MPEG-2

MPEG-2 (1994) is the designation for a group of audio and video coding standards agreed upon by MPEG (Motion Pictures Coding Experts Group), and published as ISO standard 13818. MPEG-2 is typically used to encode audio and video for broadcast signals, including digital satellite[?] and Cable TV. MPEG-2, with some modifications, is also the coding format used by standard commercial DVD movies.

MPEG-2 is similar to MPEG-1, but also provides support for interlaced video (the format used by broadcast TV systems.) MPEG-2 video is not optimized for low bit-rates (less than 1Mbps), but outperforms MPEG-1 at 3 Mbps and above. MPEG-2 also introduces and defines Transport Streams, which are designed to carry digital video and audio over unreliable media, and are used in broadcast applications. With some enhancements, MPEG-2 is also the current standard for HDTV transmission. A standards-compliant MPEG-2 decoder should be capable of playing back MPEG-1 streams.

MPEG-2 audio, defined in Part 3 of the standard, enhances MPEG-1's audio by allowing the coding of audio programs with more than two channels. Part 3 of the standard allows this to be done in a backwards compatible way, allowing MPEG-1 audio decoders to decode the two main stereo components of the presentation, or in a non backwards compatible way, which allows encoders to make better use of available bandwidth. MPEG-2 supports various audio formats, including MPEG-2 AAC.

MPEG-2 video encoding (simplified)

MPEG-2 is for the generic coding of moving pictures and associated audio and creates a video stream out of three types of frame data (intraframes, forward predictive frames and bilinear frames) arranged in a specified order called Group of Pictures (GOP). The originating material is a uncompressed video sequence at a pre-set pixel quality at 30 frames/second with sound.

Intraframes (I frames) are the key frames, they are the only parts of the video that contain an entire frame of data and every GOP starts with a I frame. I frames are created by applying the Discrete Cosine Transform algorithm (DCT, the same algorithm is used for JPEGs) to the video frame. The created I frame is divided into 16 x 16 pixel macroblocks, which also have 8 x8 data for colour (called chroma or chrominance) and light (called luminance or luma). The luma and chroma data are split into frequencies and tabulated then quantised (Huffman coded) to remove lower order bit data.

Forward predictive frames (P frames) store to diffence between the current frame and the previous I or P frame. Each macroblock in the two frames is compared and if they match motion vectors are calculated to create a frame that stores only significant differences, again compressed with the DCT algorithm. the P frames are as much as 90% smaller than I frames.

Bilinear frames (B frames) are similar to P frames but compare both the preceding and the subsequent I or P frame data. The B frames store the average of matching macroblocks or motion vectors, because the encode based on both preceding and subsequent data they are more effective at storing and displaying motion. B frames can be as small as 2% of a I frame but they require greater processing to encode and to decode.

I frames encode for spacial redundancy, P and B frames for temporal redundancy.

The sequence of different frame types is stored in a Group of Pictures (GOP) structure. There are a number of different structures but a common one is 15-frames as IBBPBBPBBPBBPBBPBB, a similar 12 frame sequence is also common. The balance of the GOP is determined by the nature of the video and the requirements of file size, although encoding time may also be an issue as a file containing many B frames can take three times longer to encode than a I frame only file.

However the influence of GOP on file size and quality is small compared to bit rate. For MPEG-2 the choice is constant or variable, with the maximum bit rate determined by the playback media - for example the DVD movie maximum is 10.4 Mbit/s. The quality of the frames is changed to match bit rate requirements, if needed the frame is quantised to remove more low order data at the cost of the macroblocks becoming more visible (mosaicing) as bit rate is reduced.

MPEG-2 audio encoding

MPEG-2 includes compressed (perceptual noise shaping) audio, usually as MPEG-1 Layer II which supports bit rates between 32 Kbit/s and 384 Kbit/s (384 Kbit/s is common for DVD movies)

MPEG-2 standards

ISO/IEC 13818-1

Systems - describes synchronization and multiplexing of video and audio.

ISO/IEC 13818-2

Video - compression codec for interlaced and non-interlaced video signals.

ISO/IEC 13818-3

Audio - compression codec for perceptual coding of audio signals. A multichannel-enabled extension of MPEG-1 audio (MP3).

ISO/IEC 13818-4

Describes procedures for testing compliance.

ISO/IEC 13818-5

Describes systems for Software simulation.

ISO/IEC 13818-6

Describes extensions for DSM-CC (Digital Storage Media Command and Control.)

ISO/IEC 13818-7

Advanced Audio Coding (AAC)

ISO/IEC 13818-9

Extension for real time interfaces.

ISO/IEC 13818-10

Conformance extensions for DSM-CC.