摘要:W ith the increasingly extensive application for the new emerging video coding standard, H.264/AVC, it becomes more and more important that we transport H.264/AVC video stream over IP Network real time. We propose a solution for it and focus on two respects—H.2642FGS encoder and rate control to explain.
关键词:H.264/AVC;IP network;fine granular scalable(FGS);real-time;rate control
摘要:The visual disturbing discontinuities at the block boundaries, which are called block artifacts,might arise in the reconstructed pictures,especially in low bit-rate applications,because the block-based motion compensation prediction,transform and quantization are adopted in the hybrid video coding standards.A deblocking filter,projections onto convex sets,and lapped orthogonal transform behave as the candidate technologies for reducing the block effects in signal processing.This paper analyzes and integrates the systematic architecture of video coding,achieves a deblocking hybrid video coding framework on the basis of a lapped orthogonal transform with time-domain pre- and post-filtering,and discusses the key technology issues including overlap mode choice within the framework.The experimental outcomes are shown to improve the subjective quality of video pictures with a comparative rate-distortion performance,compared to the traditional framework.
关键词:block artifact;lapped orthogonal transform;discrete cosine transform;time-domain pre- and post filtering
摘要:Block matching motion estimation(BMME) plays a very important role in video coding.The performance of BMME greatly affects the quality of the encoded bit stream.Full Search(FS) is the optimal algorithm for BMME.But its tremendous computation payload is unaffordable for practical applications,especially for real-time applications.To overcome this problem,a lot of fast algorithms were developed.This paper presents an adaptive fast motion-estimation algorithm based on H.264.Taking advantage of motion-vectors of the nearby macro-blocks to find the original searching central position and determine the size of the searching zone.Compared with the former algorithms with fixed searching zone,this algorithm is more flexible and can combine with any other fast motion estimation algorithm.Experimeotal results show that the algorithm can reduce computation amount variously by different motion-sequence under the premise don't reduce the PNSR.
摘要:To deliver H.264 real-time video streaming over 3G networks,an error resilient coding scheme is presented in this paper.The system first divides original sequence into two sub-sequences to encode separately and then transmits the resulting two bitstreams over different channels so that the effect of burst channel error on image quality can be decreased.The frames which are not correctly decoded are predicted by bidirectional weighted interpolation of the frames from the other sub-stream.In order to restore the decoding process of the sub-stream suffering transmission error,the reference picture of the other sub-streams correctly decoded is selected as reference for the error sub-stream so that the error propagation can be eliminated.The simulation results show that the proposed method can efficiently combat bit error and packet loss in wireless channels and improve the image quality.The performance of the proposed method gains more than 0.6dB compared with the related approaches.
摘要:In order to make motion estimation quicker,for the computation complexity of full search of inter coding is very costly in H.264 that can not be used in real-time applying,this paper proposed a quick search algorithm which is adapt to H.264.For one of the special characteristic of H.264 is multi-prediction mode,this paper utilize of the correlation of each mode to determinate a prediction;at the time,this paper also consider general regulation of motion picture to determinate a prediction,these prediction can reduce time,then use diamond search algorithm to end up motion estimation,this method can reduce much prediction time.The experimentation result show that the algorithm can raise 65% coding rate with increase no more than 1% bite-rate and descend no more 0.1dB SNR(signal noise ratio).
摘要:Packet algorithm is one of important effective technology against error.In this paper,a new scattered packet algorithm based on multiple jumpy scans is proposed,which combines the advantages of traditional raster-order scan and recent scattered slices.The proposed packet algorithm,which scans the scattered macroblocks in certain order,overcomes the limitation of macroblock's fixed numbers in scattered slices,and implements excellence scattered packets in term of bit numbers.
