entering and leaving such places as a tourist bus, an office or an university building, a public building, a supermarket or shopping mall, a culture or sport center, etc. The problem arises when the number of exiting people is not equal to the number of entering people. How many people are missing and who is missing? The paper presents an approach for people counting and detection of missing persons. This approach includes two procedures. First, the dominant color of people detected on video was analyzed. Next, the silhouette sizes were used. Both procedures finally allow us to define specific features distinctive for missing people. The results of tests performed with video recordings of people entering and exiting through the door are promising. In this approach individual’s identities are not registered; therefore, privacy violation is avoided.