Appraisal of H.264 Codec

Transcription

1 International Journal of Electronics and Computer Science Engineering 651 Available Online at ISSN Appraisal of H.264 Codec Mrs. R. N. Mandavgane 1, Dr. N. G. Bawane Department of Electronics and Communication Engineering 1 B.D. College of Engineering Sevagram, Wardha Nagpur University. 2 Principal, S. B. Jain Institute Technology, Management and Research, Nagpur Nagpur University. 1 rmandavgane@rediffmail.com Abstract- H.264/AVC, the latest video Codec standard is the most popular among Video professionals owing to its exact and unambiguous guidelines for dealing with the input stream to be transported from one place to another, having minimum loss in the transit and a very good compression ratio, both attributes highly sought after in today s world of high quality multimedia broadcast and reception. Another parameter is the bit rate with which the data has to be transmitted. This paper is an attempt to show how various input parameters affect the output of the H.264 Codec. The input video file is a.yuv file, which after encoding becomes a.264 file, which is also called a stream file because this file has to be sent as a stream to various locations owing to its small size. At the receiving end, this stream file is decoded back into.yuv file. In the entire process, there is some loss in the quality of the video transmitted. This loss has to be minimum and is indicated by a parameter called PSNR (Peak Signal to Noise Ratio). Apart from PSNR, there are some other parameters as well like sequence parameter set, picture parameter set, information about different frames as regards slices and macroblocks, all ensconced in different files. PSNR being an important parameter will be dealt with here. There are three different input profiles namely, baseline profile, main profile and extended profile for H.264. In this work, variation of bit rate versus PSNR is studied in all the three profiles. A higher PSNR for a given constant bit rate is rated as a better performance. Here, the input sequence is a yuv file (foreman_part_qcif.yuv) with one reference frame. The quantization parameter for the I, B and P slices has been set at 30. The PSNR is plotted against bit rate and it is observed that they show a striking similarity with each other. Keywords, PSNR, compression, decompression, profiles of H.264 I. INTRODUCTION Gadgets involving real time video transmission and reception for a common man, like video calling over the internet with applications such as Skype and ichat, web browsing and downloading on cell phones etc which are ubiquitous today were simply not there about 10 years ago. Advancements in hardware technology and bigger available bandwidths can be attributed to this scenario. However, the H.264/AVC and SVC Codec has further accelerated this development. Faithful transmission, reception and storage of video data is a very complex task and involves many processes, encoding and decoding of data being the most crucial one. A truly practical attempt was made in 1993 by way of introduction of the MPEG-1 standard. It evolved further and we have higher versions of the same. In 1998, VCEG (Video Coding Expert Group) introduced H.261 Codec, which was well received, which also evolved to H.263. Finally, MPEG and VCEG joined hands together to form a JVT (Joint Video Team) and H.264/MPEG4 PART 10 or H.264/AVC came into existence. The scalable extension of H.264/MPEG4 Part 10 AVC is a current standardization project as H.264/SVC (Scalable Video Coding). The research community is showing a lot of interest in this promising area and contributing to International standardization. A gradual evolution from MPEG-1 to MPEG-4 and H.261 to H.264 standard is enumerated in [3].

2 IJECSE,Volume2, Number Mrs. R. N. Mandavgane and Dr. N. G. Bawane A. Review of previous work The technical features and the profiles of H.264/AVC are described in [5], [11] and [12]. The history of the standardization process is also reviewed by the above papers giving some applications for standard. Some key advantages of H.264/AVC such as high quality video, bit rate savings up to 50%, error resilience (providing tools necessary to deal with packet loss), network friendliness (bit streams can be easily transmitted over networks) are mentioned in [12]. The description of H.264 s basic concept is written in short in [12] although it cannot be considered as a replacement of the full fledged H.264 standard document. Iain Richardson has contributed a lot on video compression in his books [1], [2], [3]. B. Paper contents The outline of paper is as follows. The brief introduction of H.264 codec is given in section II explaining the input output files of the encoder and decoder clearly. Section III describes graphically the performance of the three profiles. The paper is concluded in section IV. II. BRIEF WORKING OF THE H.264 CODEC H.264 defines the syntax of an encoded bit stream and the method of decoding it. It however, has no guidelines for the actual implementation for the same. Different companies have their own methods of implementing the hardware and software while adhering to the H.264 standards. The encoded video data (.yuv file) is in the form of a stream file with an extension of.264, which gets converted back to the.yuv format after getting decoded at the receiving end. Since video files are bulky, compression of video data is a challenge. Assuming the maximum bandwidth available to be constant, a large amount of video data to be sent demands a very high compression ratio from the encoder. The raw video data or the original input video file is organized as a stream of frames. A frame rate of more than 25 per second is considered as acceptable. However, the encoded file is very complex and is organized as a very cryptic data about the input file. In this file, each frame is divided into slices, each slice, into macroblocks and each macroblock into blocks. All this information is stored into the.264 (encoded) file as it is the file which actually streams from one point to another. Squeezing all the information about the original video file into this streaming file makes the.264 file a very complex one. Here, the macroblock is considered as the basic unit to be communicated with by the Codec and its blocks are the subunits carrying detailed information about each pixel. The macroblock contains 26 blocks [3]. Depending upon the required output quality, H.264 has different profiles or levels namely baseline profile, main profile and extended profile. The levels define the performance limits such as sample processing rate, picture size, coded bit rate and memory size. The 4:2:0 scheme of video format is used in codec, Y the luminance component, blue and red chroma i.e. Cb and Cr are transmitted, each having half the horizontal and vertical resolution of Y. The output of the encoding process is a VCL (Video Coding Layer) which can be mapped to NAL (Network Abstraction layer) prior to transmission or storage. The block diagram in fig 1 indicates the input video and different output files from codec. Here, a Trace file is generated which contains detailed information about the coding at block level. It is a human intelligible text file. The Stat file is also a text file containing all the statistics regarding the encoding process. There is also one.rec file (which is organized as a.yuv file) and is called as a reconstruction file. This file is very close to the yet to be decoded.264 file and for practical purposes of assessing its decoded quality, can be used by the viewer as a close approximation of the decoded output.yuv file. The encoder may use one or two of a previously encoded pictures as reference picture which enables the encoder for the best match for the current macroblock partition from a wider set of pictures [1].

