ISO/IEC 14496-2:1999/Cor 2:2001
(Corrigendum)Information technology — Coding of audio-visual objects — Part 2: Visual — Technical Corrigendum 2
Information technology — Coding of audio-visual objects — Part 2: Visual — Technical Corrigendum 2
Technologies de l'information — Codage des objets audiovisuels — Partie 2: Codage visuel — Rectificatif technique 2
General Information
Relations
Standards Content (Sample)
INTERNATIONAL STANDARD ISO/IEC 14496-2:1999
TECHNICAL CORRIGENDUM 2
Published 2001-02-15
INTERNATIONAL ORGANIZATION FOR STANDARDIZATION � МЕЖДУНАРОДНАЯОРГАНИЗАЦИЯПОСТАНДАРТИЗАЦИИ � ORGANISATION INTERNATIONALE DE NORMALISATION
INTERNATIONAL ELECTROTECHNICAL COMMISSION � МЕЖДУНАРОДНАЯ ЭЛЕКТРОТЕХНИЧЕСКАЯ КОМИССИЯ � COMMISSION ÉLECTROTECHNIQUE INTERNATIONALE
Information technology — Coding of audio-visual objects —
Part 2:
Visual
TECHNICAL CORRIGENDUM 2
Technologies de l'information — Codage des objets audiovisuels —
Partie 2: Codage visuel
RECTIFICATIF TECHNIQUE 2
Technical Corrigendum 2 to International Standard ISO/IEC 14496-2:1999 was prepared by Joint Technical
Committee ISO/IEC JTC 1, Information technology, Subcommittee SC 29, Coding of audio, picture, multimedia and
hypermedia information.
ICS 35.040 Ref. No. ISO/IEC 14496-2:1999/Cor.2:2001(E)
© ISO/IEC 2001 – All rights reserved
Printed in Switzerland
---------------------- Page: 1 ----------------------
ISO/IEC 14496-2:1999/Cor.2:2001(E)
In clause 3 Definitions, add the following definition with appropriate numbering in alphabetical order
"
3.XXX reference layer: A layer to be referenced for prediction in a scalable hierarchy. The video_object_id of the
reference layer should be the same value as the video_object_id of the enhancement layer. The ref_layer_id of the
enhancement layer is set to the same value as the video_object_layer_id of the reference layer.
"
Replace subclause 6.1.3.5 I-VOPs and group of VOPs with
"
6.1.3.5 I-VOPs and group of VOPs
I-VOPs are intended to assist random access into the sequence. Applications requiring random access, fast-
forward playback, or fast reverse playback may use I-VOPs relatively frequently.
I-VOPs may also be used at scene cuts or other cases where motion compensation is ineffective.
Group of VOP (GOV) header is an optional header that can be used immediately before a coded I-VOP to indicate
to the decoder:
1) the modulo part (i.e. the full second units) of the time base for the next VOP to be displayed after having
decoded a GOV header
2) if the first consecutive B-VOPs immediately following the coded I-VOP can be reconstructed properly in
the case of a random access.
In a non scalable bitstream or the base layer of a scalable bitstream, the first coded VOP following a GOV header
shall be a coded I-VOP.
"
In subclause 6.2.1 Start code, replace
"
1. Configuration information
a. Global configuration information, referring to the whole group of visual objects that will be simultaneously
decoded and composited by a decoder (VisualObjectSequence()).
b. Object configuration information, referring to a single visual object (VO). This is associated with
VisualObject().
c. Object layer configuration information, referring to a single layer of a single visual object (VOL)
VisualObjectLayer()
"
with
"
1. Configuration information
a. Global configuration information, referring to the whole group of visual objects that will be simultaneously
decoded and composited by a decoder (VisualObjectSequence()).
b. Object configuration information, referring to a single visual object (VO). This is associated with
VisualObject().
c. Object layer configuration information, referring to a single layer of a single visual object (VOL)
VideoObjectLayer().