摘要:In this paper,a new fast embedded zerotree wavelet(EZW) image coder based on relation matrix is presented.Firstly,the discrete wavelet transform is performed on the origin image.Secondly,the wavelet coefficients of the lowest frequency subband are encoded by using DPCM.Thirdly,the zerotrees are defined for the wavelet coefficients of the high frequency subbands.Finally,the initial threshold is identified and relation matrix is constructed,and the wavelet coefficients of the high frequency subbands are encoded progressively by using relation matrix.The experiment results show that the new still image compression scheme provides higher perceptual quality and less encoding time than EZW,especially at low bit rates.
摘要:In order to ensure the demultiplexing quality of MPEG-2 TS stream,it has brought in a system design for the demultiplexing based on DSP and FPGA in detail.This system belongs to real-time processing and includes both software and hardware.Multiple experiments have evidenced that this design can make the DSP clock works at 160M stably,and it can also ensure the image quality even the image processing rate is as high as 80Mbps,which meets the design goal so well.
关键词:digital signal processor(DSP);field-programmable gate array(FPGA);demultiplex;multi-program transport stream(MPTS);signal program transport stream(SPTS)
摘要:In H.264/AVC, intraprediction is used to improve the coding efficiency of I frame. There are one size of 8×8 for the chroma block and two sizes including 4×4 and 16×16 for the luma block. Each kind of block provides many prediction modes, so that this multimode selection increases the coding complexity greatly while improving coding efficiency. In this paper, a fast intraprediction mode selection algorithm with different criterion based on RDO(Rate distortion optimization) is proposed using the mode selection relation of chroma block and luma block and the relation of luma 4×4 block and luma 16×16 block. Experiment results show that our proposed algorithm can averagely speed up to times while maintaining similar PSNR and bit rate.
摘要:To describe the definite relationship between interpolation error induced by precision inefficient and noise error,an analysis of MSE performance of comb-type estimator and block-type estimator based on linear interpolation in OFDM system is discussed in this paper as there is no literature focus on it by now,and also formulae illustrating the effect of interpolation error and noise reduction to system performance are deduced then,a criterion for linear interpolation channel estimator pilot sub carries interval selection is derived.Simulation results which based on Hiperlan/2 show that the MSE performance of comb-type estimator when with the same pilot sub carriers interval is better than that of block-type estimator in indoor circumstances.
摘要:At present,the condition access(CA) has been separated from the "separation of machine and card" set-top box.The channel decoder and source decoder remain together,which disadvantages the development of the integration of digital television(DTV).In this paper,a new "separation of source decoder and channel decoder" scheme of digital set-top box is further proposed,in which the channel decoder is divided as an independent part,and only the source decoder is left.
关键词:digital television;condition access;separation of source decoder and channel decoder
摘要:AVS(audio video coding standard) is a digital audio-video coding/decoding standard proposed by China since 2002.It adopted many digital audio-video codec techniques developed by Chinese companies and universities in recent years.It has lower complexity in comparison with H.264/AVC.And very low royalty fee will be charged through one-step license to use AVS including all tools.So AVS is regard as a good and competitive candidate for Chinese satellite TV signal directly to home(DTH).In this paper,the difference of AVS and MPEG-2 is studied.We also analyze alternative approaches,such as direct AVS codec and AVS transcoder,to transmit video stream via satellite,and discuss the design of corresponding transcoding scheme.At last a realization of new generation satellite system with AVS is presented.
关键词:DTH(directly to home);AVS(audio video coding standard);satellite video system;transcoder
摘要:An implementation of the ASI loop output circuit in the DVB-IP gateway is introduced.The implementation which simply realizes the loop output of 270Mbps ASI transport stream is based on the chip PE- 65508,CLC014 and CLC006.The paper first mentions the features of ASI and SPI,then describes the circuit in detail, and gives out the test result at last.In the test,the loop output signal is amplified by 1.8 times.The loop output function is so strong that it can meet the requirements of re-transport after loop output and multilevel loop output.
