UNIVERSITY OF TEXAS AT ARLINGTON

EE 5359 Spring 2011 Final Report

AVS CHINA,IMPLEMENTATION & PERFORMANCE ANALYSIS OF ALL VIDEO

PROFILES

Instructor: Dr. K. R. Rao

Submitted by: Vamsi Krishna Vegunta

Student ID: 1000627648

Email: vamsikrishna.vegunta@mavs.uta.edu

AVS China Evolution:

High technology license fees have kept the Chinese consumer electronics industry in chains. A new audio and video compression standard will set it free. In March of 2002 video and audio coding experts from the Chinese Academy of Sciences and from a number of Chinese universities and manufacturers met with officials of the National Information Industry Ministry in Beijing to set up a Chinese national standard for audio and video coding. The standard was to be used in digital disc players, digital TV, Internet Protocol television (IPTV), satellite TV, mobile video phones, and other applications. Some of the experts suggested that China develop a standard independent of all foreign technologies, but it became clear that this was not practical. Researchers around the world have been working on audio and video coding for a long time, and their legacy technologies are associated with intellectual property that China could not do without.

That left the group, officially named the Audio and Video Coding Standard Working Group of China, with a dilemma. How could it develop a national standard that was technically competitive yet affordable—that is, with low license fees?

Five years later, the group unveiled its answer: AVS, the Audio Video Coding Standard of China. For the first time in creating an audio or video compression format, a standards body did not consider just quality and bit rate but also considered the cost of the intellectual property. The group set a price goal of 1 Yuan, or 13 cents, for the audio and video compression technology in each video player; this far undercuts the typical $2.50 license fees for the MPEG-2 compression technology used in standard DVD players today [1].

Color Space:

A color model is an abstract mathematical model describing the way colors can be represented as tuples of numbers, typically as three or four values or color components [2].

In AVS China, color of a pixel is described by two values, one (luma) essentially (but not exactly) describing its luminance (brightness), and one (chrominance). The first value in this scheme does not actually describe the luminance of the pixel’s color. As a result, it is often called “luma”, a term borrowed from the analog system used for television signals.

This term is a tip that the value does not quite describe luminance, because of its nonlinear form. And in fact, the value pair giving the chrominance is also sometimes called “chroma”, again primarily to tip us off to its nonlinear form. But here we will use the term chrominance, as it best matches normal editorial practice for the topic area we are considering. Thus, for each pixel, there are three numerical values that collectively describe its color. They are identified as Y’, , and . Y’ is the luma value, and and collectively form the chrominance value [3].

Chroma Sub sampling:

The human visual system is less sensitive to changes in color than in changes in luminance. Hence Y’ can be stored with high resolution or transmitted at high bandwidth, and two chroma components (CB and CR) that can be bandwidth-reduced, sub sampled, compressed, or otherwise treated separately for improved system efficiency. This leads to a technique called chroma sub sampling. The most popular patterns for sub sampling CR and CB values are

4:2:2

4:2:0

4:4:4 means that the 3 components (Y, Cr and CB) have the same resolution hence a sample of each is present at every pixel location. I.e. for every 4 luma samples there will be 4 CR and CB samples. In 4:4:4 each pixel is represented by 24bits. 4:4:4 sampling preserves the full fidelity of the chrominance components. In 4:2:2 sampling, there will be 2 CR and 2 CB components for every 4 luma samples. In 4:2:2 each pixel is represented by 16bits. 4:2:2 sampling is used for high quality color reproduction. 4:2:0 means that CR and CB each have half the resolution of Y. for every 4 luma samples there will be one CR and CB samples. 4:2:0 sampling is popular in mass market digital video applications like TV broad cast and video conferencing. 4:2:0 is sometimes described as 12 bits per pixel [4].

Fig1. Chroma sub sampling [5]

Data Formats:

1. Progressive Scan: AVS codes video data in progressive scan format. This format is directly compatible with all content that originates in film, and can accept inputs directly from progressive telecine machines. It is also directly compatible with the emerging standard for digital production – the so-called “24p” standard. In the next few years, most movie production and much TV production will be converted to this new standard. It will also be the standard for digital cinema, so there is convergence in the professional film and TV production industry toward a single production format offering the highest original quality. AVS also codes progressive content at higher frame rates. Such rates may be necessary for televised sports. A significant benefit of progressive format is the efficiency with which motion estimation operates. Progressive content can be encoded at significantly lower bitrates than interlaced content with the same perceptual quality. Furthermore, motion compensated coding of progressive format data is significantly less complex than coding of interlaced data. This is a significant component of the reduced complexity of AVS coding.

