Video Compression Techniques: The MPEG-1 Standard

The MPEG-1 Standard

This standard is named after the ISO committee that created it. It has a bandwidth of about 188 KBps, and consists of three layers:

MPEG video, encoded at 138 KBps
MPEG system (responsible, among other things, for A/V sync), encoded at 34 KBps
MPEG audio, the Third Layer, hence MP3, encoded at 128 kilobits per second (16 KBps)

MPEG-1 is based on progressively scanned images, so that interlaced source formats have to be converted to non-interlaced before coding. After decoding, these can be reconverted to interlaced if needed
The standard was originally defined for low bitrate coding of low spatial resolution pictures, but was subsequently improved to accept higher bitrates. Originally intended for CD-ROMs that had 1-2 Mbit/s data delivery capability, they have improved upon this with the increase in speed of CD-ROMs
As mentioned before, it is encoded in macroblocks (16x16 pixel arrays) in the YCbCr (another colour system, where Y is coded with lesser resolution than Cb and Cr) format, the algorithms operate on one of the four blocks of this macroblock at a time
It uses DCT, then quantization and finally entropy coding
There are four types of frames in each picture: I-, P-, B-, and DC- (Interpolative, Predictive, Bipredictional and DC in analogue with AC/DC ) frames are used in encoding information
The very first frame always has to be the I- frame
Each frame (if not the first frame) has a reference to the previous I- or P- frame. P- frames can have references to the NEXT I- or P- frames as well (forward prediction)
Spatial and temporal redundancy are used to compress further using techniques like Motion Compensation and Motion Compensated Prediction/Interpolation

For more information (there's a LOT!), go to the links section.

Go to the next page, or use one of the following quick links: