Multimedia Technology

H.264 - A New Technology for Video Compression

The MPEG committee (ISO WG11) has been very successful in developing the enabling media coding standards for CD, video, satellite broadcast TV, DVD and HDTV. MPEG2 is now arguably the most successful new consumer standard ever relative to its acceptance in the marketplace. Presently it is the predominant standard for existing digital video equipment worldwide.

Based on MPEG2, digital television (both standard definition, SD, and high definition, HD) has become the norm for broadcasting, replacing analog broadcast in all but standard terrestrial and cable television. Within the near future terrestrial and cable broadcast television are expected to become all-digital. Digital TV has provided the opportunity for improved quality, bandwidth efficiency and enormous flexibility versus analog NTSC. The coding efficiency (bits per image for constant quality) of digital video compression algorithms is of paramount importance to the success and competitiveness of new systems. Because television consumers are always seeking improved quality and selection, experts are continually striving to find better algorithms and technologies to improve existing systems. In addition, manufacturers and broadcasters are seeking improved technologies to get a competitive edge with new products.

In 2000, the MPEG standards committee completed work on MPEG4, Part 2 (video). However, instead of improving the coding efficiency relative to MPEG2, MPEG4 added more functionality and flexibility but did not significantly improve on the efficiency of the MPEG 2 video coding algorithm. At the same time, the International Telecommunications Union (ITU-T) which had adopted MPEG2 for video teleconferencing was unhappy with the performance of the MPEG2 algorithm, denoted as ITU-T H.262, and decided to start an independent project to develop a significantly better compression algorithm. Starting with new ideas and accepting innovative algorithms, this standard committee, VCEG, has now completed work on a completely new coding algorithm, designated H.264. Over the past two years H.264 has be continually improved and tested against the MPEG4-2 advanced simple profile relative to picture quality and coding efficiency. The results demonstrate that H.264 achieves at least a 2X improvement over MPEG4. Thus H.264 is likely to supplant MPEG2 and MPEG4 in new applications. Algorithm Comparison

MPEG4 and H.264 have a common heritage within the ISO and ITU standards committees. As a result the overall coding approach is quite similar. Both algorithms are based on a common heritage of DCT based, hybrid image coding, first used in H.261 and MPEG1. A block diagram applicable to both H.264 and MPEG coders appears in Figure 1 below. Figure 2 illustrates the structure of the MPEG and H.264 decoders.

Both algorithms can be represented by the same high-level block diagrams. Both algorithms are image transform coded, forward/backward block motion compensated prediction with entropy coded transform coefficients. However the algorithms differ significantly in the methods selected for image transforms and entropy coding. The detailed differences are summarized in Table 1 below.

                        Coding Efficiency Tests

A number of comparison tests have been performed between H.264 and MPEG2/MPEG4 on standard MPEG test material. The chart above depicts comparison tests using two standard resolution (704 X 480, 60 Hz interlaced) video sequences named “NBA” and “BUS”. In the figure, the curves from top to bottom respectively are, NBA – MPEG2, BUS – MPEG2, NBA – H.264 and BUS – H.264. The results were reported in ITU-T document VCEG-N84. The chart shows the required coding rate to achieve various quality levels as measured by the standard PSNR (Peak Signal to RMS Noise Ratio) calculation. Notice, for example, that to achieve a PSNR level of 28, NBA must be coded at a rate of 5 mbps using MPEG2, but only 1.8 mbps using H.264.

Algorithm Comparison

MPEG4 and H.264 have a common heritage within the ISO and ITU standards process. As a result the overall coding approach is quite similar. Both algorithms are based on a common heritage of block DCT transform based, predictive and entropy coding, first used in H.261 and MPEG1. More precisely, the algorithms are image transform coded, forward/backward block motion compensated prediction with entropy coded transform coefficients. However the algorithms differ significantly in the methods selected for image transforms and entropy coding. The detailed differences are summarized in Table 1 below.

Table 1: Comparison of H.264 Coding Algorithm to MPEG2/MPEG4 Algorithms

Algorithm Characteristic	MPEG2/MPEG4	H.264
General	Motion compensated predictive, residual transformed, entropy coded	Same basic structure as MPEG
Intra Prediction	None	Multi-direction, Multi-pattern
Coded Image Types	I,B,P	I,B,P, SP
Transform	8x8 DCT	4x4 DCT-like Integer Transform
Motion Estimation Blocks	16x16	16x16, 8x8, 8x4, 4x4
Entropy Coding	Multiple VLC Tables	Arithmetic Coding and adaptive VLC Tables
Frame Distance for Prediction	+/- 1	Unlimited forward/backward
Fractional Motion Estimation	1/2 Pixel (MPEG2) 1/4 Pixel (MPEG4)	1/4 Pixel
Deblocking Filter	None	Dynamic edge filters

next                   top