Overview, implementation and comparison of Audio Video Standard (AVS) China and H.264/MPEG -4 part 10 or Advanced Video Coding Standard

Multimedia Processing Term project Overview, implementation and comparison of Audio Video Standard (AVS) China and H.264/MPEG -4 part 10 or Advanced Video Coding Standard EE-5359 Class project Spring 2012 Submitted by: Kaustubh Vilas Dhonsale (10006999333) kaustubh.dhonsale@mavs.uta.edu Under the guidance of Dr. K. R. Rao

Table of Contents Acknowledgement... 4 List of acronyms in alphabetical order... 5 Introduction... 6 AVS China... 7 Introduction... 7 Encoder and decoder of AVS China... 8 Encoder... 8 Decoder... 9 Various profiles of AVS China... 10 Layered structure of AVS china... 12 Sequence... 13 Picture... 14 Slice... 14 Macroblock... 15 Block... 15 Coding tools in AVS China... 16 1. 8 X 8 spatial prediction... 16 2. Inter prediction... 17 3. Transform and Quantization... 21 4. In-loop deblocking filter... 21 5. 2D VLC (Variable Length Coding)... 22 MPEG-4 part 10 /AVC... 22 Introduction... 22 Decoder and encoder... 23 Encoder... 23 Decoder... 24 Profiles of H.264... 24 Advantages of H.264... 25 Video evaluation parameters... 26 2

Experimental results... 26 Results for AVS China... 28 AVS China results for sequence foreman_cif.yuv file... 28 AVS China results for sequence foreman_qcif.yuv file... 32 AVS China results for sequence news_cif.yuv file... 36 AVS China results for sequence news_qcif.yuv file... 40 Results for H.264... 44 H.264 results for sequence foreman_cif.yuv file... 44 H.264 results for sequence foreman_qcif.yuv file... 48 H.264 results for sequence news_cif.yuv file... 52 H.264 results for sequence news_qcif.yuv file... 56 Comparison... 60 Comparison for sequence foreman_cif.yuv file... 60 Comparison for sequence foreman_qcif.yuv file... 62 Comparison for sequence news_cif.yuv file... 64 Comparison for sequence news_qcif.yuv file... 66 Conclusions... 68 Future work... 68 References... 69 3

Acknowledgement I would like to acknowledge the guidance of Dr. K.R. Rao (Electrical Engineering Department at the University of Texas at Arlington) and his insightful support and inspiration throughout the various stages of this project. I sincerely appreciate the help and advice given by Dr. Rao which went a long way in helping me understand the key underlying concepts of the project. Several of the resources available at the University of Texas at Arlington were of great assistance in the fruitful completion of this project. I would also like to thank my fellow students for their valuable and timely inputs. 4

List of acronyms in alphabetical order AVC: AVS: CABAC: CAVLC: CIF: DCT: DPCM: HD: HDTV: IP: ISO: ITU-T: MB: MBPAFF: MPEG: MSE: NAL: PSNR: SD: SSIM: TV: VLC: QCIF: QP: Advanced video coding Audio video standard Context adaptive binary arithmetic coding Context Adaptive Variable Length Coding Common intermediate format Discrete cosine transform Differential pulse code modulation High definition High definition television Internet protocol International organization for standardization International telecommunication union s telecommunication standardization sector Macroblock Macroblock pair adaptive field frame Moving picture experts group Mean square error network adaptation layer Peak signal to noise ratio Standard definition Structural similarity index matrix Television VideoLAN client Quarter common intermediate format Quantization parameter 5

Introduction In real world applications compression is the most important part of the data transmission. Compression depends on the fact that the data is redundant in nature. It implies that up to some extent it was generated using some set of rules, and if those rules are known, the data can be accurately predicted. An encoder creates a compressed file that is ready for transmission or storage, retaining the information in the encoded data bits. Usually, video and audio files are large in nature so, compression becomes unavoidable in such conditions. The general block diagram explaining the scope of coding standardization is shown in Figure 1. Figure 1 Scope of video coding standardization [1] To play such a compressed file, an inverse algorithm is applied to produce a video that shows virtually the same content as the original source video. The time it takes to compress, send, decompress and display a file is called latency. The more advanced the compression algorithm, the higher the latency, given the same processing power. A pair of algorithms that works together is called a video codec (encoder/decoder). Video codecs that implement different standards are normally not compatible with each other; that is, video content that is compressed using one standard cannot be decompressed with a different standard. For instance, an MPEG-4 Part 2 decoder will not work with an H.264 encoder [2]. Different video compression standards use different methods to avoid data redundancy and therefore result in different bit rates, qualities and latency. 6

The purpose of this project is to implement and understand the performance of H.264 [1] and AVS China [9] in terms of parameters peak signal-to-noise ratio (PSNR), mean square error (MSE), structural similarity (SSIM) [15] and compression ratio using various test sequences at different bit rates etc. Figure 2 shows the history of the audio and video coding standards. Figure 2 History of audio / video coding standards [3]. AVS China Introduction Audio Video coding Standard (AVS) [8] is established by the Working Group of China in the same name which was approved by the Chinese Science and Technology Department of Ministry of Information Industry in June 2002 [9]. AVS comprises of two major video parts- AVS Part 2 [10] for high-definition digital video broadcasting and high-density storage media and AVS Part 7 [11] for low complexity, low resolution mobility applications. Compared with other standards, AVS has been designed to provide near optimum performance and a considerable reduction in complexity. AVS will therefore provide lowcost implementations [12]. 7

Encoder and decoder of AVS China The AVS video coding standard is based on the classic hybrid DPCM-DCT coder which was first introduced by Jain and Jain in 1979 [24]. 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. Encoder Figure 3 Block diagram of AVS encoder [13] 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 [24]. The 8

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 3 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 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 out the variable data rates produced due to 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 4 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. 9