2. Interlaced Scan: AVS also provides coding tools for interlaced scan format. These tools offer coding of legacy interlaced format video [6].

AVS China Parts:

Parts Content Part 1 System for Broadcasting Part 2 SD/HD Video Part 3 Audio Part 4 Conformance Test Part 5 Reference Software Part 6 Digital Right Management Part 7 Mobility Video Part 8 System over IP Part 9 File Format

Part 10 Mobile speech and audio coding

Table1. Parts of AVS China [4]

Different parts of AVS China are shown in Table 1. Part 2 and Part 7 target the video compression. Part 2 is for high definition video broadcasting and part 7 is for low resolution and low complexity mobile applications. AVS has been designed to give a near optimum performance with reduced complexity. AVS video applications include HDTV, HDDVD and broadband video networking [4].

Profiles, Levels and Applications: Profile is a specified subset of the coding tools. A subset of the coding tools will be implemented in each profile which target specific applications.

AVS-Jiaqiang profile (enhanced profile): To fulfill the needs of multimedia entertainment, one of the major concerns of Jiaqiang profile is movie compression for high-density storage. Relatively higher computational complexity can be tolerated at the encoder side to provide higher video quality, with compatibility to AVS-Part 2 as well [7]. Newly added tools for this profile are

1. Advanced Entropy Coding (a kind of Arithmetic Coder) 2. Weighting Quantization (add a weighting matrix for quantization)

AVS-Shenzhan profile (extended profile): The standard of AVS-Shenzhan focuses exclusively on solutions of standardizing the video surveillance applications. Especially, there are special features of sequences from video surveillance, i.e. the random noise appearing in pictures, relatively lower encoding complexity affordable, and friendliness to events detection and searching required, so corresponding techniques considering a proper process on these special features will be encouraged in the condition of compatibility to AVS-Part 2 [7]. Newly added tools for this profile are

1. Weighting Quantization 2. Constraint intra prediction 3. Slice set: put’s specified slices into a group, slices and slice set are identified by specified id 4. Slice picture header: a error resilience tool that put some essential picture header

parameters to slice header 5. Background picture, core picture, non-reference P picture: those three tools are something

similar as long-term reference picture

AVS-video Jizhun profile (base profile): Jizhun profile is defined as the first profile in the national standard of AVS-Part 2 [7], approved as national standard in 2006, which mainly focuses on digital video applications like commercial broadcasting and storage media, including high-definition applications. Typically, it is preferable for high coding efficiency on video sequences of higher resolutions, at the expense of moderate computational complexity. Jiaqiang and Shenzhan profiles are compatible with Jizhun profile.

AVS-video Jiben profile (basic profile): Jiben profile is defined in AVS-Part 7 [7] targeting to mobility video applications featured with smaller picture resolution. Thus, computational complexity becomes a critical issue. In addition, the ability on error resilience is needed due to the wireless transporting environment. AVS-Part 7 reached to final committee draft at the end of 2004. A "level" is a specified set of constraints indicating a degree of required decoder performance for a profile. For example, a level of support within a profile will specify the maximum picture resolution, frame rate, and bit rate that a decoder may be capable of using. A decoder that conforms to a given

level is required to be capable of decoding all bit streams that are encoded for that level and for all lower levels. The higher the resolution, the higher the level required. Levels supported by various profiles are Jizhun Profile: 2.0.0.08.30, 2.1.0.08.30, 4.0.0.08.30, 4.0.2.08.60, 4.2.0.08.30, 6.0.0.08.60, 6.0.1.08.60, 6.0.3.08.60, 6.0.5.08.60 and 6.2.0.08.60. Shenzhan Profile: 1.0.0.08.30, 2.1.0.08.30, 4.0.0.08.30, 4.0.0.10.30, 4.0.0.12.30 & 6.0.0.08.60. Jiaqiang Profile: 2.0.0.08.30, 2.1.0.08.30, 4.0.0.08.30, 4.0.2.08.60, 6.0.0.08.60, 6.0.1.08.60, 6.0.3.08.60 and 6.0.5.08.60 [8]

Jiben profile: 1.0, 1.1, 1.2, 1.3, 2.0, 2.1, 2.2, 3.0, 3.1 [4]

Profiles Key Applications

Jizhun profile Television broadcasting, HDTV

Jiben profile Mobility applications

Shenzhan profile Video surveillance

Jiaqiang profile Multimedia entertainment

Table2. AVS video profile applications [7]

Each profile targets a particular application depending on the subset and levels implemented in it as

shown in Table 2.

