Digital Signal Processing Reference
In-Depth Information
14
VIDEO COMPRESSION
FUNDAMENTALS
CHAPTER OUTLINE
14.1 Block Size 112
14.2 Motion Estimation 114
14.3 Frame Processing Order 116
14.4 Compressing I-frames 117
14.5 Compressing P-frames 118
14.6 Compressing B-frames 119
14.7 Rate Control and Buffering
119
14.8 Quantization Scale Factor
120
MPEG is often considered synonymous with image compres-
sion. MPEG stands for Moving Pictures Experts Group,
a committee that publishes international standards on video
compression. MPEG is a portfolio of video compression stan-
dards, which will be discussed further in the next chapter.
Image compression theory and implementation focuses on
taking advantage of the spatial redundancy present in the image.
Video is composed of a series of images, usually referred to as
frames, and so can be compressed by compressing the individual
frames as discussed in the last chapter. However, there are
temporal (or across time) redundancies present across video
frames: the frame immediately following has a lot in common
with the current and previous frames. In most videos there will be
a significant amount of repetition in the sequences of frames.
This property can be used to reduce the amount of data used to
represent and store a video sequence. In order to take advantage
of the temporal redundancy, this commonality must be deter-
mined across the frames.
This is known as predictive coding, and is effective in reducing
the amount of data that must be stored or streamed for a given
video sequence. Usually only part of an image changes from
frame to frame, which permits prediction from previous frames.
Motion compensation is used in the predictive process. If an