Figure 4 Block diagram of AVS decoder [13]. 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. Various profiles of AVS China Profiles Jizhun profile Jiben Profile Shenzhan profile Jiaqiang profile Key application Television broadcasting, HDTV Mobility applications Video Surveillance Multimedia entertainment Table 1 AVS video profiles and applications [14]. AVS video Jizhun profile Jizhun profile is defined as the first profile in the national standards of AVS China-Part 2 [14], approved as national standard in 2006. This profile mainly focuses on digital video applications like commercial broadcasting and storage media, including high-definition applications. Basically it is preferred for high 10

coding efficiency on video sequences of higher resolutions, at the expense of moderate computational complexity. AVS China Jiben profile Jiben profile is defined in AVS-Part 7 [14] taking into account mainly mobility video applications with smaller picture resolution. Thus, computational complexity becomes a critical issue. Also the ability of error resilience is needed due to the wireless transporting environment. AVS-Part 7 reached to final committee draft at the end of 2004 [14]. AVS China Shenzhan profile The profile AVS-Shenzhan focuses exclusively on standardizing the video surveillance applications. Especially, there are special features of sequences from video surveillance such as the random noise appearing in pictures, relatively lower encoding complexity affordable, and friendliness to events detection and searching required [14]. Hence corresponding techniques considering a proper process on these special features will be encouraged. AVS China Jiaqiang profile Jiaqiang profile is mainly focused to fulfill the needs of multimedia entertainment; one of the major concerns of this profile is movie compression for high-density storage [14]. Relatively higher computational complexity can be tolerated at the encoder side to provide higher video quality, with compatibility to AVS-Part 2 as well. Table 2 shows the summary of AVS China profiles. 11

Table 2 Summary of profiles in AVS China [14] Parts of AVS Contents Stage Part 1 System for broadcasting Final committee draft Part 2 SD/HD Video National Standard Part 3 Audio Final draft Part4 Conformance test Final committee draft Part5 Reference software Final committee draft Part 6 Digital right management Final committee draft Part 7 Mobility video Final draft Part 8 System over IP Final draft Part 9 File format Final draft Table 3 AVS China parts [14] Layered structure of AVS china AVS China uses coded structure for the data compression and decompression. It can be easily explained with the help of coded bit stream. Figure.5 shows the layered data structure. 12

Figure 5 Layered data structure. At the top layer, multiple frames of the video are put in sequence into a buffer. This sequence is then combined to form frames, in the layer of picture/ frame and this combined sequence is called as picture. Then in the next layer of slice, these pictures are divided into rectangular regions and these regions are called as slices. Slices are divided into macroblocks (MBs) in the subsequent layer. MB consists of set blocks which has luminance and chrominance. Sequence The sequence layer consists of a set of mandatory and optional downloaded system parameters [17]. The mandatory parameters are required to initialize decoder systems. The optional parameters are used for other system settings at the discretion of the network provider. Sometimes user data can be contained in the sequence header. The sequence layer provides an entry point into the coded video. Sequence headers should be placed in the bitstream to support user access appropriately for the given distribution medium. Repeat sequence headers may be inserted to support random access. Sequences are terminated with a sequence end code. 13

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. Three types of pictures are defined by AVS [17]: 1. Intra pictures (I-pictures) 2. Predicted pictures (P-pictures) 3. Interpolated pictures (B-pictures) Slice The slice structure provides the lowest-layer mechanism for re-synchronizing the bitstream in case of transmission error. Slices comprise of a series of MBs. Slices must not overlap, must be contiguous, must begin and terminate at the left and right edges of the Picture. It is possible for a single slice to cover the entire Picture. The slice structure is optional. Slices are independently coded and no slice can refer to another slice during the decoding process. Figure 6: Normal slice structure and flexible slice set in AVS-video storage [17] (left: normal slice structure where slice can only contain continual lines of macroblocks; right: flexible slice set allowing more flexible grouping of MBs in slice and slice 14

Macroblock Picture is divided into several macroblocks (MB). The upper-left sample of each MB should not exceed picture boundary. Macroblock partitioning is shown in Figure. 7. The partitioning is used for motion compensation. The number in each rectangle specifies the order of appearance of motion vectors and reference indices in a bitstream. Figure 7 Macroblock partitioning A MB 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 sub-sampled 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 sub-sampled by a factor of two in the horizontal dimension; therefore each chrominance component contains two 8x8 blocks. The MB header contains information about the coding mode and the motion vectors. It may optionally contain the quantization parameter (QP). 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. 15

Coding tools in AVS China The major coding tools of AVS China part-2 are listed below: 1. 8 X 8 spatial prediction Spatial prediction is used in intra coding in AVS Part 2 to exploit spatial correlation of picture. The intra prediction is based on 8x8 block. It uses decoded information in the current frame as the reference of prediction, exploiting statistical spatial dependencies between pixels within a picture. If MBPAFF is applied, intra-frame prediction can only take the MBs within the same stage as reference. There are five luminance intra prediction modes, and four chrominance intra prediction modes. Each of the four 8x8 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 [17]. 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 to describe the intra-prediction mode in video bitstream. The reconstructed pixels of neighboring blocks before deblocking filtered is used as reference pixels for the current block is shown in Figure. 8 Figure 8 Neighbor pixels in luminance intra prediction [14] 16

Four luminance intra prediction directions are illustrated in Figure 8. Five luminance intra prediction modes are illustrated in Figure 9. Figure 9 Five luminance intra prediction modes [14] 2. Inter prediction P-picture and B-picture are specified in AVS Part 2. There are four macroblock partition types for inter prediction, 16x16, 16x8, 8x16 and 8x8 [17]. Figure 10 shows different types of macroblock and sub-macroblock partitions. 17

Figure 10 Macroblock and sub-macroblock partitions [17]. 2.1 P-Prediction In P-picture, there are 5 inter prediction modes, PSkip (16x16), P_16x16, P_16x8, P_8x16, and P_8x8. For the latter 4 modes in P-frame, each partition of macroblock is predicted from one of the two candidate reference frames, which are latest decoded I- or P-frame. For the latter 4 modes in P-field, each partition of macroblock is predicted from one of the four latest decoded reference fields. 2.2 Bi-prediction There are two kinds of bi-predictions in AVS Part 2, symmetric-prediction and direct-prediction. In symmetric-prediction, only one forward motion vector is transmitted for each partition. The backward motion vector is derived from the forward one by a symmetric rule (as shown in Figure 11). In direct-prediction, forward and backward motion vectors are all derived from the motion vector of the collocated block in the backward reference picture. 18