Layered Structure: AVS is built on a layered data structure representing traditional video data. This structure is mirrored in the coded video bit stream. Figure 2 illustrates this layered structure.

Fig2. Layered structure of AVS china [6]

Sequence:

The sequence layer comprises a set of mandatory and optional downloaded system parameters. The mandatory parameters are necessary to initialize decoder systems. The optional parameters can be used for other system settings at the discretion of the network provider. In addition, user data can optionally be contained in the Sequence header. The Sequence layer provides an entry point into the coded video. Sequence headers should be placed in the bit stream to support user access appropriately for the given distribution medium. For example, they should be placed at the start of each chapter on a DVD to facilitate random access. Alternatively they should be placed every ½ second in broadcast TV to facilitate changing channels. Repeat

Sequence headers may be inserted to support random access. Sequences are terminated with a Sequence End Code [6].

Picture: The Picture layer provides the coded representation of a video frame. It comprises a header with mandatory and optional parameters and optionally with user data [6].

Slice: The Slice structure provides the lowest-layer mechanism for resynchronizing the bit stream in case of transmission error. Slices comprise an arbitrary number of raster-ordered rows of Macroblocks as illustrated in the example of Figure 3. Slices must be contiguous, must begin and terminate at the left and right edges of the Picture and must not overlap. It is possible for a single slice to cover the entire Picture. The Slice structure is optional. Slices are independently coded – no slice can refer to another slice during the decoding process.

Fig3. Slice layer example [6]

Macroblock: A Macroblock includes the luminance and chrominance component pixels that collectively represent a 16x16 region of the Picture. In 4:2:0 mode, the Chrominance pixels are subsampled by a factor of two in each dimension; therefore each chrominance component contains only one 8x8 block. In 4:2:2 mode, the Chrominance pixels are subsampled by a factor of two in the horizontal dimension; therefore each chrominance component contains two 8x8 blocks. The Macroblock layer is the primary unit of adaptivity in AVS and the primary unit of motion compensation. The Macroblock header contains information about the coding mode and the motion vectors. It may optionally contain the quantization parameter [6].

Block: The Block is the smallest coded unit and contains the transform coefficient data for the prediction errors. In the case of Intra-coded blocks, Intra prediction is performed from neighboring Blocks [6].

Figure 4 clearly illustrates the layered structure of AVS china.

Fig4. Layered structure example [4]

Types of Pictures: Three types of pictures are defined by AVS. They are

Predicted pictures (P pictures)

Interpolated pictures (B pictures)

Intra pictures (I pictures)

Predicted picture: The forward prediction process is illustrated in Figure5. Prediction of a Macroblock or block in the current picture may be from the most recent reference picture or from the second most recent reference picture. There may be any number of Interpolated pictures in between the current picture and the most recent reference picture. There may be any number of pictures in between the current picture and the second most recent reference picture. Prediction for Interlaced format is illustrated in Figure 6. In I Pictures, the second field may be predicted from the first field. In P-Pictures, prediction of the current field may be made from the four most recent fields. As shown, these fields may be in the current frame, most recent frame or second most recent frame [6].

Fig5. P picture [6]

Fig6. Interlaced P frame prediction [6]

P-prediction uses one motion vector and one reference index to locate the reference block. The motion-compensated prediction includes five macroblock modes, with partitioning down to 8×8 blocks, as listed in Table3.

Macroblock type

P_Skip

P_16*16

P_16*8

P_8*16

P_8*8

Table3. Modes of P picture [7]

Interpolated picture: The interpolated process is shown in Figure 8. A Macroblock or block in the current picture is predicted by the average of the macroblocks or blocks in the most recent and future P-Pictures that are selected by the motion vector. Prediction for interlaced format is illustrated in Figure 9. Two modes are supported for the motion vector selection: Direct mode and Symmetric mode. In direct prediction [9], both forward and backward motion vectors of current block are derived from the motion vector of its collocated block in the backward reference according to the temporal block distance between predicted and reference blocks. In symmetric prediction [10], forward motion vector needs to be transmitted for each partition of current macroblock, while backward motion vector is conducted from the forward motion vector by a symmetric rule as depicted in Fig. 10.

