HEVC中的变换系数熵编码优化算法

在高性能视频压缩标准中,由于熵编码算法的高度复杂性和变换系数的庞大计算量,使得变换系数编码成为影响视频编码效率的瓶颈之一。根据变换系数大都是零或一些绝对值较小的数,以及两类熵编码方法各自的特点,提出一种针对变换系数熵编码的优化算法,通过减少上下文自适应算术编码字节的数量达到提高变换系数熵编码效率的目的。实验结果证明,与标准的HM 10.0算法相比,采用所提算法可以在QP为2,12和22时分别减少37.31%,26.34%和20.63%的编码时间。     

基于SATD准则的HEVC快速帧内预测算法

相比于H.264/AVC视频编码标准,高效视频编码（HEVC）具有更大的压缩率和计算复杂度。HEVC更多的帧内预测模式显著地提升了编码器的率失真性能,但更高的算法复杂度使得编码时间大大增加。通过对典型序列编码的统计分析,提出一种基于变换残差绝对值和（SATD）的帧内预测模式快速选择算法,以降低HEVC帧内预测过程的计算复杂度。所提算法通过比较SATD来加速帧内预测模式的判决,避免了耗时的率失真代价（RD-Cost）计算,并且选择性地降低候选预测模式的数量,以减少总体编码时间。通过与现有其他HEVC快速算法进行比较测试,结果表明,所提算法平均减少了12.90%~19.89%的编码时间（编码器不进行任何优化）,而BD-Rate码率指标仅增加了0.5%~0.96%,几种算法的率失真性能基本一致。     

HEVC视频编码器快速算法

高效视频编码(HEVC)采用了多种先进的编码工具来大幅度提高压缩效率,但同时也带来了高的编码复杂度。考虑到LCU编码时需遍历所有的CU模式、PU模式和TU模式,来确定当前LCU的最优模式组合,其过程耗时较大。针对这点本文分别对CU、PU和TU提出模式判决过程优化算法,同时,每一种PU模式在进行运动矢量搜索时复杂度较高,继而提出运动估计优化方法。实验结果表明,相比于原始的HEVC编码方法,高速模式判决算法可将编码速度提高近15～20倍。     

HEVC快速编码深度选择算法

高效视频编码（High Efficiency Video Coding,HEVC）作为下一代新的视频编码标准,旨在有限网络带宽下传输高质量的网络视频。与现有的视频编码标准相比,高效视频编码具有更高的灵活性和压缩率。编码单元（Coding Unit,CU）是视频编码处理的基本单元,原有的算法通过四叉树递归获取最佳CU深度,在提高视频压缩性能的同时引入了较高的计算复杂度。针对该问题,提出了一种快速编码深度选择算法,该算法利用相邻CU的深度信息计算一个深度预测特征值,通过该特征值进行深度选择,以避免不必要的计算,降低计算复杂度。实验结果表明,该算法在保证视频压缩效果的同时有效降低了计算复杂度。     

HEVC中Merge模式候选决策的探究和优化

针对Merge模式中存在的运算时间长且计算复杂度高的问题,通过对候选PU的合并过程进行研究,提出了优化的候选决策算法。基于时域候选和空域候选决策算法在不同情况下对当前块的影响程度不同,自适应地改变候选决策列表来减少计算复杂度和提高预测精度。与标准算法相比,该算法在视频序列的PSNR值略有增高、码率减少的情况下,编码时间得到有效减少。而且该算法在提高HEVC编码率失真性能的同时有效降低了计算复杂度。     

3D-HEVC深度图帧内预测快速算法

目的 多视点纹理加深度视频（MVD）格式逐渐成为立体视频的主流表现形式之一。新一代高效率立体视频编码（3D-HEVC）继承了HEVC的编码结构并引入一些新的编码技术,导致深度图帧内编码过程具有较高的计算复杂度。针对这一问题,提出了一种深度图帧内编码快速算法。方法 本文算法利用深度图的特征分别对CU分割过程和粗略模式选择（RMD）过程进行优化。首先在四叉树编码结构上,利用基于纹理元的图像分析方法计算编码单元的梯度矩阵,若梯度矩阵中的梯度值之和小于给定的阈值,则终止该CU的分割进程。同时,对大尺寸的PU和小尺寸的PU分别利用纹理特征与粗略模式选择过程中Planar和DC进行低复杂度率失真计算后的最小率失真代价,跳过RMD中角度模式的检查过程。结果 实验结果表明,与原始算法相比,本文算法平均节省40.64%的深度图编码时间,而合成视点的平均比特率仅仅增加了0.17%。本文算法不仅能对平坦的CU跳过不必要的深度决策过程,而且有效地减少了RMD中需要遍历的模式数目,提高了编码器的效率。结论 该算法对CU分割进程和粗略模式选择过程都进行优化,在合成视点的视频质量几乎不变的前提下,有效降低了深度图的帧内编码复杂度。     