Figure 11 Symmetric mode of AVS Part 2 [14] 2.3 Interpolation A 1/4-pixel interpolation method named as Two Steps Four Taps interpolation (TSFT) is adopted in AVS Part 2. 1/2-pixel samples are interpolated in step 1 and 1/4-pixel samples in step 2. 1/2-pixel interpolation filter is a 4-tap filter Hi (-1/8, 5/8, 5/8, -1/8). For ordinary 1/4-pixel samples, a, c, d, f, I, k, n and q in Figure6, a 4-tap filter H2 (1/16, 7/16, 7/16, 1/16) is applied, and four special 1/4-pixel samples, e, g, p and r, are filtered by 2-tap bi-linear filter H3(1/2, 1/2)" [18]. The positions of the pixels are illustrated in Figure 12. Figure 12 Position of integer pixels, 1/2 pixels and 1/4 pixels [14]. The grey pixels are the integer pixel samples, blue ones are 1/2-pixel samples and white ones are 1/4-pixel samples 19

2.4 Reference frames Maximum of two reference pictures is allowed in inter-frame prediction of AVS-video. In particular, two reference frames are maximum when the current frame is coded with frame coding, while the reference index is no more than three (beginning from zero) when applying field coding. References are the nearest coded frames/fields of the current frame/field, except that the second field in B picture cannot take the first field of the same picture as its reference, as shown in Figure 13. Especially, reference can only be taken from the previous stages of the current macroblock pair in P-prediction when MBPAFF is applied. For video surveillance, normally the scene is not changed or changed slightly, background reference frames can replace one of the above two reference pictures and provide better coding efficiency. Figure 13 Maximum of two reference pictures in AVS-video: [14] (a) reference of current (the second) field in field-coded I picture, (b) references of current frame in frame-coded P picture, (c) references of current (the first) field in field-coded P picture, 20

(d) references of current (the second) field in field-coded P picture, (e) references of current frame in frame-coded B picture (f) references of current (the first) field in field-coded B picture and (g) references of current (the second) field in field-coded B picture 3. Transform and Quantization In a typical block based compression scheme like AVS, the residual block is transformed using a 8x8 integer transform. These integer transforms are a variation of discrete cosine transform (DCT) [22]. The transform outputs a set of coefficients, each of which is a weighting value for a standard basis pattern. Transform Quantize Coefficients 8x8 Image Block Figure 14 Quantization of the transform coefficients of the image block [17] The output of the transform, a block of transform coefficients, is quantized, i.e. each coefficient is divided by an integer value. Quantization reduces the precision of the transform coefficients according to a quantization parameter (QP). Typically, the result is a block in which most or all of the coefficients are zero, with a few non-zero coefficients. Setting QP to a high value means that more coefficients are set to zero; resulting in high compression at the expense of poor decoded image quality. Setting QP to a low value means that more non-zero coefficients remain after quantization, resulting in better decoded image quality but lower compression. 4. In-loop deblocking filter AVS uses the de-blocking filter in motion compensation loop. In-loop deblocking filter is applied in AVS Part 2, to reduce the blocking artifacts and enhance both subjective and objective performance. The de- 21

blocking process directly acts on the reconstructed reference first across vertical edges and then across horizontal edges. Obviously, different image regions and different bit rates need different smoothes. Therefore, the de-blocking filter is automatically adjusted in AVS depending on activities of blocks and QPs. 5. 2D VLC (Variable Length Coding) In AVS Part 2, an efficient context-based 2D-VLC entropy coder is designed for coding 8x8 block-size transform coefficients. 2D-VLC means that a pair of Run-Level is regarded as one event and jointly coded. Context-based is a technique, which uses the coefficient information to switch among different VLC tables. High performance can be achieved with the cost of relatively low complexity. MPEG-4 part 10 /AVC Introduction In order to produce the standard method of compression, a group of experts known as moving picture experts group (MPEG) was formed by international organization for standardization (ISO) [4] and international telecommunication union s telecommunication standardization sector [5] (ITU-T) to establish an international standard for coded representation of moving picture and associated audio on digital storage media.. Over the period of time various standards such as MPEG -1, MPEG-2, MPEG-4, MPEG-4 part 2, MPEG -4 AVC etc. [4] have been developed by this group for video as well as audio compression. This joint group involved in development of H.264 was focused on creating a simple and clean solution, limiting options and features to a minimum for video data. An important aspect of the standard, as with other video standards, is to provide the capabilities in profiles (sets of algorithmic features) and levels (performance classes) that optimally support popular productions and common formats. H.264 has seven profiles, each targeting a specific class of applications. Each profile defines what feature set the encoder may use and limits the decoder implementation complexity. Figure 15 shows the encoder and decoder for the H.264/AVC. 22

Decoder and encoder Figure 15 Block diagram of the H.264 encoder and decoder [6]. Encoder Encoder selects between intra- and inter-coding, as shown in Figure 15 for block-shaped regions of each picture. Intra-coding can provide access points to the coded sequence where decoding can begin and continue correctly. Intra-coding uses various spatial prediction modes to reduce spatial redundancy in the source signal for a single picture. Inter-coding (predictive or bi-predictive) is more efficient using 23

inter-prediction of each block of sample values from some previously decoded pictures. Inter-coding uses motion vectors for block-based inter-prediction to reduce temporal redundancy among different pictures. Prediction is obtained from deblocking filtered signal of previous reconstructed pictures. The deblocking filter reduces the blocking artifacts at the block boundaries. Motion vectors and intraprediction modes may be specified for a variety of block sizes in the picture. The prediction residual is then further compressed using integer DCT to remove spatial correlation in the block before it is quantized. Finally, the motion vectors or intra-prediction modes are combined with the quantized transform coefficient information and encoded using entropy code such as context-adaptive variable length codes (CAVLC) or context adaptive binary arithmetic coding (CABAC) [5]. Decoder The decoder receives a compressed bitstream from the NAL as shown in Figure 15. The data elements are entropy decoded and reordered to produce a set of quantized coefficients. These are rescaled and inverse transformed to give a difference macroblock. Using the header information decoded from the bit stream, the decoder creates a prediction macroblock P, identical to the original prediction P formed in the encoder. P is added to the difference macroblock and this result is given to the de-blocking filter to create the decoded macroblock [25]. Profiles of H.264 The AVC/H.264 standard defines four important Profiles: Baseline Profile Primarily used for lower-cost applications with limited computing resources, this profile is used widely in videoconferencing and mobile applications. Extended Profile Intended to use for streaming video profile, this profile has relatively high compression capability and some extra robustness to data losses and server stream switching can be achieved. 24