Fig8. Interpolated pictures [6]

Fig9. Interlaced B-Frame prediction [6]

Fig10. Example for symmetric mode [7]

Intra picture: Intra-frame prediction (intra-prediction) uses decoded information in the current frame as the reference of prediction, exploiting statistical spatial dependencies between pixels within a picture [7]. 8 × 8 Intra prediction: The technique of 8 × 8 intra prediction in AVS-video allows five prediction modes, DC, horizontal, vertical, down left and down right, for the luminance component(Fig. 11) and four prediction modes, DC, horizontal, vertical and plane, for chrominance components. Each of the four 8 × 8 luminance blocks can be predicted using one of the five intra-prediction modes. Ahead of prediction of DC mode(Mode2),diagonal down left (Mode3)mode and diagonal downright mode(Mode 4), a three-tap low-pass filter(1,2,1)is applied on the samples that will be used as references of prediction. It needs to be pointed out that in DC mode each pixel of current block is predicted by an average of the vertically and horizontally corresponding reference pixels. Hence, the prediction values of different pixels in a block might be different . This results in a fine prediction for a large block. Prediction of the most

probable mode is according to the intra-prediction modes of neighboring blocks. This will help to reduce average bits needed in describing the intra-prediction mode in video bit stream.

(a) (b) (c) (d) (e)

Fig11. 8 × 8 Intra-prediction modes for Luminance component (a): mode 4: Down right, (b): mode 3: Down left, (c): mode 0: Vertical, (d): mode 1: Horizontal, (e): mode 2: DC [7] 4 × 4 Intra prediction: In lower resolution applications, smaller block size will lead to better coding efficiency, so that AVS-video also defines 4×4 intra prediction. Some specific techniques are working together with 4×4 intra-predictions, such as direction intra-prediction (DIP); padding before prediction (PBP) and simplified chrominance intra-prediction (SCI).Prediction of most probable mode from neighboring blocks is also used. Fig. 12 illustrates all the available directional modes for 4×4 intra-prediction for both luminance component and chrominance components. One flag at macroblock level indicates the use of DIP [11, 12]. If one macroblock is marked as DIP-mode, it infers that each of the 16 luminance 4×4 sub-blocks in this macroblock takes the most probable mode as its intra- prediction mode, even though the intra-prediction mode for each 4×4 sub-block might be different, and no more mode information is transmitted in bit stream. PBP is applied for both luminance and chrominance components, during which the reference pixel r5,r6,r7andr8 are padded from r4,andc5,c6,c7 and c8 are padded from c4, so as to skip conditional test of availability of up-right and down-left reference pixels. SCI means that only DC, vertical and horizontal modes are available for chrominance components [13].

Fig12. Simplified 4×4 intra prediction (Left: directional modes for luminance component; right: directional modes for chrominance components) [7]

Encoder: The AVS video coding standard is based on the classic hybrid DPCM-DCT coder, which was first introduced by Jain and Jain in 1979[15]. Temporal redundancy is removed by motion-compensated DPCM coding. Residual spatial redundancy is removed first by spatial prediction, and finally by transform coding. Statistical redundancy is removed by entropy coding.

Fig13. AVS encoder [6]

These basic coding tools are enhanced by a set of minor coding tools that remove any remaining redundancy, code side information efficiently and provide syntax for the coded bit stream. The algorithm is highly adaptive, since video data statistics are not stationary and because perceptual coding is also used to maximize perceived quality. The adaptivity is applied at both the Picture layer and the Macroblock layer.

The encoder shown in Figure 13 accepts input video and stores multiple frames in a set of frame buffers. These buffers provide the storage and delay required by multi-frame motion estimation. The motion estimation unit can accept original frames from the input buffers or reconstructed coded frames from the forward and backward reference frame stores in the encoder. The motion estimation unit can perform motion estimation in the following ways: Forward prediction from the most recent reference frame

Forward prediction from the second most recent prediction frame

Interpolative prediction between the most recent reference frame and a future reference frame.

Motion estimation produces motion vectors used by the motion compensation unit to produce a forward prediction or interpolated prediction for the current frame. Motion vectors are coded for transmission first by a predictive encoder, and then by entropy encoding. The prediction produced by the motion compensation unit is subtracted from the current frame and the difference signal, i.e., the prediction error, is coded by the DCT and quantization units. In the case of intra-coded macroblocks, the data passes through the intra prediction process to the DCT. The signal is then VLC encoded, formatted with the motion vectors and other side information and stored temporarily in the rate buffer. The signal is also decoded by the inverse quantizer and inverse DCT, and stored in the forward or backward frame buffers for subsequent use in motion compensation. The rate buffer smoothes the variable data rate produced by coding into a constant rate for storage or transmission. A feedback path from the rate buffer controls the quantizer to prevent buffer overflow. A mode decision unit selects the motion compensation mode for pictures and macroblocks.

