&%PAGE&

-  &%page&  -
Annex 1 to Doc. AVC-177R
Annex 1 to Doc. AVC-177R

Documents for the third meeting of the Experts Group  
18-29 December 1991, Yokosuka


Normal Documents

	Note: Contributions with "*" have also been submitted to MPEG.

AVC-106R REPORT OF THE THIRD MEETING OF THE EXPERTS GROUP FOR 
         ATM VIDEO CODING IN SANTA CLARA - PART I (CHAIRMAN)

Achievements and action points obtained at the third meeting (CCITT sole 
sessions) are recorded to facilitate our discussion at this meeting.

AVC-107R REPORT OF THE THIRD MEETING OF THE EXPERTS GROUP FOR 
         ATM VIDEO CODING IN SANTA CLARA - PART II (CHAIRMAN)

Achievements and action points obtained at the third meeting (MPEG/CCITT EG 
joint sessions) are recorded to facilitate our discussion at this meeting.

AVC-108  PROPOSAL PACKAGE DESCRIPTION FOR MPEG PHASE 2 
         (CHAIR OF MPEG/REQUIREMENTS SUB-GROUP)

This is the version updated according to the outcome of the Santa Clara 
meeting in August 1991.

AVC-109  STATUS REPORT ON ATM VIDEO CODING STANDARDIZATION, 
         ISSUE 2 (EXPERTS GROUP)

Study results as well as further study items are summarized for common 
understanding among members of the the group and publicity of the group's 
activities.  This is a version updated according to the Santa Clara meeting 
outcome.  

AVC-110  MEETING REPORT (ITU IVS CO-ORDINATION MEETING)  

IVS co-ordination meeting was held to align the development of network 
standards and that of video coding standards.  Harmonization among video 
coding related standardization was also intended.  As outcome of the 
meeting, standardization timetables were produced, and a number of action 
points together with responsible persons for each point were identified.  

AVC-111  ACTIVITIES OF CCIR SG11 FOR DIGITAL TELEVISION 
         (CHAIRMAN OF WP-11B) 

This document presents CCIR SG11 activities that have been found closely 
related to ours in the IVS coordination meeting.  They are going to work on 
digital emission of HDTV (and other television?) signals, and to study 
generic video coding covering terrestrial and satellite emission as well as 
secondary distribution through cables.  We should be conscious of CCIR SG11 
in addition to MPEG and CMTT/2 for our successful work.

AVC-112  REPORT OF MEETING (WPXVIII/8-5, 8-3)

Outcome of the Rapporteur's meeting on AAL Type 1 and Type 3,4 is reported, 
listing some open items as well.

AVC-113  PICTURE FORMAT FOR CONVERSATIONAL VIDEO APPLICATIONS 
         (CANADA)

Experimental results are provided on the impact of 480 to 576 to 480 lines 
conversion (with 30 Hz frame rate) for 525-line TV systems, concluding that 
using a high-order vertical filtering results in no visible quality 
degradation and that coding efficiency loss due to the use of 576-line 
format is less than 3%.

AVC-114* ATM COMPENDIUM (THE NETHERLANDS, BELGIUM, FRG, FRANCE, 
         SWEDEN, ITALY, GREEK, UK)

Information is provided on a European living document for the area of ATM 
video coding, which is a collection of work carried out in ETSI NA5 VCM and 
Cost 211ter.

AVC-115* FEATURES OF MUPCOS (PTT RESEARCH, THE NETHERLANDS)

Implementation methods and obtained performance are provided for features 
such as MPEG1/H.261 forward/backward compatibility, upward/downward 
compatibility, low end-to-end delay, cell loss resilience, scalability.

AVC-116  BUFFERING AND DELAYS FOR A SLIDING WINDOW CONSTRAINT 
         (AT&T)

This document gives the delay and the encoder and decoder buffer sizes 
that allow any encoded bit-stream conforming to a peak-plus-sliding window 
channel rate constraint to be transmitted and decoded.  The post-coding 
buffer solution is shown as much cheaper in implementation compared to the 
display buffer solution described in AVC-90.