Main Profile Originally intended as the mainstream consumer profile for broadcast and storage applications, the importance of this profile faded when the High profile was developed for those applications.. High Profile The primary profile for broadcast and disc storage applications, particularly for high-definition television applications (this is the profile adopted into HD DVD and Blu-ray Disc, for example). Figure 16 Profiles of H.264 /AVC [8] Advantages of H.264 Up to 50% in bit rate savings: Compared to H.263v2 (H.263+) or MPEG-4 [7] Simple Profile, H.264 permits a reduction in bit rate by up to 50% for a similar degree of encoder optimization at most bit rates. High quality video: H.264 offers consistent video quality at high and low bit rates. Error resilience: H.264 provides the tools that are necessary to deal with packet losses in packet networks and bit errors in error-prone wireless networks. Network friendliness: Through the network adaptation layer (NAL), H.264 bit streams can be easily transported over different networks. Wide areas of application streaming mobile TV, HDTV over IP, extended PVR and storage options for the home user 25

Video evaluation parameters Peak signal to noise ratio (PSNR): PSNR is commonly used to measure the quality of reconstruction. The higher the PSNR higher the reconstruction quality. Mean Square Error: MSE is the difference between values estimated by estimator and true value being measured. MSE value of zero implies that the estimator has predicted the values correctly. Structural similarity (SSIM): SSIM measures the similarity between the two images. SSIM is designed to improve on traditional methods like PSNR and MSE [15]. Experimental results The software which has been used to perform AVS China is BM1.0v3 [19] obtained from the ftp server [19]. Microsoft Visual C++ 2008 express edition [20] has been used to run the code and build the project for the AVS China codec. After building the project, code will generate two application files namely lencod.exe and ldecod.exe. By running these two files using appropriate and necessary parameters final result which is a decoded file is obtained. The software used for the simulation of H.264 was FFmpeg: Fast Forwarding mpeg, version 0.6.1 [16]. FFmpeg is a open source software that produceslibraries and programs for handling multimedia data. The original file and decoded file are than evaluated using MSU video quality measurement tool. The values of PSNR, MSE and SSIM are obtained from it [21]. 26

Two types of files are used for the experiment: CIF: CIF stands for Common Intermediate Format, it is a video format used in video conference systems. As illustrated in the below figure, it specifies a data rate of 30 frames per second (fps), with each frame containing 288 lines and 352 pixels per line. Hence it has a resolution of 352 x 288 [26]. Figure 17 Chroma sampling for CIF [26] QCIF: QCIF stands for Quarter CIF, as illustrated in figure below it specifies each frame with 144 lines, with 176 pixels per line. Hence it has a resolution of 176x144. The "Quarter" terminology is meant to indicate that QCIF frames contain quarter as many pixels as the CIF frame and thus take up less bandwidth. Figure 18 Chroma sampling for QCIF [26] 27

Results for AVS China AVS China results for sequence foreman_cif.yuv file File used foreman_cif.yuv Resolution: 358X 288 Frame rate: 25 fps Original file size: 44550 Kilobytes Number of frames used: 100 Table 4 shows PSNR, MSE, SSIM and compression ratios for different bitrates. Figure 17 shows decoded images at different bitrates for AVS China. QP Compressed File size(kb) Bit rate (Kbits/s) Bits/pixel Y-PSNR (db) Y-MSE Y-SSIM Compression ratio 0 20734 17100 6.75 60.54 0.0573 0.9997 2.15 : 1 15 6213 5123.57 2.02 45.8679 1.6837 0.9942 7.17 : 1 31 1045 860.94 0.34 37.6255 11.234 0.9737 42.63 : 1 45 206 169.21 0.07 31.2507 48.7547 0.9186 216.26 : 1 63 34 27.67 0.01 23.8649 267.0482 0.7114 1310.29 : 1 Table 4 MSE, PSNR and SSIM for foreman_cif.yuv file for AVS China 28

Bitrate=17100 Kbits/s Bitrate= 860.94 Kbits/s PSNR= 60.54 db PSNR = 37.6255 db MSE= 0.0573 MSE= 11.2340 SSIM= 0.9997 SSIM= 0.9737 Compression ratio = 2.15 : 1 Compression ratio = 42.63 : 1 Bitrate = 27.67 Kbits/s PSNR = 23.8649 db MSE = 267.0482 SSIM = 0.7114 Compression ratio = 1310.29 : 1 Figure 19 Foreman_cif.yuv sequence compressed AVS China files at different bitrates. 29

MSE PSNR in db 65 PSNR Vs. Bitrate 60 55 50 45 40 35 30 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 Figure 20 PSNR Vs. Bitrate for foreman_cif.yuv sequence for AVS China 50 MSE Vs. Bitrate 45 40 35 30 25 20 15 10 5 0 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 Figure 21 MSE Vs. Bitrate for foreman_cif.yuv sequence for AVS china 30

SSIM 1 SSIM Vs. Bitrate 0.95 0.9 0.85 0.8 0.75 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 Figure 22 SSIM Vs. Bitrate for foreman_cif.yuv sequence for AVS China 31

