基于神经网络的HEVC帧间预测方法及其硬件研究

相比于H.264,高效视频编码标准(HEVC)提出了许多新技术,提高了编码性能,但是也显著提高了编码复杂度。本文从硬件实现的角度出发,对已有的帧间CU划分预测神经网络的结构进行了多方面的优化,使其参数减少了70%,加法和乘法运算分别减少了60%、58.2%。并对优化后的卷积神经网络参数采用10位定点数方案进行定点化处理,进一步有效减少硬件资源的开支。对比于HEVC参考软件(HM16.5),优化后网络引起的BD-BR和BD-PSNR平均损失为1.718%和-0.056dB,平均节省35%~52%的编码复杂度,并且定点化处理后引起的性能损失可忽略不计。

基于阵列处理器的HEVC数据流图可重构实现

提出一种基于阵列处理器的HEVC算法数据流图可重构实现方法。基于动态重构机制完成不同的划分方式、算法间的灵活切换,采用深度优先贪婪对数据流图划分后子任务时域流水的并行方式对HEVC中典型编码算法的数据流图重新划分后设计合理映射方案,以Sobel算子值为重构依据在阵列处理器上进行帧内预测算法验证。实验结果表明,与块间流水方案实现相比加速比可达14.97,各算法资源利用率及计算速度均有提升,与帧内预测模式选择快速算法相比每个时钟周期可多处理7.1个像素。

Intrusion Detection Model Using Chaotic MAP for Network Coding Enabled Mobile Small Cells

Wireless Network security management is difficult because of the ever-increasing number of wireless network malfunctions,vulnerabilities,and assaults.Complex security systems,such as Intrusion Detection Systems(IDS),are essential due to the limitations of simpler security measures,such as cryptography and firewalls.Due to their compact nature and low energy reserves,wireless networks present a significant challenge for security procedures.The features of small cells can cause threats to the network.Network Coding(NC)enabled small cells are vulnerable to various types of attacks.Avoiding attacks and performing secure“peer”to“peer”data transmission is a challenging task in small cells.Due to the low power and memory requirements of the proposed model,it is well suited to use with constrained small cells.An attacker cannot change the contents of data and generate a new Hashed Homomorphic Message Authentication Code(HHMAC)hash between transmissions since the HMAC function is generated using the shared secret.In this research,a chaotic sequence mapping based low overhead 1D Improved Logistic Map is used to secure“peer”to“peer”data transmission model using lightweight H-MAC(1D-LM-P2P-LHHMAC)is proposed with accurate intrusion detection.The proposed model is evaluated with the traditional models by considering various evaluation metrics like Vector Set Generation Accuracy Levels,Key Pair Generation Time Levels,Chaotic Map Accuracy Levels,Intrusion Detection Accuracy Levels,and the results represent that the proposed model performance in chaotic map accuracy level is 98%and intrusion detection is 98.2%.The proposed model is compared with the traditional models and the results represent that the proposed model secure data transmission levels are high.

The Application of Nicotiana benthamiana as a Transient Expression Host to Clone the Coding Sequences of Plant Genes

Coding sequences (CDS) are commonly used for transient gene expression, in yeast two-hybrid screening, to verify protein interactions and in prokaryotic gene expression studies. CDS are most commonly obtained using complementary DNA (cDNA) derived from messenger RNA (mRNA) extracted from plant tissues and generated by reverse transcription. However, some CDS are difficult to acquire through this process as they are expressed at extremely low levels or have specific spatial and/or temporal expression patterns in vivo. These challenges require the development of alternative CDS cloning technologies. In this study, we found that the genomic intron-containing gene coding sequences (gDNA) from Arabidopsis thaliana, Oryza sativa, Brassica napus, and Glycine max can be correctly transcribed and spliced into mRNA in Nicotiana benthamiana. In contrast, gDNAs from Triticum aestivum and Sorghum bicolor did not function correctly. In transient expression experiments, the target DNA sequence is driven by a constitutive promoter. Theoretically, a sufficient amount of mRNA can be extracted from the N. benthamiana leaves, making it conducive to the cloning of CDS target genes. Our data demonstrate that N. benthamiana can be used as an effective host for the cloning CDS of plant genes.

