Skip to main content
. 2010 Jun 18;10(6):6092–6114. doi: 10.3390/s100606092

Figure 7.

Figure 7.

Foreground detection from images sequences of the PETS database [17, 18] using the MOG approach[3]. The sequences in (a) and (b) correspond to cameras monitoring a train stop; the sequence in (c) corresponds to a car parking. The objects detected are not well defined, and in several frames, some objects are discarded (especially in thin object structures). The motion detection is sensitive to shadows and images are pruned with noise.