AVS China results for sequence foreman_qcif.yuv file File used foreman_qcif.yuv Resolution: 176 X 144 Frame rate: 25 fps Original file size: 11138 Kilobytes Number of frames used: 300 Table 5 shows PSNR, MSE, SSIM and compression ratios for different bitrates. Figure 21 shows decoded images at different bitrates. QP Compressed File size(kb) Bit rate (Kbits/s) Bits/pixel Y-PSNR (db) Y-MSE Y-SSIM Compression ratio 0 5762 4751.41 7.50 53.658 0.2823 0.9985 1.93 : 1 15 1843 1519.72 2.40 44.894 2.214 0.9905 6.04 : 1 31 426 350.44 0.55 36.038 16.319 0.9493 26.15 : 1 45 118 96.68 0.15 29.161 79.499 0.855 94.39 : 1 63 14 11.12 0.02 20.357 603.59 0.5186 795.57 : 1 Table 5 MSE, PSNR and SSIM for foreman_qcif.yuv file for AVS China 32

Bitrate=4751.41 Kbits/s Bitrate= 350.44 Kbits/s PSNR= 53.65 db PSNR = 36.038 db MSE= 0.2823 MSE= 16.139 SSIM= 0.9985 SSIM= 0.9493 Compression ratio = 1.93 : 1 Compression ratio = 26.15 : 1 Bitrate = 11.12 Kbits/s PSNR = 20.357 db MSE = 603.59 SSIM = 0.5186 Compression ratio = 795.57 : 1 Figure 23 foreman_qcif.yuv sequence compressed AVS China files at different bitrates. 33

MSE PSNR in db 55 PSNR Vs. Bitrate 50 45 40 35 30 25 20 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 Figure 24 PSNR Vs. Bitrate for foreman_qcif.yuv sequence for AVS China 700 MSE Vs. Bitrate 600 500 400 300 200 100 0 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 Figure 25 MSE Vs. Bitrate for foreman_qcif.yuv sequence for AVS China 34

SSIM 1 SSIM Vs. Bitrate 0.95 0.9 0.85 0.8 0.75 0.7 0.65 0.6 0.55 0.5 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 Figure 26 SSIM Vs. Bitrate for foreman_qcif.yuv sequence for AVS China 35

AVS China results for sequence news_cif.yuv file File used: news_cif.yuv Resolution: 352 X 288 Frame rate: 25 fps Original file size: 44550 Kilobytes Number of frames used: 300 Table 6 shows PSNR, MSE, SSIM and compression ratios for different bitrates. Figure 25 shows decoded images at different bitrates. QP Compressed File size(kb) Bit rate (Kbits/s) Bits/pixel Y-PSNR (db) Y-MSE Y-SSIM Compression ratio 0 17571 792.64 0.31 58.8256 0.0852 0.9993 2.54 : 1 15 7671 153.135 0.06 47.3606 1.194 0.9902 5.81 : 1 31 3044 37.914 0.01 41.5042 4.5989 0.9747 14.64 : 1 45 1196 15.9 0.01 34.3129 24.087 0.9231 37.25 : 1 63 304 9.156 0.00 26.8455 134.43 0.8022 146.55 : 1 Table 6 SSIM Vs. Bitrate for news_cif.yuv sequence for AVS China 36

Bitrate=792.64 Kbits/s Bitrate= 37.914 Kbits/s PSNR= 58.82 db PSNR = 41.5042 db MSE= 0.0852 MSE= 4.5989 SSIM= 0.9993 SSIM= 0.9747 Compression ratio = 2.54:1 Compression ratio: 14.64:1 Bitrate = 9.156 Kbits/s PSNR = 26.8455 db MSE = 134.43 SSIM = 0.8022 Compression ratio = 146.55 : 1 Figure 27 news_cif.yuv sequence compressed AVS China files at different bitrates. 37

MSE PSNR in db 60 PSNR Vs. Bitrate 55 50 45 40 35 30 25 0 100 200 300 400 500 600 700 800 Figure 28 PSNR Vs. Bitrate for news_qcif.yuv sequence for AVS China 140 MSE Vs. Bitrate 120 100 80 60 40 20 0 0 100 200 300 400 500 600 700 800 Figure 29 MSE Vs. Bitrate for news_qcif.yuv sequence for AVS China 38

SSIM 1 SSIM Vs. Bitrate 0.98 0.96 0.94 0.92 0.9 0.88 0.86 0.84 0.82 0.8 0 100 200 300 400 500 600 700 800 Figure 30 SSIM Vs. Bitrate for news_qcif.yuv sequence for AVS China 39

AVS China results for sequence news_qcif.yuv file File used: news_qcif.yuv Resolution: 176 X 144 Frame rate: 25 fps Original file size: 11138 Kilobytes Number of frames used: 300 Table 7 shows PSNR, MSE, SSIM and compression ratios for different bitrates. Figure 31 shows decoded images at different bitrates. QP Compressed File size(kb) Bit rate (Kbits/s) Bits/pixel Y-PSNR (db) Y-MSE Y-SSIM Compression ratio 0 3232 2664.59 4.21 54.17 0.2509 0.9979 3.45 : 1 15 768 633.28 1.00 45.585 1.811 0.987 14.50 : 1 31 218 179.39 0.28 38.047 10.276 0.9635 51.09 : 1 45 62 50.28 0.08 30.706 55.704 0.8919 179.65: 1 63 12 9.39 0.01 20.188 627.54 0.6015 928.17: 1 Table 7 SSIM Vs. Bitrate for news_qcif.yuv sequence for AVS China 40

Bitrate=2664.59 Kbits/s Bitrate= 179.39 Kbits/s PSNR= 54.17 db PSNR = 38.047 db MSE= 0.2509 MSE= 10.276 SSIM= 0.9979 SSIM= 0.9635 Compression ratio = 3.45:1 Compression ratio = 51.09:1 Bitrate = 9.39 Kbits/s PSNR = 20.188 db MSE = 627.54 SSIM = 0.60.54 Compression ratio = 795.57 : 1 Figure 31 news_qcif.yuv sequence compressed AVS China files at different bitrates. 41

MSE PSNR in db 55 PSNR Vs. Bitrate 50 45 40 35 30 25 20 0 500 1000 1500 2000 2500 3000 Figure 32 PSNR Vs. Bitrate for news_qcif.yuv sequence for AVS China 700 MSE Vs. Bitrate 600 500 400 300 200 100 0 0 500 1000 1500 2000 2500 3000 Figure 33 MSE Vs. Bitrate for news_qcif.yuv sequence for AVS China 42

