A Consensus Based Method for Tracking Modelling Background S(13)

Modelling of the background (“uninteresting parts of the scene”), and of the foreground, play important roles in the tasks of visual detection and tracking of objects. This paper presents an effective and adaptive background modelling method for detectin

2.3.5 Updating Background Samples

When the background scene changes, the background samples should be updated to reflect the change of the background scene. Generally speaking, the background samples should be updated so that the background model can:

adapt to light condition changes such as gradual illumination changes;

adapt to moved or inserted background objects (classifying these as background).

adapt the foreground objects, which remain static for a long time, (classifying these as background, e.g. in [1]).

There are several methods to update the background samples [3, 14]. The simplest method is to blindly add each pixel of the current frame to the background scene. However, this method also adds the foreground pixels to the background samples. Another simple but more efficient method is to selectively add only pixels marked as background pixels to the background model while neglecting the foreground pixels. This method is efficient in gradual illumination changes. However, the method also causes some problems: for example, if a background object is moved to a new place, or if a new background object is inserted to the background scene, the method can not adaptively add the corresponding pixels of the background object to the background model. In our method, we use a selective update mechanism to update the background samples. We update the background samples at both pixel level and blob level described in the following subsections. Our method can incorporate the moved/inserted background object into the background model.

2.3.5.1 Updating Background at Pixel Level

To incorporate the moved/inserted background object into the background model, we use a Time Out Map (TOM) (similar to [10] - see below). A simple counter, TOMt(m)is created for the time 13