Atsushi Sato

A moving-object extraction method robust against illumination level changes for a pedestrian counting system

The paper presents a moving object extraction method robust against illumination level changes for a real-time pedestrian counting system based on an x-t space-time image. This system, called the Ped-Counter, counts the number of pedestrians using TV camera images and image processing. The TV camera takes images of a path or the entrance of a store or an exhibition hall. A practical location for the counting system is under constant illumination conditions, such as under an arcade or in a room. If the system is used in other locations and under different conditions, for example, an outdoor path or an entrance where there is direct sunlight, then it must be robust against illumination level changes. The paper proposes two new processes for extracting moving objects from space-time images using background subtraction. The first is a background image reconstruction process using statistical characteristic analysis of temporal changes of a target pixel. The other is a moving object extraction process using delayed subtraction. These processes make the system robust against illumination level changes. This method is evaluated over a half-year period at an entrance experiencing drastic illumination level changes. We also confirm that this method can extract moving objects, that it is robust against season, weather, and time, and that the Ped-Counter is accurate.

Pedestrian counting system robust against illumination changes

This paper presents a realtime pedestrian counting system based on x-t spacetime image analysis. The system counts the number of pedestrians through the combination of three processes: moving object extraction, object counting by region labeling, and direction detection by flow estimation. In order to extract objects reliably under various illumination changes, we propose new moving objects extraction methods that extract moving regions whose shadows have been eliminated by using adaptive background image reconstruction and the color information processing of the images. For labeling and flow estimation, we use the Ortho-sectioning method that analyzes extracted regions on a spacetime slice image of the original three dimensional volume. Experimental results confirm the robustness of the system against illumination changes.