SSIM 1 SSIM Vs. Bitrate 0.95 0.9 0.85 0.8 0.75 0.7 0.65 0 500 1000 1500 2000 2500 3000 Figure 34 SSIM Vs. Bitrate for news_qcif.yuv sequence for AVS China 43

Results for H.264 H.264 results for sequence foreman_cif.yuv file File used: foreman_cif.yuv Resolution: 352 X 288 Frame rate: 25 fps Original file size: 44550 Kilobytes Number of frames used: 300 Table 8 shows PSNR, MSE, SSIM and compression ratios for different bitrates. Figure 31 shows decoded images at different bitrates. QP Compressed File size(kb) Bit rate (Kbits/s) Bits/pixel Y-PSNR (db) Y-MSE Y-SSIM Compression ratio 0 15887 10843.44 4.28 60.53 0.0574 0.9996 2.80: 1 15 2287 1557.91 0.61 44.99 2.0566 0.9888 19.48: 1 31 209 138.8 0.05 36.43 14.7843 0.9426 213.16: 1 45 54 33.62 0.01 28.89 66.6428 0.8278 825.00: 1 63 35 20.65 0.01 22.26 386.3426 0.5509 1272.86: 1 Table 8 SSIM Vs. Bitrate for foreman_cif.yuv sequence for H.264 44

Bitrate=10843 Kbits/s Bitrate= 138.8 Kbits/s PSNR= 60.53 db PSNR = 36.43 db MSE= 0.0574 MSE= 14.843 SSIM= 0.9996 SSIM= 0.9426 Compression ratio = 2.80 : 1 Compression ratio = 213.16 : 1 Bitrate = 35 Kbits/s PSNR = 22.26 db MSE = 386.3426 SSIM = 0.5509 Compression ratio = 1272.86 : 1 Figure 35 foreman_cif.yuv sequence compressed h.264 files at different bitrates. 45

MSE PSNR in db 65 PSNR Vs. Bitrate 60 55 50 45 40 35 30 25 20 0 2000 4000 6000 8000 10000 12000 Figure 36 PSNR Vs. Bitrate for foreman_cif.yuv sequence for H.264 400 MSE Vs. Bitrate 350 300 250 200 150 100 50 0 0 2000 4000 6000 8000 10000 12000 46

SSIM Figure 37 MSE Vs. Bitrate for foreman_qcif.yuv sequence for H.264 1 SSIM Vs. Bitrate 0.95 0.9 0.85 0.8 0.75 0.7 0.65 0.6 0.55 0 2000 4000 6000 8000 10000 12000 Figure 38 SSIM Vs. Bitrate for foreman_qcif.yuv sequence for H.264 47

H.264 results for sequence foreman_qcif.yuv file File used: foreman_qcif.yuv Resolution: 176 X 144 Frame rate: 25 fps Original file size: 11138 Kilobytes Number of frames used: 300 Table 9 shows PSNR, MSE, SSIM and compression ratios for different bitrates. Figure 31 shows decoded images at different bitrates. QP Compressed File size(kb) Bit rate (Kbits/s) Bits/pixel Y-PSNR (db) Y-MSE Y-SSIM Compression ratio 0 3607 2459.99 3.88 64.5322 0.0087 0.999 3.09: 1 15 499 337.29 0.53 52.563 0.837 0.9975 22.32: 1 31 73 46.01 0.07 45.3802 7.5099 0.9708 152.58: 1 45 23 11.82 0.02 26.346 120.8456 0.7736 484.26: 1 63 16 7.61 0.01 20.855 436.43 0.57 696.13: 1 Table 9 SSIM Vs. Bitrate for foreman_qcif.yuv sequence for H.264 48

Bitrate=2459.99 Kbits/s Bitrate= 46.01 Kbits/s PSNR= 64.5322 db PSNR = 45.3802 db MSE= 0.0087 MSE= 7.5099 SSIM= 0.999 SSIM= 0.9708 Compression ratio = 3.09 : 1 Compression ratio = 152.58 : 1 Bitrate = 7.61 Kbits/s PSNR = 20.855 db MSE = 436.43 SSIM = 0.5700 Compression ratio = 696.13 : 1 Figure 39 foreman_qcif.yuv sequence compressed h.264 files at different bitrates. 49

MSE PSNR in db 65 PSNR Vs. Bitrate 60 55 50 45 40 35 30 25 20 0 500 1000 1500 2000 2500 Figure 40 PSNR Vs. Bitrate for foreman_qcif.yuv sequence for H.264 450 MSE Vs. Bitrate 400 350 300 250 200 150 100 50 0 0 500 1000 1500 2000 2500 Figure 41 MSE Vs. Bitrate for foreman_qcif.yuv sequence for H.264 50

SSIM 1 SSIM Vs. Bitrate 0.95 0.9 0.85 0.8 0.75 0.7 0.65 0.6 0.55 0 500 1000 1500 2000 2500 Figure 42 SSIM Vs. Bitrate for foreman_qcif.yuv sequence for H.264 51

H.264 results for sequence news_cif.yuv file File used: news_cif.yuv Resolution: 352 X 288 Frame rate: 25 fps Original file size: 44550 Kilobytes Number of frames used: 300 Table 10 shows PSNR, MSE, SSIM and compression ratios for different bitrates. Figure 31 shows decoded images at different bitrates. QP Compressed File size(kb) Bit rate (Kbits/s) Bits/pixel Y-PSNR (db) Y-MSE Y-SSIM Compression ratio 0 11289 5801.37 2.29 67.63 0.0113 0.999 3.95: 1 15 4707 642.06 0.25 51.455 0.4688 0.996 9.46: 1 31 453 75.86 0.03 40.648 5.644 0.976 98.34: 1 45 69 19.54 0.01 30.079 64.348 0.8847 645.65: 1 63 27 9.35 0.00 23.978 262.193 0.7379 1650 : 1 Table 10 SSIM Vs. Bitrate for news_cif.yuv sequence for H.264 52