张量结构和活动值的HEVC-SCC帧内快速算法

针对新一代高效视频编码标准(high efficient video coding,HEVC)逐一划分、逐层对预测模式进行RDO过程计算复杂度高的问题,提出基于结构张量和活动值的HEVC-SCC帧内快速算法。首先利用屏幕内容图像中均匀和小的全局运动区域常用大尺寸单元CU编码,复杂或大的全局运动区域常用小尺寸CU编码的特点,通过提取能够表示CU均匀性的结构张量,研究结构张量与CTU深度划分的联系,在CTU进行遍历不同深度下的编码模式前先对当前深度CU计算结构张量值,通过结构张量值判断是否跳过当前深度下的遍历率失真优化(RDO)的过程。其次利用屏幕内容和自然内容图像纹理特性不同,屏幕内容常含有水平或垂直的边,提出了基于图像活动值的屏幕内容帧内编码模式决策。通过计算图像的编码单元(coding unit,CU)的水平活动值、垂直活动值,对CU进行判别以跳过遍历所有预测模式的过程。所提的算法经过实验测试,在全帧内(all intra)配置下,与SCM-8.8算法相比能减少26.65%的编码时间,而BDBR仅增加1.95%。

Recurrent neural network decoding of rotated surface codes based on distributed strategy

Quantum error correction is a crucial technology for realizing quantum computers.These computers achieve faulttolerant quantum computing by detecting and correcting errors using decoding algorithms.Quantum error correction using neural network-based machine learning methods is a promising approach that is adapted to physical systems without the need to build noise models.In this paper,we use a distributed decoding strategy,which effectively alleviates the problem of exponential growth of the training set required for neural networks as the code distance of quantum error-correcting codes increases.Our decoding algorithm is based on renormalization group decoding and recurrent neural network decoder.The recurrent neural network is trained through the ResNet architecture to improve its decoding accuracy.Then we test the decoding performance of our distributed strategy decoder,recurrent neural network decoder,and the classic minimum weight perfect matching(MWPM)decoder for rotated surface codes with different code distances under the circuit noise model,the thresholds of these three decoders are about 0.0052,0.0051,and 0.0049,respectively.Our results demonstrate that the distributed strategy decoder outperforms the other two decoders,achieving approximately a 5%improvement in decoding efficiency compared to the MWPM decoder and approximately a 2%improvement compared to the recurrent neural network decoder.

Single Photon Detection Technology in Underwater Wireless Optical Communication:Modulation Modes and Error Correction Coding Analysis

This study explores the application of single photon detection(SPD)technology in underwater wireless optical communication(UWOC)and analyzes the influence of different modulation modes and error correction coding types on communication performance.The study investigates the impact of on-off keying(OOK)and 2-pulse-position modulation(2-PPM)on the bit error rate(BER)in single-channel intensity and polarization multiplexing.Furthermore,it compares the error correction performance of low-density parity check(LDPC)and Reed-Solomon(RS)codes across different error correction coding types.The effects of unscattered photon ratio and depolarization ratio on BER are also verified.Finally,a UWOC system based on SPD is constructed,achieving 14.58 Mbps with polarization OOK multiplexing modulation and 4.37 Mbps with polarization 2-PPM multiplexing modulation using LDPC code error correction.

Improved Segmented Belief Propagation List Decoding for Polar Codes with Bit-Flipping

Belief propagation list(BPL) decoding for polar codes has attracted more attention due to its inherent parallel nature. However, a large gap still exists with CRC-aided SCL(CA-SCL) decoding.In this work, an improved segmented belief propagation list decoding based on bit flipping(SBPL-BF) is proposed. On the one hand, the proposed algorithm makes use of the cooperative characteristic in BPL decoding such that the codeword is decoded in different BP decoders. Based on this characteristic, the unreliable bits for flipping could be split into multiple subblocks and could be flipped in different decoders simultaneously. On the other hand, a more flexible and effective processing strategy for the priori information of the unfrozen bits that do not need to be flipped is designed to improve the decoding convergence. In addition, this is the first proposal in BPL decoding which jointly optimizes the bit flipping of the information bits and the code bits. In particular, for bit flipping of the code bits, a H-matrix aided bit-flipping algorithm is designed to enhance the accuracy in identifying erroneous code bits. The simulation results show that the proposed algorithm significantly improves the errorcorrection performance of BPL decoding for medium and long codes. It is more than 0.25 d B better than the state-of-the-art BPL decoding at a block error rate(BLER) of 10^(-5), and outperforms CA-SCL decoding in the low signal-to-noise(SNR) region for(1024, 0.5)polar codes.

