Gg3311121115

164 views

Published on

International Journal of Engineering Research and Applications (IJERA) is an open access online peer reviewed international journal that publishes research and review articles in the fields of Computer Science, Neural Networks, Electrical Engineering, Software Engineering, Information Technology, Mechanical Engineering, Chemical Engineering, Plastic Engineering, Food Technology, Textile Engineering, Nano Technology & science, Power Electronics, Electronics & Communication Engineering, Computational mathematics, Image processing, Civil Engineering, Structural Engineering, Environmental Engineering, VLSI Testing & Low Power VLSI Design etc.

Published in: Technology
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
164
On SlideShare
0
From Embeds
0
Number of Embeds
4
Actions
Shares
0
Downloads
0
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

Gg3311121115

  1. 1. K.Rajalakshmi, K.Mahesh / International Journal of Engineering Research and Applications(IJERA) ISSN: 2248-9622 www.ijera.comVol. 3, Issue 3, May-Jun 2013, pp.1112-11151112 | P a g eLossless Compression in MPEG4 VideosK.Rajalakshmi, K.MaheshResearch scholar(M.Phil),Dept. of CSE, Alagappa University, Karaikudi, Tamilnadu,India.AbstractThis work considers object repetition,which means the occurrence of same object withsimilar motion is repeated in two or moreframes. A single video is divided into severalframes. All the objects are extracted from thestationary background and their features likemotion and shape are preserved. Afterextraction, a two step compression is employedin which the algorithm checks for the continuousmotion of all objects and then the video sequenceare segmented. Now, the segmented video iscompressed. This methodology maintains thevideo quality of the compressed video to be thesame as source video.Index Terms- Video, Compression.I. INTRODUCTIONNowadays, usage of video file increasesdue to the advent of youtube, facebook etc., Thispaper considers MPEG video as input. Videocompression without quality loss is a challenge andthis paper works on it.In this work, an MPEG video is broken upinto a hierarchy of layers to help with errorhandling, random search and editing, andsynchronization, for example with an audiobitstream. From the top level, the first layer isknown as the video sequence layer, and is any self-contained bitstream, for example a coded movie oradvertisement.The second layer down is the group ofpictures, which is composed of 1 or more groups ofintra (I) frames and/or non-intra (P and/or B)pictures that will be defined later. Of course thethird layer down is the picture layer itself, and thenext layer beneath it is called the slice layer.Each slice is a contiguous sequence ofraster ordered macroblocks, most often on a rowbasis in typical video applications, but not limitedto this by the specification. Each slice consists ofmacroblocks, which are 16x16 arrays of luminancepixels, or picture data elements, with 2 8x8 arraysof associated chrominance pixels.The macroblocks can be further dividedinto distinct 8x8 blocks, for further processing suchas transform coding. Each of these layers has itsown unique 32 bit start code defined in the syntaxto consist of 23 zero bits followed by a one, thenfollowed by 8 bits for the actual start code.These start codes may have as many zerobits as desired preceding them. The MPEGencoder also has the option of usingforward/backward interpolated prediction. Theseframes are commonly referred to as bi-directionalinterpolated prediction frames, or B frames forshort. As an example of the usage of I, P, and Bframes, consider a group of pictures that lasts for 6frames, and is given as I,B,P,B,P,B,I,B,P,B,P,B. Asin the previous I and P only example, I frames arecoded spatially only and the P frames are forwardpredicted based on previous I and P frames. The Bframes however, are coded based on a forwardprediction from a previous I or P frame, as well as abackward prediction from a succeeding I or Pframe. As such, the example sequence is processedby the encoder such that the first B frame ispredicted from the first I frame and first P frame,the second B frame is predicted from the secondand third P frames, and the third B frame ispredicted from the third P frame and the first Iframe of the next group of pictures. From thisexample, it can be seen that backward predictionrequires that the future frames that are to be usedfor backward prediction be encoded and transmittedfirst, out of order.This process is summarized in Fig 1.Thereis no defined limit to the number of consecutive Bframes that may be used in a group of pictures, andof course the optimal number is