Bitrate=5801.37 Kbits/s Bitrate= 75.86 Kbits/s PSNR= 67.63 db PSNR = 40.648 db MSE= 0.0113 MSE= 5.644 SSIM= 0.999 SSIM= 0.976 Compression ratio = 3.95 : 1 Compression ratio = 98.34 : 1 Bitrate = 9.35 Kbits/s PSNR = 23.978 db MSE = 262.193 SSIM = 0.7379 Compression ratio = 1650 : 1 Figure 43 news_cif.yuv sequence compressed h.264 files at different bitrates. 53

MSE PSNR in db 70 PSNR Vs. Bitrate 65 60 55 50 45 40 35 30 25 20 0 1000 2000 3000 4000 5000 6000 Figure 44 PSNR Vs. Bitrate for news_qcif.yuv sequence for H.264 300 MSE Vs. Bitrate 250 200 150 100 50 0 0 1000 2000 3000 4000 5000 6000 Figure 45 MSE Vs. Bitrate for news_qcif.yuv sequence for H.264 54

SSIM 1 SSIM Vs. Bitrate 0.95 0.9 0.85 0.8 0.75 0.7 0 1000 2000 3000 4000 5000 6000 Figure 46 SSIM Vs. Bitrate for news_qcif.yuv sequence for H.264 55

H.264 results for sequence news_qcif.yuv file File used: news_qcif.yuv Resolution: 176 X 144 Frame rate: 25 fps Original file size: 11138 Kilobytes Number of frames used: 300 Table 11 shows PSNR, MSE, SSIM and compression ratios for different bitrates. Figure 31 shows decoded images at different bitrates. QP Compressed File size(kb) Bit rate (Kbits/s) Bits/pixel Y-PSNR (db) Y-MSE Y-SSIM Compression ratio 0 1938 1321.03 2.08 66.767 0.0138 0.999 5.75: 1 15 295 197.48 0.31 51.357 0.4794 0.997 37.76: 1 31 47 34.43 0.05 39.473 7.399 0.978 236.98: 1 45 16 7 0.01 28.089 101.74 0.8534 696.13: 1 63 12 4.48 0.01 21.78 434.99 0.6204 928.17: 1 Table 11 SSIM Vs. Bitrate for news_qcif.yuv sequence for H.264 56

Bitrate=1321.03 Kbits/s Bitrate= 34.43 Kbits/s PSNR= 66.767 db PSNR = 39.473 db MSE= 0.0138 MSE= 7.399 SSIM= 0.999 SSIM= 0.9780 Compression ratio = 5.75 : 1 Compression ratio = 236.98 : 1 Bitrate = 4.48 Kbits/s PSNR = 21.78 db MSE = 434.99 SSIM = 0.6204 Compression ratio : 928.17 : 1 Figure 47 news_qcif.yuv sequence compressed h.264 files at different bitrates. 57

MSE PSNR in db 70 PSNR Vs. Bitrate 65 60 55 50 45 40 35 30 25 20 0 200 400 600 800 1000 1200 1400 Figure 48 PSNR Vs. Bitrate for news_qcif.yuv sequence for H.264 450 MSE Vs. Bitrate 400 350 300 250 200 150 100 50 0 0 200 400 600 800 1000 1200 1400 Figure 49 MSE Vs. Bitrate for news_qcif.yuv sequence for H.264 58

SSIM 1 SSIM Vs. Bitrate 0.95 0.9 0.85 0.8 0.75 0.7 0.65 0 200 400 600 800 1000 1200 1400 Figure 50 SSIM Vs. Bitrate for news_qcif.yuv sequence for H.264 59

PSNR in db Comparison Comparison for sequence foreman_cif.yuv file File used: foreman_cif.yuv Resolution: 352 X 288 Frame rate: 25 fps Original file size: 44550 Kilobytes Number of frames used: 300 65 60 PSNR Vs. Bitrate AVS China H.264 55 50 45 40 35 30 25 20 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 Figure 51 Comparison between AVS China and H.264 based on PSNR for forman_cif.yuv sequence. 60

SSIM MSE 400 350 MSE Vs. Bitrate AVS China H.264 300 250 200 150 100 50 0 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 Figure 52 Comparison between AVS China and H.264 based on MSE for forman_cif.yuv sequence. 1 0.95 SSIM Vs. Bitrate AVS China H.264 0.9 0.85 0.8 0.75 0.7 0.65 0.6 0.55 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 Figure 53 Comparison between AVS China and H.264 based on SSIM for forman_cif.yuv sequence. 61

PSNR in db Comparison for sequence foreman_qcif.yuv file File used: foreman_qcif.yuv Resolution: 176 X 144 Frame rate: 25 fps Original file size: 11138 Kilobytes Number of frames used: 300 65 60 PSNR Vs. Bitrate AVS China H.264 55 50 45 40 35 30 25 20 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 Figure 54 Comparison between AVS China and H.264 based on PSNR for forman_qcif.yuv sequence. 62

SSIM MSE 700 600 MSE Vs. Bitrate AVS China H.264 500 400 300 200 100 0 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 Figure 55 Comparison between AVS China and H.264 based on MSE for forman_qcif.yuv sequence. 1 0.95 0.9 SSIM Vs. Bitrate AVS China H.264 0.85 0.8 0.75 0.7 0.65 0.6 0.55 0.5 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 Figure 56 Comparison between AVS China and H.264 based on SSIM for forman_qcif.yuv sequence. 63

PSNR in db Comparison for sequence news_cif.yuv file File used: news_cif.yuv Resolution: 352 X 288 Frame rate: 25 fps Original file size: 44550 Kilobytes Number of frames used: 300 70 65 60 PSNR Vs. Bitrate AVS China H.264 55 50 45 40 35 30 25 20 0 1000 2000 3000 4000 5000 6000 Figure 57 Comparison between AVS China and H.264 based on PSNR for news_cif.yuv sequence. 64

