Digital Signal Processing Reference
In-Depth Information
Fig. 6.3
Event based scalable coding (Event occurs) [
16
]
VCA
Module at
encoder
VCA
Module at
decoder
Video
CCTV
Compressed
Scalable Bitstream
Decoded
Video
SCC decoder
SCC encoder
Compressed Domain
VCA
Fig. 6.4
Scalable video codec architecture [
22
]
6.5.1
GrabcutD: Modified GrabCut Using 4 Channels
GrabcutD is a modified version of Grabcut using depth-based information. In [
20
],
for experimentation purposes, we used the video dataset frames from MSR ballet
sequence dance video. Initially, for each frame, users select a bounding box and
the pixels inside and outside rectangle are represented by foreground and back-
ground classes respectively. From each trimap selection of the foreground and
background, the histograms are formed using 4 channels information (
Red, Green,