AVC-117* CELL LOSS RESILIENCE IN A TWO-LAYER CODING SCHEME (UK)

Experimental results are reported on the effect of cell loss in a two-
layered scheme that has a guaranteed CBR path for base layer and a cell 
loss susceptible VBR enhancement layer carrying difference between the DCT 
coefficients of the base layer and their quantized values.  It is concluded 
that for cell loss ratios of 1 in 10 the picture quality degradation is 
negligible when 50% of the bits are spent in coding the base layer 
information.

AVC-118* Algorithm proposal (BT)  

AVC-119  LAYERED INTERFRAME CODING SYSTEM BASED ON MOTION 
         QUANTITY (JAPAN)

Information is provided on a H.261 based layered coding scheme where large 
movement blocks are assigned to the high priority cells and the small 
movement blocks to the low priority cells.  Two frame memories are used in 
the prediction loop, one of which stores full video signals including low 
priority blocks while the other stores only the high priority block signals.  
This scheme is characterized by the fact that full-band prediction is used 
at normal state, and that interframe prediction is applied even after cell 
loss happens in the low priority blocks.  Some experimental results are 
presented.


AVC-120* CELL LOSS RECOVERY BY CODING ALGORITHM (JAPAN)

Various cell loss recovery methods are listed with focus on the necessary 
elements in the standard; leaky prediction, layered coding, cyclic refresh, 
demand refresh, error concealment and packetization.

AVC-121  VIDEO CLOCK JUSTIFICATION (JAPAN)

Necessity of video clock transmission for H.26X codecs is pointed out.  Three 
possible methods for this purpose are outlined; use of the network clock as 
a reference, use of temporal reference in the coded video signal and use of 
the AAL function.  It is concluded that this matter should be discussed and 
decided at an early stage. 

AVC-122  COMPARISON BETWEEN COMPATIBLE AND NON-COMPATIBLE CODING 
         (JAPAN)
Simulation results are reported on a hierarchical coding scheme proposed 
for the Kurihama tests (see AVC-150).  This scheme can operate in downward 
compatible mode, upward compatible mode and non-compatible mode.  Coding 
efficiency loss due to compatible coding has been found as 1.6 dB at both 4 
and 9 Mbit/s for Flower Garden. 

AVC-123* INTERWORKING OF H.32X TERMINALS (JAPAN)

Interworking methods of H.32X terminals with various terminals are discussed 
from the bit stream level compatibility and algorithm level compatibility.  
The followings are pointed out;

     -	H.32X terminal must have bit stream level compatibility with H.261.
     -	Conversion between different service classes becomes easier by a 
unified source coding algorithm.
     -	This conversion in video multiplex and/or transmission coding should 
be kept as necessary minimum by proper design.

AVC-124* CONSIDERATIONS ON PROCESSING DELAY WITH HYBRID CODING 
         (JAPAN)
Analysis of processing delay for hybrid codecs is presented to identify the 
most demanding elements that require more than one frame time.  Format 
conversion, frame reordering for bidirectional prediction are those factors.  
Transmission buffer causes another significant delay in CBR, but it can be 
eliminated in VBR.

AVC-125  SCIF ON A SQUARE PIXEL DISPLAY (JAPAN)

Experimental results are given for displaying a 16:15 pel aspect ratio signal 
(720 pels/line x 576 lines) on a square pixel display, concluding that this 6% 
error causes distinguishable geometric distortion unless proper 
compensation is carried out. 

AVC-126  OPTIMIZATION OF 2-D VLC FOR SCIF CODING (JAPAN)

Some experimental results are provided to clarify possibilities of coding 
efficiency improvements through optimizing 2-d VLCs for coding through a 
progressive SCIF (720 pels/line, 576 lines per picture, 59.94 pictures per 
sec).  It is concluded that we can not expect to save coding efficiency loss 
due to the use of progressive SCIF by optimizing the 2-d VLC for INTER 
pictures, but that use of an optimized 2-d VLC for INTRA improves coding 
efficiency by up to 10%.

AVC-127  TRANSITION OF THE BUFFER OCCUPANCY UNDER VBR 
         ENVIRONMENTS (JAPAN)

Experimental results are reported on the leaky bucket occupancy when 89 
frames of "Edited Sequence" are coded with H.261 and rate controlled by a 
certain method (see AVC-89).  Findings are that coding rate control with 
short response time and a low coding rate provide small value of BOCmax and 
that a transmission rate slightly higher than the coding rate makes BOCmax 
quite small.  

