0x1 解码流程介绍

解码流程说明如下

0x11 tile_group_obu

对tile group的每个tile调用decode_tile()进行解码，直至tile group中包括的tile都解码完成.

for ( TileNum = tg_start; TileNum <= tg_end; TileNum++ ) {
    tileRow = TileNum / TileCols
    tileCol = TileNum % TileCols
    ......
    decode_tile( )
    exit_symbol( )
}

0x12 decode_tile

对tile中包含的每个partition进行解码。

for ( r = MiRowStart; r < MiRowEnd; r += sbSize4 ) {
    for ( c = MiColStart; c < MiColEnd; c += sbSize4 ) {
        ......
        decode_partition( r, c, sbSize )
    }
}

0x13 decode_partition

每个partition的解码过程就是对一个super block(super block的定义参考0x2)进行解码的过程.
根据partition类型对进一步split, 不能split的类型调用decode_block().

partition类型的定义如下

| Partition index | Type of partition  |
|:-------:|:------------- |
| 0 |PARTITION_NONE|
| 1 |PARTITION_HORZ|
| 2 |PARTITION_VERT|
| 3 |PARTITION_SPLIT|
| 4 |PARTITION_HORZ_A|
| 5 |PARTITION_HORZ_B|
| 6 |PARTITION_VERT_A|
| 7 |PARTITION_VERT_B|
| 8 |PARTITION_HORZ_4|
| 9 |PARTITION_VERT_4|

if ( partition == PARTITION_NONE ) {
    decode_block( r, c, subSize )
} else if ( partition == PARTITION_HORZ ) {
    decode_block( r, c, subSize )
if ( hasRows )
    decode_block( r + halfBlock4x4, c, subSize )
} else if ( partition == PARTITION_VERT ) {
    decode_block( r, c, subSize )
    if ( hasCols )
       decode_block( r, c + halfBlock4x4, subSize )
} else if ( partition == PARTITION_SPLIT ) {
    decode_partition( r, c, subSize )
    decode_partition( r, c + halfBlock4x4, subSize )
    decode_partition( r + halfBlock4x4, c, subSize )
    decode_partition( r + halfBlock4x4, c + halfBlock4x4, subSize )
} else if ( partition == PARTITION_HORZ_A ) {
    decode_block( r, c, splitSize )
    decode_block( r, c + halfBlock4x4, splitSize )
    decode_block( r + halfBlock4x4, c, subSize )
} else if ( partition == PARTITION_HORZ_B ) {
    decode_block( r, c, subSize )
    decode_block( r + halfBlock4x4, c, splitSize )
    decode_block( r + halfBlock4x4, c + halfBlock4x4, splitSize )
} else if ( partition == PARTITION_VERT_A ) {
    decode_block( r, c, splitSize )
    decode_block( r + halfBlock4x4, c, splitSize )
    decode_block( r, c + halfBlock4x4, subSize )
} else if ( partition == PARTITION_VERT_B ) {
    decode_block( r, c, subSize )
    decode_block( r, c + halfBlock4x4, splitSize )
    decode_block( r + halfBlock4x4, c + halfBlock4x4, splitSize )
} else if ( partition == PARTITION_HORZ_4 ) {
    decode_block( r + quarterBlock4x4 * 0, c, subSize )
    decode_block( r + quarterBlock4x4 * 1, c, subSize )
    decode_block( r + quarterBlock4x4 * 2, c, subSize )
    if ( r + quarterBlock4x4 * 3 < MiRows )
        decode_block( r + quarterBlock4x4 * 3, c, subSize )
} else { //PARTITION_VERT_4
    decode_block( r, c + quarterBlock4x4 * 0, subSize )
    decode_block( r, c + quarterBlock4x4 * 1, subSize )
    decode_block( r, c + quarterBlock4x4 * 2, subSize )
    if ( c + quarterBlock4x4 * 3 < MiCols )
        decode_block( r, c + quarterBlock4x4 * 3, subSize )
}

0x14 decode_block

对一个block进行解码，进一步调用residual对残差进行解码.

0x15 transform_tree

对不同size的transform块进行解码，它是一个递归函数，递归退出函数是transform_block.

transform_tree( startX, startY, w, h ) { Type
    maxX = MiCols * MI_SIZE
    maxY = MiRows * MI_SIZE
    if ( startX >= maxX || startY >= maxY ) {
        return
    }
    row = startY >> MI_SIZE_LOG2
    col = startX >> MI_SIZE_LOG2
    lumaTxSz = InterTxSizes[ row ][ col ]
    lumaW = Tx_Width[ lumaTxSz ]
    lumaH = Tx_Height[ lumaTxSz ]
    if ( w <= lumaW && h <= lumaH ) {
        txSz = find_tx_size( w, h )
        transform_block( 0, startX, startY, txSz, 0, 0 )
    } else {
        if ( w > h ) {
            transform_tree( startX, startY, w/2, h )
            transform_tree( startX + w / 2, startY, w/2, h )
        } else if ( w < h ) {
            transform_tree( startX, startY, w, h/2 )
            transform_tree( startX, startY + h/2, w, h/2 )
        } else {
            transform_tree( startX, startY, w/2, h/2 )
            transform_tree( startX + w/2, startY, w/2, h/2 )
            transform_tree( startX, startY + h/2, w/2, h/2 )
            transform_tree( startX + w/2, startY + h/2, w/2, h/2 )
        }
    }
}

0x16 Reconstruct中包括具体的反量化，反dct,重建像素和loop filter等过程.

0x2 super block

下图说明了一个frame由许多super block组成，super block的定义稍后说明.

Super block的定义如下
The top level of the block quadtree within a tile. All superblocks within a frame are the same size and are square. The
superblocks may be 128x128 luma samples or 64x64 luma samples. A superblock may contain 1 or 2 or 4 mode info
blocks, or may be bisected in each direction to create 4 sub-blocks, which may themselves be further subpartitioned,
forming the block quadtree.

通过在sequence_header_obu中的use_128x128_superblock来控制superblock的大小是128x128还是64x64.
sequence_header_obu( ) {
……
use_128x128_superblock f(1)
……
}

use_128x128_superblock的定义如下
when equal to 1, indicates that superblocks contain 128x128 luma samples. When equal to 0,
it indicates that superblocks contain 64x64 luma samples.