ABSTRACT
The background subtraction (BS) is an important task on computer vision applications such as moving object detection (e.g. to detect/segment vehicles [31] and peoples [27]), video surveillance [5] and multimedia applications [10]. Generally the BS consists of comparing an image with another image which represents an estimation of the background model. The foreground objects can be detected by the image regions that have a significant difference between the input image and the reference image (background model). Typically the BS process is defined by: a) background model initialization up to a certain threshold, b) background model maintenance (after the threshold) and c) foreground segmentation. The Figure 23.1 shows the block diagram of the background subtraction process described here.
