"... In PAGE 6: ... Another limitation is that targets which are very small ( lt; 5 5 pixels) tend to be temporally inconsistent and hence rejected. Table1 shows the results of the classification algorithm applied to over four hours of live video in an unstructured environment. The main problem with vehicle recognition is that when vehicles are partially occluded for long times, they are sometimes rejected.... ..."

"... In PAGE 6: ... Another limitation is that targets which are very small (#3C 5 #02 5 pixels) tend to be temporally inconsistent and hence rejected. Table1 shows the results of the classification algorithm applied to over four hours of live video in an unstructured environment. The main problem with vehicle recognition is that when vehicles are partially occluded for long times, they are sometimes rejected.... ..."

"... In PAGE 5: ... Gaussians with a too weak contribution to the mixture are eliminated. ) 8 ( _ dimT round g nb = (4) We evaluated the classifier on the video sequence recordings ( Table1 ) using 8-fold cross-validation. Each sequence has been used for testing once, while learning the model with the 7 remaining sequences.... In PAGE 7: ... The class with the highest counter value after all classes have been compared is selected. Again we evaluated the classifier on the video sequence recordings ( Table1 ) using 8-fold cross-validation. A radial basis function kernel with C=11.... ..."

"... In PAGE 3: ...ased rate shaping strategy for both SVC and H.264/AVC SLC. Both SVC and SLC bitstreams are generated using the JSVM software [7]. In our simulation setup, five different video sequences in CIF resolution are employed, whose characteristics are summarized in Table4 . All the video sequences are encoded at 30 fps and have a GOP size of 16 frames.... ..."

"... In PAGE 18: ...simpsons), Asterix (asterix), Mr. Bean (mr.bean), Formula 1 Car Race (race ). Table7 lists mean rate, peak rate and Hurst parameter for each of these traces. The Hurst parameter (H) is useful to determine the long-range dependence of a trace: values of H close to 1 indicate a high degree of LRD, whereas the absence of LRD is denoted by H = 0:5 (see [23]).... ..."

"... In PAGE 4: ... A subset of all investigated sequences is given in Tables 1 and 2. The video sequences given in Table 1 are used also for the statistical evaluation, while sequences in Table2 are listed because of specific character- istics found. The tables give the sequence name and video and audio information.... ..."