HEVC中基于四叉树结构的分块模式编码算法研究

高效率视频编码(High Efficiency Video Coding,HEVC)作为新一代视频编码标准,在当前高清视频压缩传输领域具有广阔的应用前景。为保证视频编码效率,降低编码计算复杂度,文章提出一种基于四叉树结构的分块模式HEVC编码算法。该算法在保证高质量编码性能的基础上,有效降低了编码算法的计算复杂度,提高了编码效率。

Recent Advances in Video Coding for Machines Standard and Technologies

To improve the performance of video compression for machine vision analysis tasks,a video coding for machines(VCM)standard working group was established to promote standardization procedures.In this paper,recent advances in video coding for machine standards are presented and comprehensive introductions to the use cases,requirements,evaluation frameworks and corresponding metrics of the VCM standard are given.Then the existing methods are presented,introducing the existing proposals by category and the research progress of the latest VCM conference.Finally,we give conclusions.

基于多尺度特征融合网络的HEVC帧内编码单元快速划分研究

高效视频编码HEVC显著提高了编码效率,但同时增加了编码复杂度,在基于四叉树结构的编码单元(CU)划分过程中尤为明显,因此研究CU快速划分具有重要意义。多尺度特征融合的网络可以实现HEVC编码单元快速划分。为此,结合U-Net和CU划分特性设计了UcuNet网络,同时为加强不同尺度像素的特征提取,采用了非对称卷积AC和CBAM注意力机制。为更好地训练深度学习模型,收集了不同分辨率的原始视频和对应的编码信息构建出大规模的数据集。最后将模型嵌入到HEVC编码架构中,提前预测CU划分的结果,跳过了原始CU划分方法中递归的率失真优化(RDO)计算过程,从而有效降低CU划分带来的编码复杂度。实验结果表明,对比HEVC官方测试模型(HM16.20),UcuNet在BD-BR仅损失2.63%的情况下,使平均编码时间缩短了68.13%。

Deeply pipelined interpolation architecture for full ultra-HD H.265/HEVC video encoding

Fractional motion estimation(FME) improves the video encoding efficiency significantly. However, its high computational complexity limits the real-time processing capability. Therefore, it is a key problem to reduce the implementation complexity of FME, especially in hardware design. This paper presents a novel deeply pipelined interpolation architecture of FME for the real-time realization of H.265/HEVC full Ultra-HD video encoder. First, a pipelined interpolation architecture together with an elegant processing order is proposed to deal with different search positions in parallel without pipeline stall and data conflict. Second, interpolation results sharing strategies are exploited among search positions to reduce the memory cost. Finally, the structure of the interpolation filter is further optimized for an area efficient implementation. As a result, the proposed design costs 41 917 slice LUTs on the Xilinx Kintex-7 FPGA platform with a 308 MHz working frequency. The measured throughput reaches a record of 1.238 Gpixels/s, which is sufficient for the real-time encoding of 8192×4320@ 30 fps video.

基于FPGA的HEVC去方块滤波硬件设计

去方块滤波(Deblocking Filtering,DBF)是高效视频编码(High Efficiency Video Coding,HEVC)的重要组成部分,能够有效地改善编码图像的主观质量,是提升视频整体编码性能的重要手段之一。但是去方块滤波技术复杂度较高。为解决该问题,设计一种HEVC去方块滤波器的硬件架构,在节省资源消耗的同时,减少处理周期并改善滤波效率。以8×4块为基本滤波单元,从输入像素到输出像素,采用四级流水线的形式进行处理,每处理一个基本滤波单元共花费5个周期。实验结果表明,所设计的去方块滤波器仅需5 212个查找表和1 291个寄存器的逻辑资源消耗,最高可达到215 MHz的工作频率,满足1 080p@60fps的高清视频实时编码。

Multi-Layer Extension of the High Efficiency Video Coding(HEVC) Standard

Multi-layer extension is based on single-layer design of High Efficiency Video Coding(HEVC) standard and employed as the common structure for scalability and multi-view video coding extensions of HEVC. In this paper, an overview of multi-layer extension is presented. The concepts and advantages of multi-layer extension are briefly described. High level syntax(HLS) for multi-layer extension and several new designs are also detailed.

基于T-CNN的3D-HEVC深度图帧内快速编码算法

3D-HEVC标准中引入了具有大面积平坦区域、陡峭边缘和低纹理复杂度特性的深度图。针对深度图编码过程中编码单元(CU)率失真优化导致编码复杂度过高这一问题,本文在分析深度图编码所具有的特点的基础上,构建了深度图划分深度数据集,并提出了一种基于两通道特征传递卷积神经网络(T-CNN)的划分深度预测算法。使用本文提出的算法替换原始编码器中各视点下深度图CU划分模块,可以在一定的率失真性能损失下,将原始HTM-16.0编码器编码时间平均减少76%左右,编码效率得到了显著提升。