"
2 © ISO/IEC 2001 – All rights reserved
---------------------- Page: 2 ----------------------
ISO/IEC 14496-2:1999/Cor.2:2001(E)
In subclause 6.2.1 Start codes, replace the following row of Table 6-3
"
Reserved C3 – C5
"
with
"
Stuffing_start_code C3
Reserved C4 – C5
"
In subclause 6.2.3 VideoObjectLayer, replace the following rows
"
…
scalability 1bslbf
if (scalability) {
hierarchy_type 1bslbf
ref_layer_id 4uimsbf
ref_layer_sampling_direc 1bslbf
5uimsbf
hor_sampling_factor_n
hor_sampling_factor_m 5uimsbf
vert_sampling_factor_n 5uimsbf
5uimsbf
vert_sampling_factor_m
enhancement_type 1bslbf
if(video_object_layer == “binary” &&
hierarchy_type== ‘0’){
use_ref_shape 1bslbf
use_ref_texture 1bslbf
shape_hor_sampling_factor_n 5uimsbf
shape_hor_sampling_factor_m 5uimsbf
shape_vert_sampling_factor_n 5uimsbf
shape_vert_sampling_factor_m 5uimsbf
}
}
}
else {
if(video_object_layer_verid !=”0001”){
scalability 1bslbf
if(scalability) {
shape_hor_sampling_factor_n 5uimsbf
shape_hor_sampling_factor_m 5uimsbf
shape_vert_sampling_factor_n 5uimsbf
shape_vert_sampling_factor_m 5uimsbf
}
}
1bslbf
resync_marker_disable
}
"
© ISO/IEC 2001 – All rights reserved 3
---------------------- Page: 3 ----------------------
ISO/IEC 14496-2:1999/Cor.2:2001(E)
with
"
…
scalability 1bslbf
if (scalability) {
hierarchy_type 1bslbf
ref_layer_id 4uimsbf
ref_layer_sampling_direc 1bslbf
5uimsbf
hor_sampling_factor_n
hor_sampling_factor_m 5uimsbf
vert_sampling_factor_n 5uimsbf
5uimsbf
vert_sampling_factor_m
enhancement_type 1bslbf
if(video_object_layer == “binary” &&
hierarchy_type== ‘0’){
use_ref_shape 1bslbf
use_ref_texture 1bslbf
shape_hor_sampling_factor_n 5uimsbf
shape_hor_sampling_factor_m 5uimsbf
shape_vert_sampling_factor_n 5uimsbf
shape_vert_sampling_factor_m 5uimsbf
}
}
}
else {
if(video_object_layer_verid !=”0001”){
scalability 1bslbf
if(scalability) {
ref_layer_id 4uimsbf
shape_hor_sampling_factor_n 5uimsbf
shape_hor_sampling_factor_m 5uimsbf
shape_vert_sampling_factor_n 5uimsbf
shape_vert_sampling_factor_m 5uimsbf
}
}
resync_marker_disable 1bslbf
}
"
4 © ISO/IEC 2001 – All rights reserved
---------------------- Page: 4 ----------------------
ISO/IEC 14496-2:1999/Cor.2:2001(E)
In subclause 6.2.3 Video Object Layer, replace the following rows
"
do {
if (next_bits() == group_of_vop_start_code)
Group_of_VideoObjectPlane()
VideoObjectPlane()
} while ((next_bits() == group_of_vop_start_code) ||
(next_bits() == vop_start_code))
}else{
short_video_header = 1
do {
video_plane_with_short_header()
} while(next_bits() == short_video_start_marker)
}
}
"
with
"
do {
if (next_bits() == group_of_vop_start_code)
Group_of_VideoObjectPlane()
VideoObjectPlane()
if ((preceding_vop_coding_type == "B" ||
preceding_vop_coding_type == "S" ||
video_object_layer_shape != "rectangular") &&
next_bits() == stuffing_start_code) {
stuffing_start_code 32 bslbf
while (next_bits() != ‘0000 0000 0000 0000 0000 0001’)
stuffing_byte 8bslbf
}
} while ((next_bits() == group_of_vop_start_code) ||
(next_bits() == vop_start_code))
}else{
short_video_header = 1
do {
video_plane_with_short_header()
} while(next_bits() == short_video_start_marker)
}
}
NOTE — preceding_vop_coding_type has the same value as vop_coding_type in the immediately preceding
VideoObjectPlane() in the decoding order.
