Game Development Reference
In-Depth Information
Fig. 3.10
Examples of the background picture and the difference frame between the original picture
and the background picture.
a
Original 6th P frame.
b
6th difference frame.
c
Background frame
prediction, as shown in Fig.
3.10
, which can improve foreground prediction perfor-
mance by excluding the background influence. It can be seen that, after background
difference prediction, the background redundancy is effectively removed. Further-
more, according to the predication modes in AVS2 compression bitstream, the blocks
of an AVS2 picture could be classified as background blocks, foreground blocks or
blocks on the edge area. Obviously, this information is very helpful for possible
subsequent vision tasks such as object detection and tracking.
To provide convenience for applications like event detection and searching, AVS2
added some novel high-level syntax to describe the region-of-interest. In the region
extension, the region number, event ID, and coordinates for top left and bottom right
corners are included to show what number the region of interest (ROI) is, what event
happened and where it lies.
AVS2 Syntax Structure
The AVS2 coded bitstream has a hierarchical structure, and it can be segmented
into three layers, e.g., sequence, picture, slice. Each layer is started with a 32bits
start code. Table
3.2
shows the start code definition in AVS2. And slice is further
composed of a series of LCUs. In one LCU, CU is the basic coding and bitstream
organization unit. For each CU, the prediction information of the PUs in it, including
prediction partition, prediction mode, motion vector, etc., are transmitted first. Then
the transform coefficients are transmitted.
3.3.1 Sequence and Picture
Sequence is the highest syntax structure of the coded bitstream,which is composed of
consecutive coded pictures. It starts with a sequence header, followed by one or more
coded pictures. The initial sequence header can be allowed to appear repeatedly in
the bitstream—which is called repeated sequence header, and makes random access
into the potential sequence.
In the sequence, each coded picture starts with a picture header, and is ordered
accordingly in the bitstream. The order of these pictures in the bitstream is the same
