Paused Object Detection with Background Updating

[IC 4] Won-Du Chang, Jungpil Shin, "Paused Object Detection with Background Updating", Proceedings of the Ninth International Conference on Humans and Computers, pp.13-18, Aizu-Wakamatsu City, Japan, Sep. 2006. [pdf]

Abstract: Background subtraction algorithm is one of two main methods of extracting object from the image sequence. It has a merit of being able to detect much region of objects region clearly even when the object is paused where the background image is known to the computer, while the method based on frame difference algorithm can detect moving region only. In real world application, however, background is continuously changing and it makes harder to extract the stable background image. Though there have been researches on extracting background, most of them are focused on updating the background and lost its own merit of detecting paused object. In this paper, we propose a method to detect paused object updating background image continuously. After obtaining the first background image, the image regions are classified into three groups of moving, paused and non-moving region using background subtracting and frame difference algorithms. Consequently, background updating is applied differently for each group region. Paused moving regions are updated only with average changing amount of illuminance, while general method is applied for the non-moving region. Through the experiment, proposed algorithm is able to detect paused object without interfering the detection of other moving objects, while the background image is continuously updated.