浅析HEVC参考帧设置与块划分快速算法研究

HEVC作为新一代的视频编码标准,在传统的混合视频编码框架下,采用了更精细的变换单元、运动融合技术等先进技术,使得HEVC的压缩性能得到很大的提高,尤其针对高分辨率的视频压缩,将达到更加出色的效果。然而,HEVC的压缩性能越强,计算复杂度也就越高,为了降低压缩编码时间,本文探讨了参考帧设置以及块划分的解决方案,通过对HEVC的参考帧集对图像缓存中的已经解码的图像进行管理方法以及对四叉树的块划分策略进行深入分析,结果表明通过参考帧设置以及块划分能够有效的解决HEVC的编码时间,提高了压缩效率。     

高效视频编码中变换跳过模式的快速选择

目的新一代高效视频编码(HEVC)标准采用了灵活的块结构和大量新颖的编码工具,有效提高了视频编码效率。变换跳过模式作为一种新增加的模式,可以有效地提高编码效率,但是也显著地增加了编码的复杂度,增加了实时编码应用的难度。因此提出一种针对变换跳过模式的提前剪枝算法。方法通过分析不同率失真代价下是否选择变换跳过模式的残差块的分布情况,最终选取率失真代价的平方根作为阈值,并建立了量化参数与阈值之间的模型。之后可以根据量化参数提前计算得到阈值大小,减少变换跳过模式的编码次数,从而降低变换跳过模式的复杂度。结果由于最终只需要对少量的块进行变换跳过模式编码,并且使用模型得到经验阈值并不会额外增加复杂度,因此能减少编码器的计算复杂度。实验结果表明,与标准编码器相比,该算法对于不同场景的标准测试序列平均峰值性噪比和平均比特率变化都非常小,平均减少了70%的变换跳过模式编码的次数。结论该变换跳过模式的剪枝算法,选取率失真代价的平方根作为阈值,根据本文模型获取不同量化参数下的经验阈值,对是否需要进行变换跳过模式提前判断。实验结果表明,该算法能在保证视频编码质量的前提下有效地降低由于加入变换跳过模式增加的编码复杂度。     

基于纹理特性的全景视频快速帧内编码算法

为了提高对360°全景视频的编码效率,联合探索专家组研发了基于HEVC的下一代视频编码标准——多功能视频编码标准(Versatile Video Coding,VVC)。相对于HEVC,VVC具有更高的编码效率,但是也引入了更高的时间复杂度。因此为了降低其编码的计算复杂度,提出了一种针对VVC的帧内模式快速决策算法。通过分析图像块的纹理特性来减少帧内编码的候选模式数量从而减少模式选择中的冗余计算。实验结果表明,提出的算法可以节省24. 08%的编码时间同时只有0. 80%的BDrate损失。     

基于HEVC视频标准的熵编码过程

HEVC(High Efficiency Video Coding)作为最新一代国际视频编解码标准,主要针对高清或超高清视频的编解码,在保证视频图像质量的前提下,可以实现将视频码率再次压缩一半。文章分析了HEVC标准中采用的一些关键技术:灵活的编码结构、大尺寸变换单元结构、改进的去方块滤波技术以及HEVC的并行性处理,然后对HEVC标准的码流结构以及编码单元的熵编码过程做了重点分析。最后,简要地对即将进行的硬件设计工作提出了一些方法。     

Simplified algorithms for rate-distortion optimization in high efficiency video coding

HEVC is the latest coding standard to improve the coding efficiency by a factor of two over the previous H.264/AVC standard at the cost of the increased complexity of computation rate-distortion optimization (RDO) is one of the computationally demanding operations in HEVC and makes it difficult to process the HEVC compression in real time with a reasonable computing power. This paper aims to present various simplified RDO algorithms with the evaluation of their RD performance and computational complexity. The algorithms for the simplified estimation of the sum of squared error (SSE) and context-adaptive binary arithmetic coding (CABAC) proposed for H.264/AVC are reviewed and then they are applied to the simplification of HEVC RDO. By modifying the previous algorithm for H.264/AVC, a new simplified RDO algorithm is proposed for modifying the previous algorithm for H.264/AVC to be optimized for the hierarchical coding structure of HEVC. Further simplification is attempted to avoid the transforms operations in RDO. The effectiveness of the existing H.264/AVC algorithms as well as the proposed algorithms targeted for HEVC is evaluated and the trade-off relationship between the RD performance and computational complexity is presented for various simplification algorithms. Experimental results show that reasonable combinations of RDO algorithms reduce the computation by 80–85% at the sacrifice of the BD-BR by 3.46–5.93% for low-delay configuration.     