applicationdependent. Most broadcast quality applicationshowever, have tended to use 2 consecutive Bframes (I,B,B,P,B,B,P) as the ideal trade-offbetween compression efficiency and video quality.B-Frame EncodingThe main advantage of the usage of Bframes is coding efficiency. In most cases, Bframes will result in less bits being coded overall.Quality can also be improved in the case of movingobjects that reveal hidden areas within a videosequence. Backward prediction in this case allowsthe encoder to make more intelligent decisions onhow to encode the video within these areas.Also, since B frames are not used topredict future frames, errors generated will not bepropagated further within the sequence. Onedisadvantage is that the frame reconstructionmemory buffers within the encoder and decodermust be doubled in size to accommodate the 2anchor frames.This is almost never an issue for therelatively expensive encoder, and in these days of
  2. 2. K.Rajalakshmi, K.Mahesh / International Journal of Engineering Research and Applications(IJERA) ISSN: 2248-9622 www.ijera.comVol. 3, Issue 3, May-Jun 2013, pp.1112-11151113 | P a g einexpensive DRAM it has become much less of anissue for the decoder as well. Another disadvantageis that there will necessarily be a delay throughoutthe system as the frames are delivered out of order.Most one-way systems can tolerate these delays, asthey are more objectionable in applications such asvideo conferencing systems.II. Motion EstimationThe temporal prediction technique used inMPEG video is based on motion estimation. Thebasic premise of motion estimation is that in mostcases, consecutive video frames will be similarexcept for changes induced by objects movingwithin the frames.In the trivial case of zero motion betweenframes (and no other differences caused by noise,etc.), it is easy for the encoder to efficiently predictthe current frame as a duplicate of the predictionframe. When this is done, the only informationnecessary to transmit to the decoder becomes thesyntactic overhead necessary to reconstruct thepicture from the original reference frame.When there is motion in the images, thesituation is not as simple. Fig 1 shows an exampleof a frame with 2 stick figures and a tree. Thesecond half of this figure is an example of apossible next frame, where panning has resulted inthe tree moving down and to the right, and thefigures have moved farther to the right because oftheir own movement outside of the panning.The problem for motion estimation to solve is howto adequately represent the changes, or differences,between these two video frames.Fig 1: Video SequenceThe way that motion estimation goes about solvingthis problem is that a comprehensive 2-dimensionalspatial search is performed for each luminancemacroblock. Motion estimation is not applieddirectly to chrominance in MPEG video, as it isassumed that the color motion can be adequatelyrepresented with the same motion information asthe luminance.It should be noted at this point that MPEGdoes not define how this search should beperformed. This is a detail that the system designercan choose to implement in one of many possibleways.It is well known that a full, exhaustivesearch over a wide 2-dimensional area yields thebest matching results in most cases, but thisperformance comes at an extreme computationalcost to the encoder. As motion estimation usually isthe most computationally expensive portion of thevideo encoder, some lower cost encoders mightchoose to limit the pixel search range, or use othertechniques such as telescopic searches, usually atsome cost to the video quality.After a predicted frame is subtracted fromits reference and the residual error frame isgenerated, this information is spatially coded as in Iframes, by coding 8x8 blocks with the DCT, DCTcoefficient quantization, run-length/amplitudecoding, and bitstream buffering with rate controlfeedback.This process is basically the same withsome minor differences, the main ones being in theDCT coefficient quantization. The defaultquantization matrix for non-intra frames is a flatmatrix with a constant value of 16 for each of the64 locations. This is very different from that of thedefault intra quantization matrix which is tailoredfor more quantization in direct proportion to higherspatial frequency content.As in the intra case, the encoder maychoose to override this default, and utilize anothermatrix of choice during the encoding process, anddownload it via the encoded bitstream to thedecoder on a picture basis. Also, the non-intraquantization step function contains a dead-zonearound zero that is not present in the intra version.This helps eliminate any lone DCT coefficientquantization values that might reduce the run-length amplitude efficiency. Finally, the motionvectors for the residual block information arecalculated as differential values and are coded witha variable length code according to their statisticallikelihood of occurrence.