Decoder: The decoder shown in Figure 14 accepts the constant rate signal from the storage or transmission and stores it temporarily in a rate buffer. The data is read out at a rate demanded by the decoding of each macroblock and picture. The signal is parsed to separate the quantization parameter, motion vectors and other side information from the coded data signal.

Fig14. AVS decoder [6]

The signal is parsed to separate the quantization parameter, motion vectors and other side information from the coded data signal.

The motion vectors are decoded, reconstructed and used by the motion compensation unit to produce a prediction for the current picture. This is added to the reconstructed prediction error to produce the output signal. In the case of intra-coded macroblocks, the data passes from the DCT through the intra prediction process.

Sequences:

Name Resolution Frames/Sec Number of Frames

File size(MB)

Crew 704 × 576 30 100 57.4

City 704 × 576 30 100 58

Bus 352 × 288 30 100 14.5

News 352 × 288 30 100 14.2

Foreman 176 × 144 30 100 3.62

Tempete 176 × 144 30 100 3.62

Surveillance 352 × 288 30 100 14.5

Snow gate 352 × 288 30 100 14.5

Table 4 . List of sequences used in performance evaluation

(a) (b) (c)

(d) (e) (f)

(g) (h) Fig 15. Sequences used in performance evaluation (a): Crew, (b): City, (c): Bus, (d): News, (e): Foreman, (f): Tempete, (g): Surveillance, (h): Snow gate

Results:

QP PSNR-Y MSE-Y SSIM MSSSIM 3-SSIM Compression

ratio

0 61.169 0.05 0.99955 0.9999 0.99995 2 : 1

15 46.614 1.385 0.9875 0.99745 0.99495 7 : 1

31 38.052 9.062 0.9308 0.9778 0.9637 52 : 1

47 31.864 41.520 0.8340 0.9140 0.8362 315 : 1

63 24.365 253.23 0.6742 0.5944 0.4715 1435:1

Table 5. Performance analysis for Jizhun profile (Sequence: Crew)

(a) (b)

0 20 40 60 8020

30

40

50

60

70

QP

PS

NR

0 20 40 60 800

50

100

150

200

250

300

QP

MS

E

(c) (d)

(e)

Fig 16. Performance evaluation plots for Jizhun profile (Sequence: Crew)

QP PSNR-Y MSE-Y SSIM MSSSIM 3-SSIM Compression ratio

0 60.6 0.0532 0.997 0.9999 1 2:1

15 45.36 1.9 0.990 0.9971 0.9952 6:1

31 36.27 15.32 0.946 0.9850 0.9632 50:1

47 29.08 80.22 0.7803 0.9086 0.8040 326:1

63 21.59 450 0.4249 0.4515 0.315 1350:1

Table 6. Performance analysis for Jizhun profile (Sequence: City)

0 20 40 60 800.65

0.7

0.75

0.8

0.85

0.9

0.95

1

QP

SS

IM

0 20 40 60 800.5

0.6

0.7

0.8

0.9

1

QP

MS

SS

IM

0 20 40 60 800.4

0.5

0.6

0.7

0.8

0.9

1

QP

3-S

SIM

(a) (b)

(c) (d)

(e)

Fig 17. Performance evaluation plots for Jizhun profile (Sequence: City)

0 20 40 60 8020

30

40

50

60

70

QP

PS

NR

0 20 40 60 800

100

200

300

400

500

QP

MS

E

0 20 40 60 800.4

0.5

0.6

0.7

0.8

0.9

1

QP

SS

IM

0 20 40 60 800.4

0.5

0.6

0.7

0.8

0.9

1

QP

MS

SS

IM

0 20 40 60 80

0.4

0.5

0.6

0.7

0.8

0.9

1

QP

3-S

SIM

QP PSNR-Y MSE-Y SSIM MSSSIM 3-SSIM Compression ratio

0 52.5647 0.3602 0.99923 0.99986 0.99959 2:1

15 43.1380 3.1570 0.99117 0.99819 0.99524 7:1

31 35.4738 18.437 0.96045 0.99011 0.97453 25:1

47 27.4192 117.80 0.81584 0.93532 0.85520 122:1

63 21.3616 475.24 0.54136 0.72683 0.53317 518:1

Table 7. Performance analysis for Jiaquiang profile (Sequence: Bus)

(a) (b)

(c) (d)