"
© ISO/IEC 2001 – All rights reserved 5
---------------------- Page: 5 ----------------------
ISO/IEC 14496-2:1999/Cor.2:2001(E)
In subclause 6.2.5.1 Complexity Estimation Header, replace the last 8 rows of the
read_vop_complexity_estimation_header() syntax
"
if (npm) dcecs_npm 8uimsbf
if (forw_back_mc_q) dcecs_forw_back_q 8uimsbf
if (halfpel2) dcecs_halfpel2 8uimsbf
if (halfpel4) dcecs_halfpel4 8uimsbf
if (interpolate_mc_q) dcecs_interpolate_mc_q 8uimsbf
}
}
}
"
with
"
if (npm) dcecs_npm 8uimsbf
if (forw_back_mc_q) dcecs_forw_back_mc_q 8uimsbf
if (halfpel2) dcecs_halfpel2 8uimsbf
if (halfpel4) dcecs_halfpel4 8uimsbf
if (interpolate_mc_q) dcecs_interpolate_mc_q 8uimsbf
}
}
}
"
In subclause 6.3.3, replace the semantics of use_ref_shape with
"
use_ref_shape: This is one bit flag which indicate procedure to decode binary shape for spatial scalability. If it is
set to ‘0’, scalable shape coding should be used. If it is set to ‘1’ and enhancement_type is set to ‘0’, no shape data
is decoded and up-sampled binary shape of reference_layer should be used for enhancement layer. If
enhancement_type is set to ‘1’ and this flag is set to ‘1’,binary shape of enhancement layer should be decoded as
the same non-scalable decoding process. When video_object_layer_verid == ‘0001’, the value of use_ref_shape is
set to ‘1’.
"
In subclause 6.3.3, replace the semantics of use_ref_texture with
"
use_ref_texture: Reserved flag for future extension. This flag shall be 0 in the case of video_object_layer_ver_id
is "0001" or "0010".
"
6 © ISO/IEC 2001 – All rights reserved
---------------------- Page: 6 ----------------------
ISO/IEC 14496-2:1999/Cor.2:2001(E)
In subclause 6.3.3 Video Object Layer, add the following paragraphs at the end of the subclause
"
stuffing_start_code: This is the bit string ‘000001C3’ in hexadecimal. It is used in conjunction with possibly
following stuffing_byte(s) for the purpose of stuffing bits to guaranty the VBV buffer regulation.
stuffing_byte: This is the 8-bit string in which the value is ’11111111’.
"
In subclause 6.3.5, replace the semantics of modulo_time_base with
"
modulo_time_base: This value represents the local time base in one second resolution units (1000 milliseconds).
It consists of a number of consecutive ‘1’ followed by a ‘0’.Each ‘1’ represents a duration of one second that have
elapsed. For I-, S(GMC)-, and P-VOPs of a non scalable bitstream and the base layer of a scalable bitstream, the
number of ‘1’s indicate the number of seconds elapsed since the synchronization point marked by time_code of the
previous GOV header or by modulo_time_base of the previously decoded I-, S(GMC)-, or P-VOP, in decoding
order. For B-VOP of a non scalable bitstream and a base layer of a scalable bitstream, the number of ‘1’s indicates
the number of seconds elapsed since the synchronization point marked in the previous GOV header, or I-VOP,
S(GMC)-VOP, or P-VOP, in display order. For I-, P-, or B-VOPs of enhancement layer of scalable bitstream, the
number of ‘1’s indicate the number of seconds elapsed since the synchronization point marked in the previous GOV
header, I-VOP, P-VOP, or B-VOP, in display order.
"
In subclause 6.3.5, replace the semantics of vop_vertical_mc_spatial_ref with
"
vop_vertical_mc_spatial_ref: This is a 13-bit signed integer which specifies, in pixel units, the vertical position of
the top left of the rectangle defined by vertical size of vop_height. The value of vop_vertical_mc_spatial_ref shall be
divisible by two for progressive and divisible by four for interlaced motion compensation. This is used for decoding
and for picture composition.
