A study on analysis of road image including a pedestrian crossing

An analyzing method of walker behavior on a pedestrian crossing is considered. As the walking speed and the moving directions of a walker is analyzed by the proposed method based on Freeman´s chain code, a pedestrian´s face dangerous situations and a warning message to a driver can be given. The road image, including pedestrian crossing, is acquired by using a digital video camera which is set up on the fourth floor of a building such as that a human region is easily obtained. The extraction of the individual walker regions is executed by subtracting the background image from the enquired image. Finally, the extracted walker region is pursued by following the track of the chain code stringy and the dangerous measurement is calculated by the track. The simulation result of the walker extraction show a 88.6% accuracy. Another simulation shows that the automatic system of the proposed method can judge whether the walker situation is normal or not.