  3. 3. K.Rajalakshmi, K.Mahesh / International Journal of Engineering Research and Applications(IJERA) ISSN: 2248-9622 www.ijera.comVol. 3, Issue 3, May-Jun 2013, pp.1112-11151114 | P a g eIII. RELATED WORKIn this section, we are presenting theresearch work of some prominent authors in thesame field and explaining a short description ofvarious techniques used for video compression.G.Suresh, P.Epsiba, Dr.M.Rajaram,Dr.S.N.Sivanandam “A LOW COMPLEXSCALABLE SPATIAL ADJACENCY ACC-DCTBASED VIDEO COMPRESSIONMETHOD”,2010 proposed a video compressionapproach which tends to hard exploit the temporalredundancy in the video frames to improvecompression efficiency with less processingcomplexity.It produces a high video compressionratio. Many experimental tests had been conductedto prove the method efficiency especially in highbit rate and with slow motion video. The proposedmethod seems to be well suitable for videosurveillance applications and for embedded videocompression systems. Tzong-Jer Chen, Keh-Shih Chuang “A Pseudo Lossless ImageCompression Method”,2010 present a losslesscompression which modifies the noise componentof the bit data to enhance the compression withoutaffecting image quality. Data compressiontechniques substantially reduce the volume of theimage data generated and thus increase theefficiency of the information flow. Method isinformation lossless and as a result, thecompression ratio is smaller.Qiang L iu, Robert J . Sclabassi , Mark L. Scheuer, and Mingui Sun “A Two-step MethodFor Compression of Medical MonitoringVideo”2010 present a two-step method to compressmedical monitoring video more efficiently. In thefirst step, a novel algorithm is utilized to detect themotion activities of the input video sequence.Then, the video sequence is segmented into severalrectangle image regions (video object planes),which contain motion activities restricted withinthese windows. In the second step, the generatedvideo object planes are compressed. Ourexperimental results show that the two-step methodimproves the compression ratio significantly whencompared with the existing algorithms while stillretaining the essential video quality.Raj Talluri, , Karen Oehler, ThomasBannon, Jonathan D. Courtney, Arnab Das, andJudy Liao “A Robust, Scalable, Object-BasedVideoCompression Technique for Very Low Bit-RateCoding” 1997 describes an object-based videocoding scheme (OBVC) this technique achievesefficient compression by separating coherentlymoving objects from stationary background andcompactly representing their shape, motion, and thecontent. In addition to providing improved codingefficiency at very low bit rates, the techniqueprovides the ability to selectively encode, decode,and manipulate individual objects in a videostream. Applications of this object-based videocoding technology include video conferencing,video telephony, desktop multimedia, andsurveillance video.Ian Gilmour, R. Justin Dávila “LosslessVideo Compression for Archives: Motion JPEG2kand Other Options”2011 algorithm is clearly forend-user distribution through narrow bandwidths,and where no subsequent re-coding or re-puposingis required. The optimisation of image qualitywithin individual frames allows true lossless data-reduction for applications such as archiving, whereno loss of image quality is acceptable.Yucel Altunbasak, A. Murat Tekalp, andGozde Bozdagi “TWO-DIMENSIONAL OBJECT-BASED CODING USING A CONTENT-BASEDMESH AND AFFINE MOTIONPARAMETERIZATION”1995 present a completesystem for 2-D object-based video compressionwith a method for 2-D content-based triangularmesh design, two connectivity preserving affinemotion parameterization schemes, two methods fortemporal mesh propagation, a polygon-basedadaptive model failure detection/coding scheme,and bitrate control strategies.Raj Talluri “A HYBRID OBJECT-BASED VIDEO COMPRESSIONTECHNIQUE”1996 describes a hybrid object-based video coding scheme that achieves efficientcompression by separating coherently movingobjects from stationary background and compactlyrepresenting their shape, motion and the content.In addition to providing improved codingefficiency at very low bit rates the techniqueprovides the ability to selectively encode, decodeand manipulate individual objects in a video streamApplications of this object- based video codingtechnology include video conferencing, videotelephony, desktop multimedia, and surveillancevideo. Video conferencing, video telephony,desktop multimedia and surveillance video.Zixiang Xiong “Video CompressionBased on Distributed Source Coding Principles”2009 This paper applies distributed source codingprinciples to three areas of video compression,leading to Witsenhausen-Wyner video coding,layered Wyner- Ziv video coding, andmultiterminal video coding. Compared toH.264/AVC and H.26L-FGS video coding, theadvantage of Witsenhausen-Wyner video codingand layered Wyner-Ziv video coding is errorrobustness when the compressed video has to bedelivered over a noisy channel. The price for thisadded error robustness is a small loss ofcompression performance.