"
In subclause 6.3.5, replace the semantics of resync_marker with
"
resync_marker: This is a binary string of at least 16 zero’s followed by a one‘0 0000 0000 0000 0001’. For an I-
VOP or a VOP where video_object_layer_shape has the value “binary_only”, the resync marker is 16 zeros followed
by a one. The length of this resync marker is dependent on the value of vop_fcode_forward, for a P-VOP or a
S(GMC)-VOP, and the larger value of either vop_fcode_forward and vop_fcode_backward for a B-VOP. For a P-
VOP and a S(GMC)-VOP, the resync_marker is (15+fcode) zeros followed by a one; for a B-VOP, the
resync_marker is max(15+fcode,17) zeros followed by a one. It is only present when resync_marker_disable flag is
set to ‘0’. A resync marker shall only be located immediately before a macroblock and aligned with a byte
"
© ISO/IEC 2001 – All rights reserved 7
---------------------- Page: 7 ----------------------
ISO/IEC 14496-2:1999/Cor.2:2001(E)
In subclause 6.3.5 Video Object Plane and Video Plane with Short Header, replace the semantics of
header_extension_code with
"
header_extension_code: This is a 1-bit flag which when set to ‘1’ indicates the presence of additional fields in the
header. When header_extension_code is set to ‘1’, modulo_time_base, vop_time_increment and vop_coding_type
are also included in the video packet header. If video_object_layer_shape is not “rectangular”, VOP header fields
used for the shape decoding (vop_width, vop_height, vop_horizontal_mc_spatial_ref, vop_vertical_mc_spatial_ref,
change_conv_ratio_disable and vop_shape_coding_type) are also present. if video_object_layer_shape is not
“binary only”, intra_dc_vlc_thr is also present. Furthermore, if the vop_coding_type is equal to either a P, S or B
VOP, the appropriate fcodes are also present. Additionally, if the current VOP is an S(GMC)-VOP,
sprite_trajectory() is included. And if reduced_resolution_vop_enable is equal to one, vop_reduced_resolution is
also present.
"
In subclause 6.3.6 Macroblock related, replace the semantics of cbpb with
"
cbpb: This is a 3 to 6 bit code representing coded block pattern in B-VOPs, if indicated by modb. Each bit in the
code represents a coded/no coded status of a block; the leftmost bit corresponds to the top left block in the
macroblock. For each non-transparent blocks with coefficients, the corresponding bit in the code is set to ‘1’.In
case no coefficients are coded for all the non-transparent blocks in the macroblock, modb shall be set to the value
indicating cbpb is not present (i.e. modb==‘1’ or ‘01’) and cbpb shall not be included in the bitstream for this
macroblock.
"
In subclause 7.4.1.2 Other coefficients, replace
"
When short_video_header is 0, the variable length code table is different for intra blocks and inter blocks. The most
commonly occurring EVENTs for the luminance and chrominance components of intra blocks in this case are
decoded by referring to Table B-16. The most commonly occurring EVENTs for the luminance and chrominance
components of inter blocks in this case are decoded by referring to Table B-17. The last bit “s” denotes the sign of
level, “0” for positive and “1” for negative. The combinations of (LAST, RUN, LEVEL) not represented in these
tables are decoded as described in subclause 7.4.1.3.
"
with
"
When short_video_header is 0, the variable length code table is different for intra blocks and inter blocks. The most
commonly occurring EVENTs for the luminance and chrominance components of intra blocks in this case are
decoded by referring to the intra columns of Table B-23 when reversible_vlc is set to ‘1’ in I-, P-, or S(GMC)-VOPs,
and by referring to Table B-16, otherwise. The most commonly occurring EVENTs for the luminance and
chrominance components of inter blocks in this case are decoded by referring to the inter columns of Table B-23
when reversible_vlc is set to ‘1’ in I-, P-, or S(GMC)-VOPs, and by referring to Table B-17, otherwise. The last bit “s”
denotes the sign of level, “0” for positive and “1” for negative. The combinations of (LAST, RUN, LEVEL) not
represented in these tables are decoded as described in subclause 7.4.1.3.
"
8 © ISO/IEC 2001 – All rights reserved
---------------------- Page: 8 ----------------------
ISO/IEC 14496-2:1999/Cor.2:2001(E)
In subclause 7.4.1.3 Escape code, replace
"
Many possible EVENTS have no variable length code to represent them. In order to encode these statistically rare
combinations an Escape Coding method is used. The escape codes of DCT coefficients are encoded in five modes.
The first three of these modes are used when short_video_header is 0 and in the case that the reversible VLC
tables are not used, and the fourth is used when short_video_header is 1. In the case that the reversible VLC tables
are used, the fifth escape coding method as in Table B-23 is used. Their decoding process is specified below.
