Technologies de l'information — Technologies vidéo MPEG — Partie 1: Exigences d'exactitude pour l'implémentation de la transformation cosinus inverse discrète de sortie du nombre entier 8x8

A number of image and video coding related standards include a requirement for decoders to implement an integer-output 8x8 inverse discrete cosine transform (IDCT) for the generation of inverse-transformed sample differences with a nominal range from -2^B to (2^B)-1 for some integer number of bits B, where B is greater than or equal to 8. ISO/IEC 23002-1:2006 specifies conformance requirements for establishing sufficient accuracy in such an integer-output IDCT implementation. It is intended to be suitable for reference to establish partial or complete requirements for IDCT accuracy for conformance to other standards that require IDCT use.

The accuracy requirements specified in the main body of ISO/IEC 23002-1:2006 are essentially the same as those previously specified in IEEE 1180-1990 (which has since been withdrawn), in Annex A of ITU-T Rec. H.261, and in Annex A of ITU-T Rec. H.263. These requirements have been specified herein to resolve normative references to IEEE 1180-1990 in MPEG standards after its withdrawal and to provide improved clarity for the specification of IDCT accuracy requirements.

An additional requirement on encoded-bitstream intra refresh frequency was also previously specified in IEEE 1180-1990, establishing a requirement of bitstream conformance that each macroblock be intra-coded at least once within each series of 132 times that it is coded in a predicted picture without an intervening intra picture. That additional requirement is not specified in ISO/IEC 23002-1:2006 , in order to confine its scope to the domain of decoder conformance specification.

Some allowances for modification of the specified accuracy requirements are made within ISO/IEC 23002-1:2006. Additional accuracy requirements that may be invoked by a referencing specification are specified in Annexes A and B.