Real-time detection of human body in videos

The paper describes a new method of detecting human figures in the video scene in real time. This problem can be found, for example, in the protection of buildings where unauthorized persons have access, surveillance of persons in common areas such as shopping centers, airport lounges, etc. For the detection of the contour of a human figure the HOG algorithm is often used which detects the human figure in static images. Its major disadvantage is the time-consuming calculation. The new method significantly speeds up the time for the identification of a human figure, and therefore it can be used for videos. The speeding up is based on reducing the search space when comparing moving objects against a static background.