"
with
"
Many possible EVENTS have no variable length code to represent them. In order to encode these statistically rare
combinations an Escape Coding method is used. The escape codes of DCT coefficients are encoded in five modes.
The first three of these modes are used when short_video_header is 0 and in the case that the reversible VLC
tables are not used, and the fourth is used when short_video_header is 1. In the case that the reversible VLC tables
are used, the fifth escape coding method as in Table B-23 is used. Use of escape sequence of the reversible VLC
(Table B-24 and Table B-25) for encoding the combinations listed in Table B-23 is prohibited. Their decoding
process is specified below.
"
In subclause 7.4.4.6 Summary of quantiser process for method 1, replace
"
for (v=0; v<8;v++) {
for (u=0; u<8;u++) {
if (QF[v][u] == 0)
F’’[v][u] = 0;
else if ( (u==0) && (v==0) && (macroblock_intra) ) {
F''[v][u]= dc_scaler * QF[v][u];
}else{
if ( macroblock_intra ) {
F''[v][u]=( QF[v][u]* W[0][v][u]* quantiser_scale * 2 ) / 32;
}else{
F''[v][u]=((( QF[v][u]* 2) + Sign(QF[v][u]))* W[1][v][u]
* quantiser_scale ) / 32;
}
}
}
}
"
with
"
for (v=0; v<8;v++) {
for (u=0; u<8;u++) {
if (QF[v][u] == 0)
F’’[v][u] = 0;
else if ( (u==0) && (v==0) && (macroblock_intra) ) {
F''[v][u]= dc_scaler * QF[v][u];
}else{
© ISO/IEC 2001 – All rights reserved 9
---------------------- Page: 9 ----------------------
ISO/IEC 14496-2:1999/Cor.2:2001(E)
if ( macroblock_intra ) {
F''[v][u]=( QF[v][u]* W[0][v][u]* quantiser_scale * 2 ) / 16;
}else{
F''[v][u]=((( QF[v][u]* 2) + Sign(QF[v][u]))* W[1][v][u]
* quantiser_scale ) / 16;
}
}
}
}
"
In subclause 7.5.2.1.2 P- and B-, and S(GMC)-VOPs, replace
"
The decoding of the current bab_type is dependent on the bab_type of the co-located bab in the reference VOP.
The reference VOP is either a forward reference VOP or a backward reference VOP. The forward reference VOP is
defined as the most recent non-empty (i.e. vop_coded != 0 ) I- or P-, or S(GMC)-VOP in the past, while the
backward VOP is defined as the most recently decoded I- or P-, or S(GMC)-VOP in the future. If the current VOP is
a P-, or S(GMC)-VOP, the forward reference VOP is selected as the reference VOP. If the current VOP is a B-VOP
the following decision rules are applied:
1. If one of the reference VOPs is empty, the non-empty one (forward/backward) is selected as the reference VOP
for the current B-VOP.
2. If both reference VOPs are non-empty, the forward reference VOP is selected if its temporal distance to the
current B-VOP is not larger than that of the backward reference VOP, otherwise, the backward one is chosen.
"
with
"
The decoding of the current bab_type is dependent on the bab_type of the co-located bab in the reference VOP.
The reference VOP is either a forward reference VOP or a backward reference VOP. The forward reference VOP is
defined as the most recent non-empty (i.e. vop_coded != 0 ) I- or P-, or S(GMC)-VOP in the past, while the
backward VOP is defined as the most recently decoded I- or P-, or S(GMC)-VOP in the future. If the current VOP is
a P-, or S(GMC)-VOP, the forward reference VOP is selected as the reference VOP. If the current VOP is a B-VOP
the following decision rules are applied:
1. If the backwards reference VOPs is empty, the non-empty one (forward) is selected as the reference VOP for the
current B-VOP.
2. If both reference VOPs are non-empty, the forward reference VOP is selected if its temporal distance to the
current B-VOP is not larger than that of the backward reference VOP, otherwise, the backward one is chosen.
"
In subclause 7.5.2.4 Motion compensation, replace
"
For inter mode babs (bab_type = 0,1,5 or 6), motion compensation is carried out by simple MV displacement
according to the MVs.
