Table 1 - uploaded by Yi-Ming Lin
Content may be subject to copyright.
Source publication
In H.264/AVC, the context-based adaptive variable length coding (CAVLC) is used for lossless compression. Direct table-lookup implementation requires higher cost because it employs a larger memory to produce the encoded results. A carefully reduced table is constructed and used for encoding in the design. With the helps of symbol mapping and codewo...
Context in source publication
Context 1
... applying our technique to the implementation of three VLC tables, we can calculate the practical memory required for coeff_token. Table 1 shows the comparisons of memory sizes needed for our CAVLC implementation and others [9][10][11][12][13] where a pointer lookup technique is used in [9], a low-cost section- based VLC technique is used in [10], and the direct table-lookup technique is used in [11][12][13]. Evidently, the required memory of our method is less than the others. ...
Citations
... However, as the clock period shrinks, the CAVLC coder often encounters tight timing constraints due to the presence of a significant amount of look-up tables (LUTs). Since the use of LUTs introduces considerable circuit area and delay overhead, there have been many approaches to completely remove LUTs [4], [5], [6] or to at least reduce them [7] . However, those architectures still bring out a considerable delay; thereby, they are not suitable for high-end H.264 encoders supporting high bitrates or full HD resolution. ...
In H.264/AVC and the variants, the coding of context-based adaptive variable length codes (CAVLC) requires demanding operations, particularly at high bitrates such as 100 Mbps. This letter presents two approaches to accelerate the coding operation substantially. Firstly, in the architectural aspect, we propose component-level parallelism and pipeline techniques capable of processing high-bitrate video data in a macroblock (MB)-level pipelined codec architecture. The second approach focuses on a specific part of the coding process, i.e., the residual block coding, in which the coefficient levels are coded without using look-up tables so we minimize the pertaining logic depth in the critical path, and we achieve higher operating clock frequencies. Additionally, two coefficient levels are processed in parallel by exploiting a look-ahead technique. The resulting architecture, merged in the MB-level pipelined codec system, is capable of coding up to 100 Mbps bitstreams in real-time, thus accommodating the real-time encoding of 1080p@60 Hz video.
... The goal is to attenuate the high frequencies, i.e. those to which the human eye is less sensitive. The quantized coefficients are then zig-zag parsed following their magnitude and finally coded using an entropy coder, such as Run-Length Coding (RLC) [130], Huffman coding [102], Context-based Adaptive Variable-Length Coding (CAVLC) [119] or Context-based Adaptive Binary Arithmetic Coding (CABAC) [128]. The motion-vector fields are losslessly coded, using the Variable Length Coding (VLC) [90] method. ...
The recent progress in wavelet-based video coding schemes led to the emergence of a new generation of scalable video codecs, whose performance is comparable to the best hybrid codecs. The t+2D wavelet-based schemes exploit the temporal interframe redundancy by applying an open-loop temporal wavelet transform over the frames of a video sequence. Temporally filtered subband frames are further spatially decomposed and can be encoded by different entropy-coding algorithms. Because of their inherent multiresolution signal representation, wavelet-based coding schemes have the potential to support temporal, spatial and SNR scalability. This is the reason for which we have chosen the scalable lifting-based wavelet-coding paradigm as the conceptual development framework for this thesis. The objective of this work consists of the analysis and design of a scalable video coding system. In a first time, we are interested in the construction and optimization of new motion-compensated temporal coding schemes, in order to increase the efficiency of objective and subjective video coding. Moreover, we describe a better representation of temporal subbands, by using anisotropic spatial decompositions, for a better capture of the orientation of spatial details. Finally, we propose a method for improving the entropy coding by designing a graph-based solution, in order to optimize the minimization of the Lagrangian rate-distortion functional.
Le transcodage est un élément clé dans la transmission vidéo permettant à une séquence vidéo de passer d'un type de codage à un autre afin de s'adapter au mieux aux capacités de transport d'un canal de transmission. L'intérêt de ce type de traitement est de faire profiter un maximum d'utilisateurs possédant des terminaux variés dont la résolution spatiale, la résolution temporelle affichable, et le type de canal utilisé pour accéder au média varient fortement, et cela à partir d'une seule source de qualité et résolution maximale, stockée sur un serveur, par exemple. Le transcodage est adapté dans les cas où l'on souhaite envoyer une séquence vidéo vers un destinataire et dont le chemin serait constitué de divers canaux de transmission. Nous avons réalisé un transcodeur par requantification ainsi qu'un transcodeur par troncature. Ces deux méthodes ont été comparées et il apparait qu'en termes de qualité d'image l'une ou l'autre de ces méthodes est plus efficace selon le contexte. La suite de nos travaux consiste en l'étude du standard scalable dérivé de H.264 AVC, le standard SVC (Scalable Video Coding). Nous avons souhaité étudier un transcodeur en qualité, mais aussi en résolution spatiale qui permettra de réécrire le flux SVC en un flux AVC décodable par les décodeurs du marché actuel. Cette transposition est réalisée grâce à une architecture reconfigurable permettant de s'adapter aux nombreux types de flux pouvant être conformes au standard SVC d' H.264. L'étude proposée a aboutie à une implémentation partielle d'un transcodeur du type SVC vers AVC. Nous proposons dans cette thèse une description des implémentations de transcodage concernant les formats AVC puis SVC
In this paper a new Context-Adaptive Variable Length Coding (CAVLC) encoder architecture is proposed aimed to be implemented in embedded systems and field programmable logic. The design proposes novel Arithmetic Table Elimination (ATE) techniques, along with a table compression technique applied to those tables that cannot be eliminated by arithmetic manipulations. These approaches allows to halve the total number of tables requested by CAVLC algorithm and bring to an overall memory saving of about 87% with respect to an unoptimized implementation of the tables. Computational performances of the encoder have been improved by increasing the degree of parallelism through the use of priority cascading logic. With the proposed approaches the CAVLC encoder is capable of real time compression of 1080p HDTV video streams, coded in YCbCr 4:2:0, when it is implemented with a low-end Xilinx Spartan 3 FPGA, where the encoder achieves an operation frequency of 63MHz and requires an area occupancy of 2200 LUTs.