  4. 4. K.Rajalakshmi, K.Mahesh / International Journal of Engineering Research and Applications(IJERA) ISSN: 2248-9622 www.ijera.comVol. 3, Issue 3, May-Jun 2013, pp.1112-11151115 | P a g eIV. PROPOSED WORKOur work follows the below givenalgorithm, which successfully segments the videoand compresses it without any quality loss.Break video into layers;Compute B frames;Estimate motion;if(motion=0)duplicate frame;elsesubtract the predicted frame from previous frame;end if;obtain motion vectors;end;Fig:2 AlgorithmIn this work, an MPEG video is broken upinto a hierarchy of layers to help with errorhandling, random search and editing, andsynchronization, for example with an audiobitstream. From the top level, the first layer isknown as the video sequence layer, and is any self-contained bitstream, for example a coded movie oradvertisement.The second layer down is the group ofpictures, which is composed of 1 or more groups ofintra (I) frames and/or non-intra (P and/or B)pictures that will be defined later. Of course thethird layer down is the picture layer itself, and thenext layer beneath it is called the slice layer.Each slice is a contiguous sequence ofraster ordered macroblocks, most often on a rowbasis in typical video applications, but not limitedto this by the specification. Each slice consists ofmacroblocks, which are 16x16 arrays of luminancepixels, or picture data elements, with 2 8x8 arraysof associated chrominance pixels.The macroblocks can be further dividedinto distinct 8x8 blocks, for further processing suchas transform coding. Each of these layers has itsown unique 32 bit start code defined in the syntaxto consist of 23 zero bits followed by a one, thenfollowed by 8 bits for the actual start code.These start codes may have as many zerobits as desired preceding them. The MPEGencoder also has the option of usingforward/backward interpolated prediction. Theseframes are commonly referred to as bi-directionalinterpolated prediction frames, or B frames forshort.The main advantage of the usage of Bframes is coding efficiency. In most cases, Bframes will result in less bits being coded overall.Quality can also be improved in the case of movingobjects that reveal hidden areas within a videosequence. Backward prediction in this case allowsthe encoder to make more intelligent decisions onhow to encode the video within these areas. Afterall the processes mentioned in Introduction part,the motion vectors for the residual blockinformation are calculated as differential values andare coded with a variable length code according totheir statistical likelihood of occurrence.V. CONCLUSIONThus, this work segments the input videoand then estimates the motion. After extraction, atwo step compression is employed in which thealgorithm checks for the continuous motion of allobjects and then the video sequence are segmented.Now, the segmented video is compressed. Thismethodology maintains the video quality of thecompressed video to be the same as source video.REFERENCES[1] Raj Talluri, Member, IEEE, KarenOehler, Member, IEEE, Thomas Bannon,Jonathan D. Courtney, Member, IEEE,Arnab Das, and Judy Liao “A Robust,Scalable, Object-Based VideoCompression Technique for Very Low Bit-Rate Coding”, IEEE TRANSACTIONSON CIRCUITS AND SYSTEMS FORVIDEO TECHNOLOGY, VOL. 7, NO. 1,FEBRUARY 1997.[2] G.Suresh, P.Epsiba, Dr.M.Rajaram,Dr.S.N.Sivanandam “A LOW COMPLEXSCALABLE SPATIAL ADJACENCY ACC-DCT BASED VIDEO COMPRESSIONMETHOD”, 2010 Second Internationalconference on Computing,Communication and NetworkingTechnologies.[3] Tarek Ouni, Walid Ayedi, Mohamed Abid“New low complexity DCT based videocompression Method”, 2009 IEEE.[4] K.Uma, P.Geetha palanisamy, P.Geethapoornachandran “Comparison of ImageCompression using GA, ACO and PSOtechniques” , IEEE-InternationalConference on Recent Trends inInformation Technology, ICRTIT 2011.[5] Tzong-Jer Chen, Keh-Shih Chuang “APseudo Lossless Image CompressionMethod”, 2010 3rd InternationalCongress on Image and SignalProcessing.[6] Nasir D. Memon, Khalid Sayood“Asymmetric Lossless ImageCompression”, 2011IEEE

×