Specifically, when bab_type is equal to 0 or 1 i.e. for the no-update modes, a displaced block of 16x16 pixels is
copied from the binary alpha map of the previously decoded I or P-, or S(GMC)- VOP for which vop_coded is not
equal to ‘0’. When the bab_type is equal to 5 or 6 i.e. when interCAE decoding is required, then the pixels
immediately bordering the displaced block (to the left, right, top and bottom) are also copied from the most recent
10 © ISO/IEC 2001 – All rights reserved
---------------------- Page: 10 ----------------------
ISO/IEC 14496-2:1999/Cor.2:2001(E)
valid reference VOP’s (as defined in subclause 6.3.5) binary alpha map into a temporary shape block of 18x18
pixels size (see Figure 7-12). If the displaced position is outside the bounding rectangle, then these pixels are
assumedtobe “transparent”.
If the current VOP is a B-VOP the following decision rules are applied:
� If one of the reference VOPs is empty (i.e. VOP_coded is 0), the non-empty one (forward/backward) is selected
as the reference VOP for the current B-VOP.
� If both reference VOPs are non-empty, the forward reference VOP is selected if its temporal distance to the
current B-VOP is not larger than that of the backward reference VOP, otherwise, the backward one is chosen.
"
with
"
For inter mode babs (bab_type = 0,1,5 or 6), motion compensation is carried out by simple MV displacement
according to the MVs.
Specifically, when bab_type is equal to 0 or 1 i.e. for the no-update modes, a displaced block of 16x16 pixels is
copied from the binary alpha map of the previously decoded I or P-, or S(GMC)- VOP for which vop_coded is not
equal to ‘0’. When the bab_type is equal to 5 or 6 i.e. when interCAE decoding is required, then the pixels
immediately bordering the displaced block (to the left, right, top and bottom) are also copied from the most recent
valid reference VOP’s (as defined in subclause 6.3.5) binary alpha map into a temporary shape block of 18x18
pixels size (see Figure 7-12). If the displaced position is outside the bounding rectangle, then these pixels are
assumedtobe “transparent”.
If the current VOP is a B-VOP the following decision rules are applied:
� If the backwards reference VOPs is empty, the non-empty one (forward) is selected as the reference VOP for
the current B-VOP.
� If both reference VOPs are non-empty, the forward reference VOP is selected if its temporal distance to the
current B-VOP is not larger than that of the backward reference VOP, otherwise, the backward one is chosen.
"
Replace subclause 7.5.4.2 Decoding of enhancement layer with
"
7.5.4.2 Decoding of enhancement layer
When spatial scalability is enabled (scalability is set to 1 and hierarchy_type is set to 0) with enhancement_type ==
0 or When spatial scalability is enabled with enhancement_type == 1 and use_ref_shape == 0, scalable shape
coding process is used for decoding of binary shape.
If spatial scalability is enabled, use_ref_shape is set to 1 and enhancement_type is set to 1, the same non-scalable
decoding process is applied for binary shape of enhancement layer. In this case, the following rules are applied for
enhancement layer.
1. In PVOP, Inter shape coding should be done as bab_type of co-located MB in the reference VOP (lower layer)
is “Opaque”.
2. In BVOP, forward reference VOP, most recently decoded non-empty VOP in the same layer, is always
selected as reference VOP of shape coding.
© ISO/IEC 2001 – All rights reserved 11
---------------------- Page: 11 ----------------------
ISO/IEC 14496-2:1999/Cor.2:2001(E)
If spatial scalability is enabled, use_ref_shape is set to 1 and enhancement_type is set to 0, then the up-sampled
binary shape from reference layer is used for the binary shape of enhancement layer. The up sampling and down
sampling process of this purpose also follows up-down sampling method described in the subclause 7.5.4.4.
When spatial scalability is enabled and enhancement_type is set to 0, in the enhancement layer, the forward
prediction in P-VOP and the backward prediction in B-VOP are used as the spatial prediction. In that case the
shape information is coded by scan interleaving (SI) based method. For the forward prediction in B-VOP a motion
compensated temporal prediction is made from the reference VOP in the enhancement layer. In that case the
shape information is coded by the CAE method as like in base layer except that the shape motion vectors are
obtained from those of the collocated bab in the lower layer. Motion vector and shape coding mode(bab_type) of
collocated bab in the lower layer are used for decoding the enhancement layer bab.
