Scene Video Coding - Advanced Video Coding Systems - page 153

Table 8.8 Performance of surveillance coding tools of AVS2
Surveillance tools
Y(%)

U(%) V(%)

Low-delay
Crossroad − 22.10 − 56.10 − 51.20

Overbridge − 33.40 − 48.30 − 47.80

− − −

Office 12.00 34.50 30.80

Mainroad − 54.10 − 56.20 − 51.10

Intersection − 21.10 − 22.60 − 21.70

− − −

SD 22.50 46.30 43.30

1200p − 37.60 − 39.40 − 36.40

Overall − 28.50 − 43.50 − 40.50

Random access
Crossroad − 50.30 − 66.30 − 64.60

Overbridge − 71.50 − 76.90 − 75.80

Office − 34.70 − 44.60 − 42.90

Mainroad − 72.90 − 76.50 − 73.00

Intersection − 33.70 − 40.50 − 39.00

SD − 52.20 − 62.60 − 61.10

1200p − 53.30 − 58.50 − 56.00

Overall − 52.60 − 61.00 − 59.10

Random access with B
Crossroad − 41.30 − 63.30 − 60.90

Overbridge − 65.80 − 72.10 − 71.30

Office − 28.20 − 43.10 − 41.00

Mainroad − 65.00 − 72.30 − 69.60

Intersection − 31.80 − 35.30 − 34.40

SD − 45.10 − 59.50 − 57.70

1200p − 48.40 − 53.80 − 52.00

− − −

Overall 46.40 57.20 55.40

While employing the tools for better utilizing the MBG on HEVC [method is
namely BHO (Zhang et al. 2014c )], the average bit savings of BHO versus HEVC
encoder (HM) are 39.09, 65.63, and 60.47% for Y, U, and V components at low-
delay configurations, as shown in Table 8.9 . The RD curve examples are shown in
Fig. 8.13 .