Screen Content Coding in HEVC and Beyond

Screen content is video or picture captured from a computer screen typically by reading frame buffers or recording digital display output signals of a computer graphics device.Screen content is an extremely comprehensive and diverse class of content and includes traditional photosensor captured pictures as a small subset.Furthermore,screen content has many unique characteristics not seen in traditional content.By exploring these unique characteristics,new coding techniques can significantly improve coding performance of screen content.Today,more than ever,screen content coding(SCC)is becoming increasingly important due to the rapid growth of a variety of networked computers,clients,and devices based applications such as cloud computing and Wi-Fi display.SCC is the ultimate and most efficient way to solve the data transferring bottleneck problem in these applications.The solution is to transfer screen pixel data between these computers,clients,and devices.This paper provides an overview of the background,application areas,requirements,technical features,performance,and standardization work of SCC.

Deep Learning Aided SCL Decoding of Polar Codes with Shifted-Pruning

Recently,a generalized successive cancellation list(SCL)decoder implemented with shiftedpruning(SP)scheme,namely the SCL-SP-ωdecoder,is presented for polar codes,which is able to shift the pruning window at mostωtimes during each SCL re-decoding attempt to prevent the correct path from being eliminated.The candidate positions for applying the SP scheme are selected by a shifting metric based on the probability that the elimination occurs.However,the number of exponential/logarithm operations involved in the SCL-SP-ωdecoder grows linearly with the number of information bits and list size,which leads to high computational complexity.In this paper,we present a detailed analysis of the SCL-SP-ωdecoder in terms of the decoding performance and complexity,which unveils that the choice of the shifting metric is essential for improving the decoding performance and reducing the re-decoding attempts simultaneously.Then,we introduce a simplified metric derived from the path metric(PM)domain,and a custom-tailored deep learning(DL)network is further designed to enhance the efficiency of the proposed simplified metric.The proposed metrics are both free of transcendental functions and hence,are more hardware-friendly than the existing metrics.Simulation results show that the proposed DL-aided metric provides the best error correction performance as comparison with the state of the art.

基于视频阵列处理器的3D-HEVC视差估计算法并行设计与实现

三维高效视频编码(3D High Efficiency Video Coding,3D-HEVC)中视差估计算法存在处理数据量大、运算时间长和资源消耗大的问题,进一步提高算法执行效率对于3D-HEVC的推广应用具有十分重要的意义。在深入分析视差估计算法的并行性的基础上,基于项目组开发的视频阵列处理器(DPR-CODEC),提出一种新的并行实现方案。在可重构阵列结构中完成了视差估计算法的并行映射、功能仿真和FPGA测试,显著减少了视差估计算法的执行时间。实验结果表明,所提出的并行实现方案相比于串行单PE执行时间节省了59%,基于可编程可重构阵列的并行实现在具有较高的执行效率的同时也具有较好的灵活性。

HEVC帧内率失真优化预测模式的并行流水线硬件设计

近年来,随着人们对视频数据需求的不断增加,视频的分辨率和帧率也在不断地提高,而实时视频序列的压缩编码速度往往受到帧率和分辨率的影响,分辨率和帧率越大,编码所需要的时间越长。为了实现更高分辨率和更高帧率的视频序列实时压缩编码,文中设计了一种新的帧内率失真优化预测模式的并行流水线硬件架构,该架构支持最大64×64编码树单元的帧内预测编码。首先设计了9路预测模式并行方案;然后,按照Z型扫描顺序实现以4×4块为基本处理单元的流水线硬件架构,并复用32×32预测单元的预测数据,用以代替64×64预测单元的预测数据,减少运算量;最后,基于该流水线架构,提出了一种新的哈达玛变换电路,用以实现高效的流水线处理。实验结果表明:在Altera Arria 10系列的现场可编程门阵列上,该9路模式并行架构仅占用75 kb的查找表和55 kb的寄存器资源,主频可以达到207 MHz,完成一个64×64编码树单元的预测仅需要4096个时钟周期,最大能够支持1080 P分辨率99 f/s全I帧的实时编码;与已有设计方案相比,文中方案能够用更小的电路面积实现更高帧率的1080 P实时视频编码。

H.265/HEVC熵解码的分组并行流水线实现

针对高效视频编码(H.265/HEVC)中CABAC熵解码模块的高资源消耗和数据依赖性,设计了一种多路并行的高效FPGA实现结构。根据不同类型语法元素特性,采用分组并行数据调度方法,减少数据处理等待时间和内部存储器访问次数,同时利用流水线技术实现运算加速。评估验证结果表明,熵解码模块吞吐量可以达到1.64位元/时钟周期,满足当下超高清视频实时解码的要求。