The location of collocated bab in the lower layer can be found by following method.
collocated_MBX
= min ( max ( 0, current_MBX*shape_hor_sampling_factor_m/shape_hor_sampling_factor_n ),
NumMBXLower-1 );
collocated_MBY
= min ( max( 0, current_MBY*shape_ver_sampling_factor_m/shap_ever_sampling_factor_n ),
NumMBYLower-1);
For the current MB location [current_MBX, current_MBY], the location of collocated bab in the reference layer is
denoted as [collocated_MBX, collocated_MBY]. current_MBX, current_MBY, collocated_MBX and collocated_MBY
are the MB-unit coordinations. NumMBXLower and NumMBYLower denote the number of micro-blocks in the lower
layer VOP on horizontal and vertical directions, respectively.
"
In subclause 7.5.4.4 Spatial prediction, replace
"
The spatial prediction is made by resampling the lower reference layer reconstructed VOP to the same sampling
grid as the enhancement layer. For the resampling, repetition is used on the the lower layer.
For enhancement layer encoding/decoding, the base layer VOP should be up-sampled as the sampling ratio, which
is included in the VOL syntax. In VOL syntax for enhancement layer, there are four fields for the up-sampling ratio,
i.e., shape_hor_sampling_factor_n, shape_hor_sampling_factor_m, shape_vert_sampling_factor_n and
shape_vert_sampling_factor_n.
"
with
"
The spatial prediction is made by resampling the lower reference layer reconstructed VOP to the same sampling
grid as the enhancement layer. For the resampling, repetition is used on the lower layer.
For enhancement layer encoding/decoding, the reference layer VOP should be up-sampled as the sampling ratio,
which is included in the VOL syntax. In VOL syntax for enhancement layer, there are four fields for the up-sampling
ratio, i.e., shape_hor_sampling_factor_n, shape_hor_sampling_factor_m, shape_vert_sampling_factor_n and
shape_vert_sampling_factor_n.
"
12 © ISO/IEC 2001 – All rights reserved
---------------------- Page: 12 ----------------------
ISO/IEC 14496-2:1999/Cor.2:2001(E)
In subclause 7.5.4.8 Intra coded enhancement layer decoding, replace
"
Intra coded enhancement layer decoding uses scan interleaving algorithm before performing intra-mode CAE. The
decoding order with SI scanning is as follows:
1. Copy B from base layer
2. Decoding order with Vertical scanning : Vr --> Vp1 --> Vp2 --> . --> Vpk
3. Decoding order with Horizontal scanning : Hr --> Hp1 --> Hp2 --> . --> Hpl
where,
B : Piexl that can be copied from collocated pixel in the base layer.
"
with
"
Intra coded enhancement layer decoding uses scan interleaving algorithm before performing intra-mode CAE. The
decoding order with SI scanning is as follows:
1. Copy B from reference layer
2. Decoding order with Vertical scanning : Vr --> Vp1 --> Vp2 --> . --> Vpk
3. Decoding order with Horizontal scanning : Hr --> Hp1 --> Hp2 --> . --> Hpl
where,
B : Piexl that can be copied from collocated pixel in the reference layer.
"
In subclause 7.6.3, replace
"
if(quarter_pel==1)
”
with
“
if(quarter_sample==1)
”
In subclause 7.6.3, add immediately following the pseudocode
"
The value of mv_data (i.e., horizontal_mv_data and vertical_mv_data) is equal to two times the value found in the
'vector differences' column of Table B-12 associated with the received codeword.
"
© ISO/IEC 2001 – All rights reserved 13
---------------------- Page: 13 ----------------------
ISO/IEC 14496-2:1999/Cor.2:2001(E)
In subclause 7.6.3, add a footnote attached to the last entry of Table B-12
"
The last code shall not be used when vop_fcode=1.
”
In subclause 7.6.3, replace
"
The parameters in the bitstream shall be such that the components of the reconstructed differential motion vector,
MVDx and MVDy, shall lie in the range [low:high].