AVC-128  COMPARISON BETWEEN SLIDING WINDOW AND LEAKY BUCKET AS A 
         UPC MECHANISM (JAPAN)

This document compares two average rate monitoring methods, "sliding 
window" and "leaky bucket," assuming a simple square wave source model for 
information generation.  It is concluded that "leaky bucket" is a more 
suitable UPC mechanism for video coding at least from transmission 
efficiency.

AVC-129  COMPARISON OF MULTIMEDIA MULTIPLEX (JAPAN)

Three multimedia multiplex methods, cell multiplex (VCI approach), SAR 
multiplex (packet approach) and user multiplex approach (H.221 approach) are 
compared in terms of channel utilization, compatibility with H.320 and MPEG, 
multimedia aspect, cell loss influence, etc.  The logical interface between 
virtual media control layer and upper layer (e.g. video codec) is illustrated 
and necessary functions of the media control layer are listed for each of 
the three approaches.  It is concluded that we need further study to 
definitely adopt the VCI approach.   

AVC-130  REQUIREMENTS FOR SGXVIII (JAPAN)

The following items are listed for request (a,b) and question (c,d);

    a.	video clock transmission, network clock to be provided for the 
terminal, timing information in AAL
    b.	definition of standard DOS for international connections
    c.	UPC mechanism and its standardization
    d.	Negotiation for use of VCI's in multimedia communication

AVC-131  NETWORK MODELS (SWEDEN, BELGIUM, FRANCE, FRG, GREEK, 
         ITALY, THE NETHERLANDS, UK)

The following three ATM network models are presented which provide the 
congestion probability Psat and cell loss ratio CLR;

     -	Method of large deviation (as in the reference model)
     -	Gaussian (Normal Distribution) model
     -	Method of equivalent bursts

The first model gives more pessimistic statistical multiplex gain than the 
other two models that are both based on mean and variance.  It is stated 
that though Psat and CLR are different in 2 orders of magnitude, the number 
of allowable calls is only 5% different.  Some calculation examples for the 
number of allowable calls are also provided. 


AVC-132  POLICING FUNCTION FOR VBR CODING (SWEDEN, BELGIUM, 
         FRANCE, FRG, GREEK, ITALY, THE NETHERLANDS, UK)

It is proposed that the SGXV undertakes some action to actively try to 
specify some policing functions for VBR video coding.

AVC-133  Algorithm proposal (PTT Research) 

AVC-134  PICTURE FORMAT CONSIDERATIONS (AUSTRALIA) 

The concept of flexible spatial resolutions and frame rates as proposed in 
AVC-36 is supported.  A number of questions raised at the previous meeting 
are discussed in detail, leading to the above conclusion.  It is stressed 
that the ability to code sources which have frame rates not related to 59.94 
Hz coder rate, proposed for SCIF, by integer factors, is essential. 

AVC-135  FLEXIBLE LAYERED CODING EFFICIENCY (AUSTRALIA)

A layered coding structure for interworking between different service 
classes, which implements the flexible layering concept, is outlined.  
Several potential sources of inefficiency in layered coders have been 
identified; appropriate motion compensation in different layers, critical 
sampling, incremental signal coding in the higher layer.  Work is currently 
being carried out to remove these sources of inefficiency. 

AVC-136  B-ISDN MULTIMEDIA SERVICE INTERWORKING (AUSTRALIA)

The following three multimedia multiplex methods are compared in terms of 
interworking between H.32X and H.320 terminals for different communication 
scenarios;

     -	User multiplex that multiplexes the multimedia into H.221 frame 
structure
     -	Cell-base multiplex that supports the multimedia with different VCs
     -	Hybrid multiplex that combine features of the user multiplex and the 
cell-base multiplex

It is concluded that the importance of cell-base multiplexing for longer 
term terminal interworking scenarios is strongly identified, together with 
the possible interim, short term role of user and hybrid multiplexing.