0 20 40 60 8020

25

30

35

40

45

50

55

QP

PS

NR

0 20 40 60 800

100

200

300

400

500

QP

MS

E

0 20 40 60 800.5

0.6

0.7

0.8

0.9

1

QP

SS

IM

0 20 40 60 800.7

0.75

0.8

0.85

0.9

0.95

1

QP

MS

SS

IM

(e)

Fig 18. Performance evaluation plots for Jiaquiang profile (Sequence: Bus)

QP PSNR-Y MSE-Y SSIM MSSSIM 3-SSIM Compression ratio

0 58.8256 0.0852 0.99927 0.99989 0.99977 7:1

15 47.3606 1.1940 0.99022 0.99815 0.99561 40:1

31 41.5042 4.5989 0.97475 0.99396 0.98420 169:1

47 34.3129 24.087 0.92312 0.96673 0.90882 473:1

63 26.8455 134.43 0.80223 0.86389 0.63852 947:1

Table 8. Performance analysis for Jiaquiang profile (Sequence: News)

(a) (b)

0 20 40 60 800.5

0.6

0.7

0.8

0.9

1

QP

3-S

SIM

0 20 40 60 8025

30

35

40

45

50

55

60

QP

PS

NR

0 20 40 60 800

20

40

60

80

100

120

140

QP

MS

E

(c) (d)

(e)

Fig 19. Performance evaluation plots for Jiaquiang profile (Sequence: News)

QP PSNR-Y MSE-Y SSIM MSSSIM 3-SSIM Compression ratio

0 58.0515 0.1018 0.99949 0.99994 0.99982 2:1

15 46.4563 1.4704 0.99285 0.99916 0.99698 4:1

31 37.1725 12.468 0.96006 0.99396 0.97740 12:1

47 29.1086 79.838 0.86077 0.96668 0.87389 39:1

63 20.3233 603.59 0.51872 0.75155 0.44279 402:1

Table 7. Performance analysis for Jiben profile (Sequence: Foreman)

0 20 40 60 800.8

0.85

0.9

0.95

1

QP

SS

IM

0 20 40 60 800.86

0.88

0.9

0.92

0.94

0.96

0.98

1

QP

MS

SS

IM

0 20 40 60 80

0.65

0.7

0.75

0.8

0.85

0.9

0.95

1

QP

3-S

SIM

(a) (b)

(d) (e)

(f)

Fig 20. Performance evaluation plots for Jiben profile (Sequence: Foreman)

0 20 40 60 8020

25

30

35

40

45

50

55

60

QP

PS

NR

0 20 40 60 800

100

200

300

400

500

600

700

QP

MS

E

0 20 40 60 800.5

0.6

0.7

0.8

0.9

1

QP

SS

IM

0 20 40 60 800.75

0.8

0.85

0.9

0.95

1

QP

MS

SS

IM

0 20 40 60 800.4

0.5

0.6

0.7

0.8

0.9

1

QP

3-S

SIM

QP PSNR-Y MSE-Y SSIM MSSSIM 3-SSIM Compression ratio

0 53.3170 0.3029 0.99938 0.99991 0.99963 2:1

15 43.8232 2.6962 0.99558 0.99930 0.99667 5:1

31 33.3007 30.409 0.96311 0.99269 0.96823 24:1

47 24.6740 221.65 0.78055 0.93618 0.78448 145:1

63 17.9758 1036.3 0.30354 0.55060 0.29145 402:1

Table 8. Performance analysis for Jiben profile (Sequence: Tempete)

(a) (b)

(c) (d)

0 20 40 60 8015

20

25

30

35

40

45

50

55

QP

PS

NR

0 20 40 60 800

200

400

600

800

1000

1200

QP

MS

E

0 20 40 60 80

0.4

0.5

0.6

0.7

0.8

0.9

1

QP

SS

IM

0 20 40 60 800.5

0.6

0.7

0.8

0.9

1

QP

MS

SS

IM

(f)

Fig 21. Performance evaluation plots for Jiben profile (Sequence: Tempete)

QP PSNR-Y MSE-Y SSIM MSSSIM 3-SSIM Compression ratio

0 55.4173 0.1867 0.99909 0.99985 0.99967 2:1

15 42.7762 3.4313 0.98147 0.99671 0.99188 9:1

31 36.7841 13.635 0.94137 0.98676 0.96800 199:1

47 29.9413 65.909 0.80113 0.93054 0.82377 1812:1

63 23.7803 272.29 0.55678 0.70113 0.44946 3625:1

Table 9. Performance analysis for Shenzhan profile (Sequence: Surveillance)

(a) (b)