"
with
"
The parameters in the bitstream shall be such that the components of the reconstructed differential motion vector,
MVDx and MVDy, shall lie in the range [low:high], at the time of their use in calculating the values of MVx and MVy
(i.e., intermediate values of MVDx and MVDy may occur that are outside the range [low:high]).
"
Replace subclause 7.8.5 Warping with
"
7.8.5 Warping
(i ,j ), G (i ,j )) are computed as described in the
For any pixel (i, j) inside the VOP boundary, (F(i, j), G(i, j)) and (F
c c c c c c
following. These quantities are then used for sample reconstruction as specified in subclause 7.8.6. The following
notations are used to simplify the description:
I=i- i ,
0
J=j- j ,
0
I = 4 i - 2 i + 1,
c c 0
J = 4 j - 2 j + 1,
c c 0
When no_of_sprite_warping_point == 0,
(F(i, j), G(i, j)) = (si,s j),
(F (i ,j ), G (i ,j )) = (si ,s j ).
c c c c c c c c
When no_of_sprite_warping_point == 1 and sprite_enable == ‘static,
(F(i, j), G(i, j)) = (i ’+sI, j ’+s J),
0 0
(F (i ,j ), G (i ,j )) = i ’ /// 2 + s (i � i /2),j ’ /// 2 + s ( j � j / 2)).
c c c c c c 0 c 0 0 c 0
When no_of_sprite_warping_point == 1 and sprite_enable == ‘GMC’,
(F(i, j), G(i, j)) = (i ’+sI, j ’+s J),
0 0
(F (i ,j ), G (i ,j )) = (((i ’ >> 1) | (i ’&1)) + s (i � i /2), ((j ’ >> 1) | (j ’&1)) + s ( j � j /2))
c c c c c c 0 0 c 0 0 0 c 0
14 © ISO/IEC 2001 – All rights reserved
---------------------- Page: 14 ----------------------
ISO/IEC 14496-2:1999/Cor.2:2001(E)
When no_of_sprite_warping_points == 2,
(F(i, j), G(i, j)) = ( i ’+((�ri ’ + i ’’) I +(rj ’� j ’’) J) /// (W’ r),
0 0 1 0 1
j ’+((�rj ’ + j ’’) I +(�ri ’ + i ’’) J) /// (W’ r)),
0 0 1 0 1
(F (i ,j ), G (i ,j )) = (((�ri ’ + i ’’) I +(rj ’� j ’’) J +2 W’ri ’� 16W’) /// (4 W’ r),
c c c c c c 0 1 c 0 1 c 0
((�rj ’ + j ’’) I +(�ri ’ + i ’’) J +2 W’rj ’� 16W’) /// (4 W’ r)).
0 1 c 0 1 c 0
� �
According to the definition of W’ and H’ (i.e. W’ =2 and H’ =2 ), the divisions by “///” in these functions can be
replaced by binary shift operations. By this replacement, the above equations can be rewritten as:
�+�-1
(F(i, j), G(i, j)) = ( i ’ + (((�ri ’ + i ’’) I +(rj ’� j ’’) J+ 2 )>>(�+�)) ,
0 0 1 0 1
�+�-1
j ’+(((�rj ’ + j ’’) I +(�ri ’ + i ’’) J+ 2 )>>(�+�)),
0 0 1 0 1
+ +1
� �
(F (i ,j ), G (i ,j )) = (((�ri ’ + i ’’) I +(rj ’� j ’’) J +2 W’ri ’� 16W’ + 2 )>>(�+�+2),
c c c c c c 0 1 c 0 1 c 0
+ +1
� �
((�rj ’ + j ’’) I +(�ri ’ + i ’’) J +2 W’rj ’� 16W’ + 2 )>>(�+�+2)),
0 1 c 0 1 c 0
�
where 2 =r.
When no_of_sprite_warping_points == 3,
(F(i, j), G(i, j)) = (i ’+((�ri ’ + i ’’) H’ I +(�ri ’+ i ’’)W’ J) /// (W’H’r),
0 0 1 0 2
j ’+((�rj ’ + j ’’) H’ I +(�rj ’+ j ’’)W’ J) /// (W’H’r)),
0 0 1 0 2
(F (i ,j ), G (i ,j )) = (((�ri ’
...
Questions, Comments and Discussion
Ask us and Technical Secretary will try to provide an answer. You can facilitate discussion about the standard in here.