A Consensus Based Method for Tracking Modelling Background S(21)

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

Figure 6 and Table 1 show the results obtained by SACON, and five other state-of-the-art methods. There are some isolated examples and measures where some methods have relative advantages. For the LS sequence, the Wallflower method and the Eigen-based method achieve less total error. For the FA sequence, the Wallflower method achieves the most accurate result based on the measure te. For MO sequence, except for the Eigen-based background model, almost all other methods achieve accurate results. However, our method achieves the most accurate overall performance on TE and TE* among the six competitive methods. SACON also obtains the best results of total error (te) for the five image sequences of MO, TOD, WT, C and B.

3.2 The Influence of the Parameters

There are two important parameters that are crucial to SACON: the value of T1 (in Equation 6) and the number of the background samples at each pixel. In this section, we investigate the influence of these two parameters on the results of SACON. We evaluate the results by TE* and corresponding total error (te) for each of the image sequences.

3.2.1 The Influence of T1 on the Results of SACON

In this experiment, we changed the value of T1 from 2 to 30, with interval 1.

From Figure 7, we can see that when the value of T1 varies, the influence on the results for various types of image sequences is different. The fluctuation of the te for B image sequence is relatively large, while relatively small for other image sequences. For the overall performance (see Figure 7 (b)), we can see that TE* fluctuates with the increase of T1. The shape of TE* in Figure 7 (b) is roughly the same as the shape of the te of B in Figure 7 (a). This is because it is the sum of several quantities which, except for the "Bootstrapping", remain relatively flat. Moreover, TE* is 21