3 Appraisal of H.264 Codec 653 Figure 1. Block schmatic for codec In this experiment, the input file is foreman_part_qcif.yuv and is in 4:2:0 format, the first frame of which is shown in Figure 2. The size of this file is 112 KB and has 3 frames. The.264 file generated has a size of only 4 KB. This shows a tremendous compression of the input file. A. Baseline profile Figure 2. Frame from foreman sequence III. PSNR CURVES H.264 is operated for baseline profile on the YUV file of foreman sequence. The single reference frame is used for predictions. The foreman sequence is encoded with different bitrates and the graph is shown in fig 3. In this profile the first frame is an IDR and next frames are P frames as per the constraint of baseline profile. Figure 3. Bitrate vs PSNR curve for baseline profile

4 IJECSE,Volume2, Number B. Main profile Mrs. R. N. Mandavgane and Dr. N. G. Bawane The same sequence of 10 frames is operated with main profile. In this profile first frame is an IDR and then P and B slices are generated alternately for the next frames if hierarchy is not mentioned. Fig 4 shows the bitrate vs. PSNR performance. Figure 4. Bitrate vs PSNR curve for main profile C. Extended profile For the extended profile also the observations are almost the same with the same compression. The first frame is encoded as IDR. P and B slices follow alternately for the next frames. D. Observation Figure 5. Bitrate vs PSNR curve for extended profile A comparison between the three figures (figs. 3, 4 and 5) clearly shows that, as the bitrate increases, the PSNR also increases. Also, the performance of baseline profile, main profile and extended profile shows that, they are in ascending order. So, the codec s rate distortion performance shows the tradeoff between PSNR and bitrate. This shows that, higher the bandwidth, better the is the resulting quality. The encoding parameters of the encoder are defined in a configuration file encoder.cfg, which allows the user to set the desired parameters for input to the encoder.

5 Appraisal of H.264 Codec 655 Figure file from the codec Figure 7. Encoder trace file

6 IJECSE,Volume2, Number Mrs. R. N. Mandavgane and Dr. N. G. Bawane Figure 8. Diagram showing statistics IV. CONCLUSION The codec compresses the YUV file with a good degree of compression. The original file of 3 frames is of 112 KB, where as the output.264 file has 4 KB size. The.264 file displayed in fig 6 as a hex dump. A unique bit pattern of is a header for the different fields in the file. The first slot of bytes are of sequence parameter set, the next slot of bytes are picture parameter set and the three slices as the IDR, P, P for baseline profile. YUV file generated by the encoder is the reference or reconstructed file. The stat file in fig. 8 displays the number of frames coded, frequency for the encoded bitstream, image format and many relevant things. A good rate distortion factor and low processing power are the most One of the desirable properties of a codec includes good rate distortion performance and low processing power required to code the video sequence.

7 Appraisal of H.264 Codec 657 REFERENCE [1] Iain E. G. Richardson, Wiley 2003, H.264 and MPEG 4 video compression. [2] Iain E. G. Richardson, Wiley 2010, The H.264 advanced video compression standard. [3] Iain E. G. Richardson, Wiley 2002, Video codec design. [4] H.264/ AVC Reference Software Manual (Revised for JM 18.0). [5] T. Wiegand, G. J. Sullivan, G. Bjontegaard, and A. Luthra, Overview of the H.264/AVC Video Coding Standard IEEE Transactions on Circuits and Systems for Video Technology, Vol. 13, No. 7, July. [6] M. Fiedler, Implementation of a basic H.264/AVC Decoder Seminar Paper, Chemnitz university of technology, June 1, [7] D. Marpe, T. Wiegand, G. J. Sullivan, The H.264/MPEG4 Advanced Video Coding Standard and its Applications Standards Report. [8] Iain Richardson, H.264 / AVC Picture Management VCodex, White Paper, [9] International Standard ISO/IEC , Part 10 Advanced Video Coding second edition, [10] S. Wenger, M.M. Hannuksela, T. Stockhammer, M. Westerlund, D. Singer, RTP Payload Format for H.264 Video February [11] R. Schafer, T. Wiegand, Heiko Schwarz, The Emerging H.264/AVC Standard Heinrich Hertz Institute, Berlin, Germany. [12] MPEG 4 part 10 AVC(H.264) Video Encoding, Scientific Atlanta, June 2005.