0 20 40 60 800.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

QP

3-S

SIM

0 20 40 60 8020

25

30

35

40

45

50

55

60

QP

PS

NR

0 20 40 60 800

50

100

150

200

250

300

QP

MS

E

(c) (d)

(e)

Fig 22. Performance evaluation plots for Shenzhan profile (Sequence: Surveillance)

QP PSNR-Y MSE-Y SSIM MSSSIM 3-SSIM Compression ratio

0 55.3226 0.1909 0.99919 0.99989 0.99970 2:1

15 43.0736 3.2041 0.98715 0.99798 0.99365 9:1

31 36.3743 14.984 0.95223 0.98996 0.97216 125:1

47 29.5485 72.148 0.82469 0.94822 0.86587 725:1

63 23.1127 317.54 0.55524 0.75196 0.51585 1450:1

Table 10. Performance analysis for Shenzhan profile (Sequence: Snow gate)

0 20 40 60 800.5

0.6

0.7

0.8

0.9

1

QP

SS

IM

0 20 40 60 800.7

0.75

0.8

0.85

0.9

0.95

1

QP

MS

SS

IM

0 20 40 60 800.4

0.5

0.6

0.7

0.8

0.9

1

QP

3-S

SIM

(a) (b)

(c) (d)

(e)

Fig 23. Performance evaluation plots for Shenzhan profile (Sequence: Snow gate)

0 20 40 60 8020

25

30

35

40

45

50

55

60

QP

PS

NR

0 20 40 60 800

50

100

150

200

250

300

350

QP

MS

E

0 20 40 60 800.5

0.6

0.7

0.8

0.9

1

QP

SS

IM

0 20 40 60 800.75

0.8

0.85

0.9

0.95

1

QP

MS

SS

IM

0 20 40 60 800.5

0.6

0.7

0.8

0.9

1

QP

3-S

SIM

Conclusion: All the video profiles in AVS China are implemented. Performance evaluation for each profile is done by choosing sequences specific to each profile based on targeted application. Performance measures chosen for evaluation are PSNR (Peak signal to noise ratio), MSE (Mean squared error), SSIM (Structural similarity index), 3-SSIM (3 component structural similarity index) [29] and MSSSIM (Multi scale structural similarity index) [31]. These performance measures are plotted for different values of QP (Quantization parameter). As QP is increased PSNR, SSIM, MSSSIM, 3-SSIM are decreased and MSE is increased due to the error between original frames and compressed frames.

Further Implementation:

Yidong profile can be implemented and evaluated.

Scene mode which supports day, night and infrared mode can be implemented and evaluated.

References:

1. http://spectrum.ieee.org/consumer-electronics/standards/why-china-wants-its-own-digital-

video-standard : IEEE Spectrum Online Edition about AVS china

2. http://en.wikipedia.org/wiki/Color_space : Color Space

3. http://dougkerr.net/pumpkin/articles/Subsampling.pdf : Chroma sub sampling article.

4. http://www-ee.uta.edu/Dip/Courses/EE5359/index.html : UTA Multimedia processing website

5. http://lea.hamradio.si/~s51kq/subsample.gif : Chroma sub sampling gif.

6. http://www.avs.org.cn/reference/AVS%20NAB%20Paper%20Final03.pdf : AVS Document

7. L. Yu et al, Overview of AVS-video coding standards, Signal Processing: Image Communication,

Vol 24, Issue 4, Special Issue on AVS and its Application, April 2009, Pages 247-262.

8. Information technology-Advanced coding of audio and video-Part 2: Video: AVS Document.

9. X. Ji et al, "New bi-prediction techniques for B pictures coding [video coding]," ICME '04. 2004

IEEE International Conference on Multimedia and Expo, Vol 1, Pages 101- 104, June 2004.

10. X. Ji et al, B-picture coding in AVS video compression standard, Signal Processing Image

Communication, Vol 23,Issue 1,Pages 31-41, Nov 2008.

11. F. Yi et al, An improvement of intra prediction mode coding, AVS Doc. AVS M1456, 2004.

12. Y. Shen et al, A simplified intra prediction method, AVS Doc. AVS-M1419, 2004.

13. L. Xiong, Improvement of chroma intra prediction, AVS Doc. AVS-M1379, 2004.

14. J R Jain, A K Jain, “Interframe Adaptive Data Compression Techniques for Images”, Signal and

Image Processing lab., Dep. Electrical & Computer Eng., UC Davis Tech Rep., Aug 1979.

15. Special issue on 'AVS and its Applications' Signal Processing: Image Communication, vol. 24, pp.

245-344, April 2009.