In this study, a new context-adaptive variable length-coding encoder architecture is proposed particularly aimed to be implemented with field programmable logics (FPL) like FPGAs. The design implements different approaches in order to minimise the area cost as well as to speed up the coding efficiency, which allows real-time compression of 1080 p video streams coded in YCbCr 4:2:0 format. Priority cascading logics have been implemented in order to increase the parallelisation degree of the pre-coding stage, thus favouring the limitation of the number of clock cycles needed for the extraction of symbols from the input data, whereas the employment of the arithmetic table elimination technique has allowed a large-area reduction of the encoder thanks to the elimination of 18 of the 38 tables needed for the encoding stage. The design achieves real time elaboration with an operation frequency of 63 MHz and occupies 2200 look-up table (LUT)s when implemented on a low-cost, low-end XILINX Spartan 3 FPGA, thus overcoming the most recent FPL implementation and making this encoder quite comparable both in terms of area and speed with some recently proposed ASIC implementations, so that it turns out to be a valid alternative also for application specific implementations.
In the H.264 standards, the default block size is reduced to 4times4. As a result, it becomes quite possible that a quantized residue block in the current P-frame is matched (either identical or very similar) to one of the previously-coded blocks. Thus, the encoding of any current block that contains a few non-zero coefficients could be made more efficient if we transmit the relevant information of the matched block instead of coding it with standard variable-length coding. To this end, all previously-coded blocks should be carefully maintained so as to construct a codebook. In this paper, we study how such codebook is formed initially, how to use this newly formed codebook to encode some blocks in the current frame, and how this codebook is updated dynamically during the encoding process. We also propose various modifications on the H.264 encoding table in order to maintain the synchronization between the modified encoder and decoder after the lookup-table is added to the system. Testing is performed based on different codebook setup. Experiment results show that the bit-count per frame can be saved by about 6-9% after applying the proposed codebook technique with very little degradation on the video quality.
In this paper, architecture of variable length code decoder (VLD) with combination of software and hardware for AVS and H.264 dual standards is proposed. Fixed length code, unsigned or signed Exp-Golomb code in AVS and H.264 can respectively be decoded by software. Hardware is in charge of context-based adaptive 2D-VLC (CA-2D-VLC) code decoding for AVS, CABAC and CAVLC decoding for H.264. In order to share as more modules as possible between two standards, a module named VLD is devised which can decode 0-th Exp-Golomb code. A new decoding method for k-th (k>0) Exp-Golomb code is presented as the combination of 0-th Exp-Golomb code decoding and fixed length code decoding. Control and information streams are designed to make connection between software and hardware. Finally, parsing results of syntax elements from AVS and H.264 bitstream are output correctly.
In this paper, a combined kernel architecture for efficiently decoding the residual data in the H.264/AVC baseline decoder is proposed. The kernel architecture in the H.264/AVC decoder consists of context-based adaptive variable length code (CAVLC) decoder, inverse quantization (IQ), and inverse transforms (IT) units. Since the decoding speeds of these kernel units vary with data, traditional methods require data buffers between these units. The first proposed architecture efficiently combines CAVLC decoding and IQ procedures. The multiple 2-D transforms architecture is applied to all inverse transforms, including the 4times4 inverse integer transform, the 4times4 inverse Hadamard transform and the 2times2 inverse Hadamard transform, to attain fewer gate counts than those of existing transform designs. Simulation results show that the total number of gates is 14.1 k and the maximum operating frequency is 130 MHz. For real-time requirements, in the worst case, the proposed architectures can achieve the operation speed of the H.264/AVC decoder up to 4VGA@30 frames/sec in 4:2:0 format.
Modern medical imaging requires storage of large quantities of digitized clinical data. Due to the constrained bandwidth and storage capacity, however, a medical image must be compressed before transmission and storage. Among the existing compression schemes, Integer based discrete cosine transform coding is one of the most effective strategies. Image data in spatial domain is transformed into spectral domain after transformation to attain higher compression gains. Based on the quantization strategy, coefficients of low amplitude in the transformed domain are discarded using a threshold technique: set partitioning in hierarchical trees (SPIHT) where in only significant coefficients are retained to increase the compression ratio without inducing salient distortion. In this paper, we used two advanced coding engines context adaptive variable length coding (CAVLC) and embedded block coding with optimal truncation (EBCOT) to code the significant coefficients. Recording or transmitting the significant coefficients instead of the whole coefficients achieves the goal of compression.. Simulations are carried out on different medical images, which include CT skull, angiogram and MR images. Consequent images demonstrate the performance of two coding engines in terms of PSNR & bpp without perceptible alterations. Simulation results showed that the Integer DCT with SPIHT and CAVLC coding has shown better results compared to JPEG & JPEG2000 schemes. Therefore, our proposed method is found to preserve information fidelity while reducing the amount of data.
In this paper, we implement the configurable processor for CAVLC function module of a H.264/AVC baseline profile decoder as the starting point to implement the H.264/AVC decoder system in a multiprocessor platform. The requirements of the implementations are the low-power processor and speed optimized algorithms tailored to the processor architecture. An arithmetic formula mapping method for fast CAVLC algorithms and a dual-issue VLIW processor architecture with custom instructions are proposed. The experiment results show that the synthesized processor has about 75 K gates and can carry out the decoding of 30-fps CIF (352x288 pixels) images around 120 Mega cycles.
