Ahmed Mahmoud HAMAD^1,2* and Norimichi TSUMURA¹

¹Graduate School of Advanced Integration Science, Chiba University, Chiba 263-8522, Japan ²Department of Information Technology, Faculty of Computers and Information, Menofia University, Egypt

(Received October 3, 2011; Accepted January 27, 2012)

Background model updating is a vital process for any background subtraction technique. This paper presents an updating mechanism that can be applied efficiently to any background subtraction technique. This updating mechanism exploits the color and spatial features to characterize each detected object. Spatial and color features are used to classify each detected object as a moving background object, a ghost, or a real moving object. The starting position of each detected object is the cue for updating background images. In addition, this paper presents a hybrid scheme to detect and remove cast shadows based on texture and color features. The robustness of the proposed method and its effectiveness in overcoming challenging problems such as gradual and sudden illumination changes, ghost appearance, non-stationary background objects, the stability of moving objects most of the time, and cast shadows are verified quantitatively and qualitatively.

Key words: background subtraction, statistical modeling, texture analysis

^*E-mail address: hamad@graduate.chiba-u.jp