16. W. Gao et al “Special issue on AVS and its applications: Guest editorial”, Signal Process: Image

Communications, vol. 24, Issue 4, pp 245-344, April 2009.

17. R. Schafer and T. Sikora, “Digital video coding standards and their role in video

communications”, Proc. of the IEEE, vol. 83, pp. 907-924, June 1995.

18. M. Liu and Z. Wei “A fast mode decision algorithm for intra prediction in AVS-M video coding”

Vol 1, International Conference on Wavelet Analysis and Pattern Recognition apos;07, Issue 2-

4, pp.326 –331, Nov. 2007.

19. S.Hu et al, “Efficient Implementation of Interpolation for AVS”, Image and Signal Processing,

2008. Vol 3, pp.133 – 138,May 2008,

20. K.R.Rao and D Ny Kim, "Current video coding standards: H.264/AVC, Dirac, AVS China and VC-

1," System Theory (SSST), 2010 42nd Southeastern Symposium on , vol., no., pp.1-8, 7-9 March

2010

21. X. Jin et al, “Platform-independent MB-based AVS video standard implementation”, Signal

Processing: Image Communication, Vol. 24, Issue 4, pp. 312-323, April 2009.

22. W. Gao et al, “Context-based entropy coding in AVS video coding standard”, Signal Processing:

Image Communication, Vol. 24, Issue 4, pp. 263-276, April 2009.

23. G. Su et al ,“Low-Cost Hardware-Sharing Architecture of Fast 1-D Inverse Transforms for

H.264/AVC and AVS Applications”, IEEE Trans. on Circuits and Systems II: Express Briefs, Vol. 55,

Issue 12, pp.1249 – 1253, Dec. 2008.

24. W. Gao et al , “A Real-Time Full Architecture for AVS Motion Estimation”, IEEE Trans. on

Consumer Electronics, Vol. 53, Issue 4, pp. 1744 –1751, Nov. 2007.

25. B. Tang et al,“ AVS Encoder Performance and Complexity Analysis Based on Mobile Video

Communication”, WRI International conference on Communications and Mobile Computing,

CMC '09, Vol. 3, pp. 102 – 107, . 6-8 Jan. 2009.

26. X. Wang and D. Zhao, “Performance comparison of AVS and H.264/AVC video coding

standards”, Journal of Computer Science & Technology, Vol.21, No.3, pp.310-314, May 2006.

27. X. Zhang and L. Yu, “An area-efficient VLSI architecture for AVS intra frame encoder”, Visual

Communications and Image Processing 2007, Proc. of SPIE-IS & T Electronic Imaging, SPIE vol.

6508, pp. 650822, Jan. 29, 2007.

28. Proceedings of the 37th IEEE Asilomar conference on signals, systems and computers, Pacific

Grove ,CA, Nov. 9-12, 2003.

29. C. Li et al, Content-weighted video quality assessment using a three-component image model,

Journal of Electronic Imaging, Vol 19,Issue 1,Jan 2010, Pages 1-9.

30. Z. Wang and Q. Li, "Video quality assessment using a statistical model of human visual speed

perception," Journal of the Optical Society of America A, vol. 24, no. 12, pp. B61-B69, Dec. 2007.

31. Z. Wang et al, “Multi-scale structural similarity for image quality assessment,” Invited

Paper, IEEE Asilomar Conference on Signals, Systems and Computers, Nov. 2003.

32. Z. Wang et al, “Structural Approaches to image quality assessment,” in Handbook of Image and

Video Processing 2nd edition, Academic Press, June 2005.

33. Y. Huang, "Perceptual Rate-Distortion Optimization Using Structural Similarity Index as Quality

Metric," Circuits and Systems for Video Technology, IEEE Transactions on , vol.20, no.11,

pp.1614-1624, Nov. 2010

34. http://www.avs.org.cn/english/ : AVS website

35. AVS FTP Server: Source code and test video sequences.

36. http://forum.doom9.org/showthread.php?t=135034 : Consolidated list of test video clip

resources.

37. http://filezilla-project.org/ : FTP client

38. http://vspc.ee.cuhk.edu.hk/~ele5431/AVS.pdf : AVS China PPT by K. Ngan, Department of Electronic Engineering, The Chinese University of Hong Kong.

39. http://www.slideshare.net/sanjivmalik/video-compression-basics : Video compression basics by Sanjiv malik, Samsung Ltd.

40. http://compression.ru/video/index.htm : MSU Video quality measurement tool.