AVC-137  B-ISDN NETWORK STANDARDIZATION - ITS IMPACT ON THE 
         DEVELOPMENT OF ATM VIDEO CODING (AUSTRALIA)

After a review on recent activity in defining the scope and intent of B-ISDN 
network standards, particularly staged service support capabilities, 
resource allocation strategies, charging principles and B-ISDN performance, 
it is proposed for the Experts Group to take a similarly staged approach to 
develop appropriate video coding standards.  It is also proposed that SGXV 
should provide an indication of the likely range of required cell loss 
performance values to ensure flexible and efficient coding.  

AVC-138  SECOND PROGRESS REPORT TO WPXV/1 (CHAIRMAN)

There are reported major achievements since February, toward defining 
Recommendation H.26X for high quality video coding in the ATM environments.  
Particular items needing consideration of WPXV/1 are also raised; formal 

aspects of collaboration with other standardization bodies and study on 
system aspects such as multimedia multiplexing. 

AVC-139  TRANSPORT AND ERROR CONCEALMENT FOR MPEG-2 
         (DAVID SARNOFF RESEARCH, THOMSON CONSUMER ELECTRONICS)

A packet-oriented transport format for a specific MPEG-2 algorithm proposal 
(see AVC-159) is described.  The transport format contains "data link level" 
containing generic transport information (such as priority indicator, service 
ID) and "adaptation level" containing video related information fields (such 
as slice/MB pointer, slice/MB number, picture #, picture type) that aids 
error recovery at the video decoder.  The use of HP/LP data types allow 
partitioning subjectively important and less important bit-streams when 
appropriate transmission facilities are available.  Possible error 
concealment methods as well as cell loss simulation results are described.   

AVC-140  STATISTICAL MULTIPLEX GAINS FOR VARIABLE BIT RATE
         (BELGIUM, FRANCE, FRG, GREEK, ITALY, THE NETHERLANDS, 
          SWEDEN, UK)

It is argued that if both provides the same quality, VBR has always larger 
statistical multiplex gain than CBR since VBR is an "eroded" version of CBR.

Several factors impacting the statistical multiplex gain are listed up; codec 
adaptability, noise in source, source bit rate, link bit rate, change in image 
contents, integration periods.  It is concluded that there is no need to 
make a generic choice between CBR and VBR; CBR is a particular case of VBR. 

AVC-141 to 168  Algorithm proposals  

AVC-169  PICTURE FORMAT FOR HIGH QUALITY INTERACTIVE VIDEO 
         SERVICES (BELLCORE)  

Simulation results are presented on a coding efficiency comparison between 
a SCIF with 576x720 spatial resolution and a SCIF with 528x720 spatial 
resolution (square pixel), concluding that the latter format is on average 
5.5% more efficient than the former.

AVC-170  ATM CELL HEADER FIELD (BELLCORE)

Several questions on the ATM cell header field are brought to the Experts 
Group's attention; availability of CLP field to the receiver end user, 
tagging compliant cells at the edge of network, support of user information 
by the existing ATM cell header field, and number of cell loss performance 
objectives.

AVC-171  STATISTICAL ANALYSIS OF VIDEO TELECONFERENCE TRAFFIC 
         (BELLCORE)

Histograms and Q-Q plots are provided for the number of cells per frame in 
"viconf" and "viphone" sequences which were provided by Alcatel Bell (see 
AVC-99), concluding that it follows a gamma distribution.

AVC-172  DSM/ATM WORKSHOP (EXPERTS GROUP)

Collection of the following four talks given during the MPEG/CCITT EG joint 
sessions at JVC-Kurihama on 20 November 1991;

	M. Biggar	Broadband ISDN evolution
	O. Poncin	ATM network characteristics
	T. Tanaka	VBR aspects
	G. Morrison	Video coding techniques suitable for utilizing ATM network 
			characteristics

AVC-173  PRELIMINARY FUNCTIONAL REQUIREMENTS FOR SECONDARY
         DISTRIBUTION OF TV AND HDTV SIGNALS (TG CMTT/2 SRG)

This is an update of the previous document AVC-67.  

AVC-174  REPORT OF THE ACTIVITIES OF THE GROUPS OF EXPERTS 
         ASSISTING THE SPECIAL RAPPORTEUR (TG CMTT/2 SRG)

Work progress is reported covering 4 meetings since May 1991.  It is noted 
that CMTT/2 SRG are confident;

     -	that a family of formats can be defined for 50 or 59.94/60 Hz systems,
     -	that the same algorithm scheme can be adopted for both the families, 
     -	that compatibility between 50 Hz and 59.94 Hz is not a requirement at 
       the decoder.  