摘要:With the development of UWB technology,it is possible for wireless mobile multimedia to apply to MANET.The paper designs a new routing protocol for QoS of wireless mobile multimedia.The protocol is described in detail.Consider three parameters of QoS: transmission speed,time delay and jam of link,the protocol is based on the basic routing protocols of MANET-DSR and AODV.The protocol makes use of interconversion between speed and distance in UWB technology.Thus it can not only obtain right route for transmission,but also forecast the halt of link to switch other route.Compared with present routing protocols based on UWB technology,the protocol is more adaptive to the transmission of wireless mobile multimedia.
关键词:ultra wide band(OWB);mobile ad hoc networks;multimedia;routing protocol;quality of service
摘要:For transportation of voice and video over network,uses Microsoft's DirectSound and DirectShow subsystem of DirectX to study and realize network transmission of voice and point-to-point video.Discusses related issues in software development.
摘要:In order to reach the goal of controlling IP-TS(IP transmission) stream multicast through software,the IP multicast controlling module is programmed based on SOAP(simple object access protocol) and after the creation of the MySQL database the PHP(hypertext preprocessor) web script is programmed.Being put into practice,this software can accomplish the commands from the web server sending by users while manipulating the PHP web.
摘要:In our practice of the IPTV test system construction,many technical problems related directly to subscriber quality of experience,networking cost and the scale of operation are encountered.Two key problems among them,i.e.,the CDN node storage problem and the live broadcasting QoS problem,are investigated in this paper.Through performance and cost comparison between several popular CDN storage schemes,the superiority of SAN becomes clear.Channel switch and multicast packet loss,the two critical points involved in QoS of live broadcasting service,are discussed in detail,and practical measures to improve the associated QoS are then obtained based on the analysis.
关键词:internet protocol television(IPTV);content delivery network;storage area network;channel switch
摘要:Stereo matching is currently one of the most active research topics in domain of computer vision.A new feature-point-based stereo matching algorithm is proposed in this paper,in order to match feature points more efficiently and accurately.It is independent of feature detection algorithms,using the scan line as the matching unit,using the robust function as the matching cost function and using the ordering constraint to detect and correct errors in each matching unit in the disparity map.Experiments show that the matching accuracy of this algorithm is better than the traditional NCC(normalized cross-correlation) based stereo matching algorithm and the running time is quite fast.So,it can be used in the independent hardware feature-point-based stereo vision system.
摘要:This paper gives out a watershed-based granular analyzing method for digital color images.The new watershed-based method which improves on the original watershed transform to some extent,can overcome the shortcoming of some conventional methods that they cannot mark the grain exactly and give a exact boundary or mark to every grain in the image.The features of grains can be acquired from these exactly marked areas and then further work can be done more easily.The paper finally applies the proposed method in some digital microscopic mineral images and obtains good results.
摘要:It is the important part of license plate location in license plate recognition technology.In this paper,orients an algorithm of license plate location on the basis of texture characteristic of license plate binary image combined with line scanning.Studies have shown that the algorithm of license plate location can get better result,and the algorithm has better interference immunity.
摘要:In this paper,we shall investigate multi-channel filter banks (MFB,M=3),introduce the analysis,synthesis Equations and transfer Equations of MFB,design its optimal restriction and "least-squares" iteration algorithm,and discuss the application to image decomposition.
摘要:Model construction and control are focus in virtual reality(VR),cooperative controls help objects in VR work together efficiently;based on research in these areas a virtual human-resource-market(VHRM) has been implemented in VRML and Java,at the same time cooperative controls have been applied in it.
关键词:virtual reality(VR);model construction in VR;cooperative control;virtual reality model language(VRML)
摘要:In this paper,a method of camera self-calibration based on the projective invariant is proposed.By the invariability of Cross Ratio in projective transform,we can compute the image of circular points via the cross ratio of circular points and vanish points of intersecting lines;With the image of circular points in three images of the planar pattern from different orientation,we can get IAC(Image of Absolute Conic),because of the invariability of IAC in rigid motion;Using Cholesky factorization on IAC,we get internal parameters of the camera.The result of experiments shows that this method is effective and has high precision.
