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

The Number

of Frames

Foreground

Region Size

Test Image

MO TOD LS WT C B FA 300 1851 226 248 252 300 230 0 1548 3247 5881 10565 2925 5022

Ground Truth

SACON

Tracey LAB

LP

Mixture of

Gaussian

Bayesian

decision

Eigen

background

Wallflower

Figure 6: Experimental results by several methods on the seven canonical background problems of the Wallflower benchmarks. The first row shows the number of frames that are used for each image sequence; The second row shows the size of the foreground regions (in pixels); The third row shows the evaluated frames of each image sequences; the fourth row shows the hand-segmented ground truth; the fifth row shows the results of SACON. The sixth row shows the results of Tracey reported in [17]; the seventh to the tenth rows show the results reported in [27].

19