Two examples of compatible coding structures are given; both of which use a 
progressive or interlaced format for CTV and EDTV, while compatibility 
between this format and HDTV format is achieved by pyramidal coding.

AVC-175  RESOLUTIONS ADOPTED DURING THE FIRST PLENARY MEETING OF 
         ISO/IEC JTC1/SC29, 21-23 NOVEMBER 1991, IN TOKYO, JAPAN  
         (SECDREATRIAT ISO/IEC JTC1/SC29)

Decisions made at the first meeting of ISO/IEC JTC1/SC29 are brought to the 
Experts Group's attention; acivities of the groups under its umbrella and 
the relation to other groups. 

AVC-176  PICTURE PERFORMANCE OF 1.3, 4 AND 9 MBIT/S BELLCORE 
         ALGORITHMS AND OTHER TEST PROCESSES (BELLCORE, NTT)  

Subjective study results are reported on the picture quality evaluation of 
compressed bit rate algorithms operating at 1.3, 4 and 9 Mbit/s, and the 
picture quality of other benchmark test processes (NTSC, VHS VCR).  The 
subjective test plan is also detailed. 

ALGORITHM PROPOSAL DOCUMENTS SUBMITTED FOR THE KURIHAMA TESTS

	Note: "#nn" indicates the MPEG pre-registration number.

AVC-118  MPEG91/207	#08		BT Labs
AVC-133  MPEG91/215,216	#18,19		PTT Research, LEP-Philips, PRL-Philips
AVC-141  MPEG91/201	#02		AT&T
AVC-142  MPEG91/202	#03		Aware (I)
AVC-143  MPEG91/203	#04		Aware (II)
AVC-144  MPEG91/204	#05		Bellcore, NTT
AVC-145  MPEG91/206	#06		BBC et al.
AVC-146  MPEG91/206	#07		NTA
AVC-147  MPEG91/208	#09		CCETT et al.
AVC-148  MPEG91/209	#11		GCT et al.
AVC-149  MPEG91/210	#12		HHI
AVC-150  MPEG91/211	#13		Hitachi, Fujitsu
AVC-151  MPEG91/212	#14		IBM
AVC-152  MPEG91/213	#16		JVC
AVC-153  MPEG91/214	#17		KDD
AVC-154  MPEG91/217	#22		Matsushita
AVC-155  MPEG91/218	#23		MIT
AVC-156  MPEG91/219	#24		Mitsubishi
AVC-157  MPEG91/220	#25		NEC et al.
AVC-158  MPEG91/221	#26		NHK
AVC-159  MPEG91/223,255	#27,28,35	TCE, PCE, David Sarnoff, PTT Research
AVC-160  MPEG91/224	#30		RTT, UCL
AVC-161  MPEG91/225	#31		Sharp
AVC-162  MPEG91/226	#33		SONY
AVC-163  MPEG91/227	#34 		Thomson-CSF/LER
AVC-164  MPEG91/228	#36 		Toshiba
AVC-165  MPEG91/229	#37		UCL
AVC-166  MPEG91/230	#38		Waseda University
AVC-168  MPEG91/231	#39		Columbia University
AVC-151  MPEG91/232	#40		EPFL


Temporary Documents

TD-1	Agenda for the fourth meeting in Yokosuka (Chairman)
TD-2	Available documents (Chairman)
TD-3	List of tape demonstrations (Chairman)
tD-4	Summary of proposals (MPEG/VIDEO and REQUIREMENTS)
TD-5	Subjective test results of MPEG-2 (MPEG/TEST)
TD-6	Summary of discussion (MPEG/VIDEO and REQUIREMENTS)
TD-7	Meeting report (Chairman, WPXV/1)
TD-8	Guiding principles for the Singapore meeting (Small group) 
TD-9	Definition of end to end delay (Small group)
TD-10	Addendum to liaison statement from CCITT SGXV Experts Group on Video 
Coding for ATM Networks (Small group)
TD-11	Draft report of the fourth meeting in Yokosuka (Chairman)


END