摘要:The computer supported cooperative work (CSCW) can improve people's working efficiency,the Computer supported collaborative learning(CSCL) is an important application of CSCW in education.The thesis utilized the combining of JAVA and XML to realize the storing and management of graphics learning resources with the computer supported graphics design(CSGD) in CSCL system,ensuring the practicality,reality,continuity of CSCL.
摘要:Redeye is an annoying defect in photographs caused by red light reflected from blood vessels on the retina when a flashgun is used.As the flashgun is close to the lens in a small sized digital camera,the problem becomes even worse.This paper studies detection and removal of red-eyes based on color space analysis,and proposes an automatic red-eye removal technique.Taking into account the prior knowledge of human face,and through experiments on a number of images,an existing approach to face detection is modified.Parameters in an empirical criterion for face detection are determined,resulting in satisfactory segmentation of the skin area.This is used as a basis for further location of eyes and correction of red eyes.Experimental results confirm the effectiveness of the proposed method.
摘要:This paper proposes an edge detection scheme formillimeterwave images, which has low resolution and large noise. It introduces standard gradient strength according to statistical signal processing theory. The basic idea is to employ cubic Bspline as edge detector. Thewavelet transform results reflect the variations for images—along horizontal edges and vertical edges, which are usedwith statistical information together to get the standard gradientmagnitude. For each millimeter wave image a fixed and identical threshold is adopted to detect the image edges roughly. Finally, non2maximum suppression phase and a filter are introduced to get the specific edges. The experiment results demonstrate that the scheme is effective and feasible for detecting the edge of the hidden objects on the background of the human body, which meet the real time requirement in custom.
摘要:Disparity match is the key technology that realizes intermediate view synthesis,multiocular video coding and object recognition.In this paper,a predict recursion disparity match algorithm based on object is put forward.Firstly,foreground area is distinguished from background,and then the disparity of only object area is estimated with information of neighbor block.Experimental results show that this method can obtain disparity map with high quality.
关键词:disparity estimation;predict recursion;image segmentation;based on object
摘要:Face detection and facial features localization are important parts of face analysis techniques.It's task is to search facial features in an image.Although human can finish this work without any effort,it's still a very difficult work for computer.In the past years,this technique has developed vastly and has been successfully applied in the fields of face recognition,pose recognition,expression recognition and face cartoon,etc.In this paper,we use a multi-stage facial features detection algorithm proposed by David Cristinacce and Tim Coote to implement a real-time facial features localization system.Meanwhile,we also improve the origin facial features localization algorithm to run much faster with only a little loss in accuracy.
摘要:Histogram is commonly the statistical information about gray level or other chromatic components of image.Analyzing the histogram of image is a useful method in image processing.Adopting several probability density functions(PDFs) with Gaussian distribution to approximate the histogram is one way for histogram analyzing.But how to acquire the parameters of these distributions remains a hard issue.This paper uses expectation maximization algorithm to estimate the parameters by converting histogram apporximation problem into Guassian mixture models problem in statistics,and then introduces its application in optimal thresholding and histogram component analysis.
摘要:A new kind of filters——Zero SpectralMomentFilters are intriduced, which properties are better than that of the common FIR filters, particularly for flatness on passband and stopband. It includes the known wavelet filters, for examples of conjugate filters and biorthogonal filters.
摘要:Geometric distortion is a simple and effective attack to digital audio watermarking.This paper proposes a novel watermarking scheme,in which the correlation between samples of audio signal is used and watermark signal derived from the local characteristic of host signal is embedded self-synchronously.Simulation results show the proposed scheme,with robustness enough against general signal processing,can effectively resist geometric distortion attacks.
摘要:Digital watermarking technology of video image is a hot topic in present watermarking researching area.In this paper,a adaptive video watermarking scheme based on MPEG-2 motion vectors is proposed,using matrix scrambling computing target group of pictures for watermark clip,combided human visual system,chosing target component motion for watermarks bit.Experimental show that it is robust to special video watermarks attack,and has faintness influence on the visual effect of pictures.
关键词:video watermarking;component motion;matrix scrambling;group of pictures