SSIM MSE 300 250 MSE Vs. Bitrate AVS China H.264 200 150 100 50 0 0 1000 2000 3000 4000 5000 6000 Figure 58 Comparison between AVS China and H.264 based on MSE for news_cif.yuv sequence. 1 0.95 SSIM Vs. Bitrate AVS China H.264 0.9 0.85 0.8 0.75 0.7 0 1000 2000 3000 4000 5000 6000 Figure 59 Comparison between AVS China and H.264 based on SSIM for news_cif.yuv sequence. 65

PSNR in db Comparison for sequence news_qcif.yuv file File used: news_qcif.yuv Resolution: 176 X 144 Frame rate: 25 fps Original file size: 11138 Kilobytes Number of frames used: 300 70 65 60 PSNR Vs. Bitrate AVS China H.264 55 50 45 40 35 30 25 20 0 500 1000 1500 2000 2500 3000 Figure 60 Comparison between AVS China and H.264 based on PSNR for news_qcif.yuv sequence. 66

SSIM MSE 700 600 MSE Vs. Bitrate AVS China H.264 500 400 300 200 100 0 0 500 1000 1500 2000 2500 3000 Figure 61 Comparison between AVS China and H.264 based on MSE for news_qcif.yuv sequence. 1 0.95 SSIM Vs. Bitrate AVS China H.264 0.9 0.85 0.8 0.75 0.7 0.65 0 500 1000 1500 2000 2500 3000 Figure 62 Comparison between AVS China and H.264 based on SSIM for news_qcif.yuv sequence. 67

Conclusions It can be concluded from the results that H.264 performs better in terms of PSNR, MSE, SSIM as well as compression ratio than AVS China. The graphs and tabulations clearly show that the PSNR, MSE and SSIM of the video sequences for H.264 improve as the bit rate increases, while the bit rate is varied using the quantization parameter. Future work Further work can be carried out in studying and comparing video coding standards: H.264/AVC and AVS China for all profiles i.e. main, baseline and high profiles. Test video sequences of SD and HD formats can also be used at various bit rates for better analysis of performance of the video coding standards. The problem in terms of testing SD and HD video sequences is that they consume a lot of time. 68

References 1. T. Wiegand, G. J. Sullivan, G. Bjøntegaard, and A. Luthra, Overview of the H.264/AVC Video Coding Standard, IEEE transaction on circuits and systems for video technology, vol. 13, pp. 560-576, JULY 2003. 2. H.264 video compression standard : New possibilities within video surveillance, Axis Commuincation,2008, Feb. 22nd 2012, <http://www.axis.com/files/whitepaper/wp_h264_31669_en_0803_lo.pdf> 3. ITU H.263 Video Compression : http://www.h263l.com/ 4. ITU-T: http://en.wikipedia.org/wiki/itu-t 5. S-K Kwon, A. Tamhankar and K.R. Rao Overview of H.264 / MPEG-4 Part 10, J. Visual Communication and Image Representation, vol. 17, pp. 186-216, April 2006. 6. X. Zhou, E. Q. Li, and Y.-K. Chen, Implementation of H.264 decoder on general purpose processors with media instructions, SPIE Conference on Image and Video Communications and Processing, vol. 5022, pp.224-235, May 2003. 7. J. Watkinson, The MPEG Handbook, Second Edition, Elsevier/Focal Press, pp. 1, 2004. 8. A. Puri, X. Chen and A. Luthra, Video coding using the H.264/MPEG- 4 AVC compression standard, Signal Processing: Image Communication, vol. 19, pp. 793-849, Oct. 2004. 9. Audio Video Coding Standard Group of China (AVS), Advanced Audio Video Coding Standard Part 2: Video, Doc. AVS-N1063, Dec, 2003. 10. L. Fan, S. Ma and F. Wu, Overview of AVS Video Standards, IEEE International Conference on Multimedia and Expo (ICME), pp. 423-426, 2004. 11. Audio Video Coding Standard Workgroup of China (AVS), Video Coding Standard FCD1.0, NOV.2003. 12. AVS Video Expert Group, Information Technology Advanced Audio Video Coding Standard Part 7: Mobility Video, Audio Video Coding Standard Group of China (AVS), Doc. AVS-N1151, Dec.2004. 13. W. Gao et al, AVS The Chinese next-generation video coding standard, NAB, Las Vegas, 2004. 14. L. Yu, S. Chen and J. Wang, Overview of AVS-video coding standards, Signal Processing: Image Communication, Vol. 24, Issue 4, Special Issue on AVS and its Application, pp. 247-262, April 2009. 69

15. J.B. Maretens and L. Meesters, Image dissimilarity, Signal Processing, vol. 70, pp. 155-176, Nov. 1998. 16. FFmpeg ver. 0.6.1 for H.264 software: http://www.ffmpeg.org/ 17. Project on "Low Complexity AVS-China Part-2 video using data mining techniques" by Jennie Abraham: http://www-ee.uta.edu/dip/courses/ee5359/thesis%20project%20table%20docs/ jennieproposal.doc 18. L. Yu et al, Overview of AVS-Video: Tools, performance and complexity, SPIE VCIP, vol. 5960, pp. 596021-1~ 596021-12, Beijing, China, July 2005. 19. AVS China software: ftp://124.207.250.92 (password required) 20. Microsoft Visual C++ 2008 express edition: http://msdn.microsoft.com/enus/express/future/bb421473 21. MSU video quality measurement tool: http://compression.ru/video/quality_measure/vqmt_download_en.html#start 22. C. Zhang et al, The Technique of Prescaled Integer Transform: Concept, Design and Applications, IEEE Transaction on circuits and systems for video technology, vol. 18, pp. 84-97, January 2008 23. I.E. Richardson, The H.264 Advanced Video compression standard, second edition, Wiley, 2010 24. Special issue on 'AVS and its Applications' Signal Processing: Image Communication, vol. 24, pp. 245-344, April 2009. 25. Apple Inc., Technology Brief Quicktime and MPEG-4 : http://images.apple.com/quicktime/pdf/h264_technology_brief.pdf 26. J.E. Caviedes, The Evolution of Video Processing Technology and Its Main Drivers, Frontiers of audiovisual communications: convergence of broadband, computing and rich media, vol. 100, pp. 872-877, April, 2012 70