Resultados por GIXRD obtenidos para el acero 1.4981

The Baseline Profile supports coded sequences containing I- and P-slices. I-slices contain intra-coded macroblocks in which each 16× 16 or 4 × 4 luma region and each 8 × 8 chroma region is predicted from previously-coded samples in the same slice. P-slices may contain intra-coded, inter-coded or skipped MBs. Inter-coded MBs in a P slice are predicted from a number of previously coded pictures, using motion compensation with quarter-sample (luma) motion vector accuracy.

After prediction, the residual data for each MB is transformed using a 4× 4 integer transform (based on the DCT) and quantised. Quantised transform coefficients are reordered and the syntax elements are entropy coded. In the Baseline Profile, transform coefficients are entropy coded using a context-adaptive variable length coding scheme (CAVLC) and all other

Figure A.2: Syntax of a slice.

When an I picture is lost, e.g., due to transmission errors, the recon- structed video quality is very bad as all the subsequent frames are en- coded depending on the I picture. To reduce the propagated error, it is possible to randomly insert intra-coded MBs in the bitstream. (In the H.264/AVC reference software this is denoted by an encoding parame- ter called RandomIntraM BRef resh). In this dissertation we neglect the occurrence of errors or frame drops in the video stream. As such, RandomIntraM BRef resh is set to 0, meaning no random Intra-coded MBs are inserted in the bitstream.

Intra Prediction

Intra prediction, in H.264/AVC, is conducted in the spatial domain. To predict a block, neighbouring samples of previously-coded blocks which are to the left and/or above the block, are used.

Two main types of Intra coding exist: Intra-4×4 and Intra-16×16. When Intra- 4×4 is used, each 4×4 block of luminance samples of a MB is predicted sepa- rately using one of nine prediction modes. This mode is well suited for coding parts of pictures with much details. For Intra-16×16, four uniform prediction modes are defined to predict the luma component of a whole MB. Homoge- neous areas of a picture are best coded with this mode. Figure A.3 shows the first four Intra-4×4 prediction modes.

I J K L A B C D M E F G H mode 0: vertical I J K L A B C D M E F G H

mode 3: diagonal down-left

I J K L A B C D M E F G H mode 1: horizontal I J K L A B C D M E F G H

mode 4: diagonal down-right

Figure A.3: Four of the nine Intra-4×4 prediction modes in H.264/AVC.

Inter Prediction

MBs which are inter predicted are assigned a MB type. This type denotes a specific partitioning of the 16×16 block. The partitions can have a size of 16×16, 16×8, 8×16, or 8×8 pixels. If the latter partitioning is chosen the 8×8 blocks can be further partitioned in partitions of 8×4, 4×8, or 4×4 pixels. For each partition a motion vector (MV) is created to denote the best match of the current partition with a block of corresponding size in a reference picture. The MV components are differentially coded using either median or directional prediction from neighbouring blocks.

A MB can also be coded as a skipped MB, called a P Skip MB. In this case no prediction residual is coded, nor a MV or reference picture is signalled. The MV, used for reconstructing such a P Skip MB, is calculated based on the MVs of neighbouring blocks. These types of MBs are very useful when large areas of no change between pictures occur in the sequence, since this can be coded with very few bits.

Transform and Quantization

H.264/AVC applies transform coding to the prediction residual. The transform coding happens on 4×4 blocks and consists of a separable integer transform.

Figure A.4: Profiles in H.264/AVC.

After the transformation, H.264/AVC applies zig-zag scanning, scaling and rounding as a quantization step. The latter is determined by a quantization parameter (QP), which influences the amount of compression.

Entropy Coding

For transmitting the quantized transform coefficients, two entropy coding methods are defined: Context-Adaptive Variable Length Coding (CAVLC) and Context-based Adaptive Binary Arithmetic Coding (CABAC). Both methods use statistics about previously coded symbols as a basis for the prediction for the current symbol (context).

Profiles

As in most other specifications for digital video coding, H.264/AVC uses the concept of profiles and levels. A profile defines a set of coding tools or al- gorithms that may be used to generate a compliant bitstream whereas a level imposes constraints on certain key parameters such as bit rate, resolution, or frame rate. Three profiles were defined in the first version of the H.264/AVC specification, each targeting a different range of applications.

Baseline Profile The Baseline Profile tries to minimize the complexity of the coding tools while including many tools for robustness as it is intended to be used in a broad range of transmission networks where the quality of service is not guaranteed. It targets wireless communication and low- delay (interactive) video applications such as video conferencing. Main Profile This profile focuses on high compression efficiency. The target

applications are in the domain of entertainment such as digital video broadcast and storage (e.g., on next-generation DVDs).

Extended Profile The Extended Profile is defined to be used in streaming sce- narios as it adds tools for an enhanced coding efficiency and for better network robustness to the set of tools of the Baseline Profile. In fact, the Extended Profile is a superset of the Baseline Profile.

The relationship between these three profiles and the coding tools of H.264/AVC is shown in Figure A.4.

The first amendment of H.264/AVC (FRExt) defines four additional profiles which are all supersets of the Main Profile. These four profiles are the High Profile, the High 10 Profile, the High 4:2:2 Profile, and the High 4:4:4 Profile. More information about these profiles and the tools they include can be found in [133, 134].