Improved rate-distortion optimized video coding using non-integer bit estimation and multiple Lambda search

Many modern video encoders use the Lagrangian rate-distortion optimization (RDO) algorithm for mode decisions during the compression procedure. For each encoding stage, this approach involves minimizing a cost, which is a function of rate, distortion and a multiplier called Lambda. This paper proposes to improve the RDO process by applying two modifications. The first modification is to increase the accuracy of rate estimation, which is achieved by computing a non-integer number of bits for arithmetic coding of the syntax elements. This leads to a more accurate cost computation and therefore a better mode decision. The second modification is to search and adjust the value of Lambda based on the characteristics of each coding stage. For the encoder used, this paper proposes to search multiple values of Lambda for the intra-4×4mode decision. Moreover, a simple shift in Lambda value is proposed for motion estimation. Each of these modifications offers a certain gain in RDO performance, and, when all are combined, an average bit-rate saving of up to 7.0% can be achieved for the H.264/AVC codec while the same concept is applicable to the H.265/HEVC codec as well. The extra added complexity is contained to a certain level, and is also adjustable according to the processing resources available.     

Fast motion estimation using priority-based inter-prediction mode decision method in high efficiency video coding

The latest video coding standard, high efficiency video coding (HEVC), is developed to acquire a more efficient coding performance than the previous standard, H.264/AVC. To achieve this coding performance, elaborate coding tools were implemented in HEVC. Although those tools show a higher coding performance than H.264/AVC, the encoding complexity is heavily increased. Especially, motion estimation (ME) requires the most computational complexity because that is always performed on three inter-prediction modes: uni-directional prediction in List 0 (Uni-L0), uni-directional prediction in List 1 (Uni-L1), and bi-prediction (Bi). In this paper, we propose a priority-based inter-prediction mode decision method to reduce the complexity of ME caused by inter-prediction. The proposed method computes the priorities of all inter-prediction modes and decides whether ME is performed or not. Experimental results show that the proposed method reduces the computational complexity of ME up to 55.51% while maintaining similar coding performance compared to HEVC test model (HM) version 10.1.     

基于运动复杂度的快速模式选择算法

H.264/AVC是ITU-T和ISO/IEC联合制定的最新视频压缩标准。它采用了可变块运动估计和率失真优化模式判决。H.264需要对10 种模式进行率失真优化计算才能得到一个宏块的最优编码模式,极大地增加了编码器的复杂度。为了提高模式选择效率,提出了一种基于运动复杂度的快速模式选择算法。实验结果表明,在PSNR仅有微小下降情况下,该算法可大幅提升编码速度。     

H.264/AVC帧间预测模式的快速分类方法

H.264/AVC采用RDO技术进行帧间模式选择，提高了编码效率，但增加了计算复杂度。该文提出了快速分类算法，利用相邻宏块率失真特征以及预测模式之间的时空相关性，缩小了模式选择的范围，降低了帧间预测的复杂度。结果表明，该方法可减少25%~46%的序列编码时间，保持了H.264/AVC的编码性能与视觉效果。     

一种H.264帧内预测模式的快速选择算法

H.264/AVC视频编码标准采用码率—失真度最优化(RDO)算法,以保证更高的压缩效率和更好的图像质量,但同时也大大增加了编码器的复杂度和运算负担。本文提出了一种新型的帧内预测模式的快速选择算法,该算法基于图像内部信息的变化剧烈程度进行判断,并通过方向矢量来减少候选帧内预测模式。试验结果表明,该算法可以将整个编码时间减少到原来采用RDO算法时的40%~50%,而且和RDO算法比较,最终压缩后的视频图像质量基本没有下降。     

一种快速的H.264/AVC帧内预测模式选择算法

H.264/AVC采用RDO技术选择帧内和帧间预测模式以进一步提高编码的效率,取得显著效果,但同时大幅度增加了预测的复杂度。该文提出一种RDO模式下的快速Intra_4x4模式选择算法,该算法根据SATD特征以及相邻块的预测模式之间的相关性,可以预先排除约50%可能性小的Intra_4x4模式,避免了不必要的代价(RD_Cost)计算,从而大幅度降低帧内预测的复杂度,同时基本保持了H.264/AVC的编码性能。     

H.264/AVC帧内预测模式选择算法研究

H.264/AVC采用空间域上的帧内预测技术,进一步提高了编码效率,但由于H.264/AVC支持的帧内预测模式数较多,使预测的复杂度大幅度增加。详细分析了帧内预测模式的选择过程,提出一种快速的率失真优化(RDO)模式下的快速Intra_4×4模式选择算法,该算法根据SATD(SumofAbsoluteTransformedDifference)以及相邻块的预测模式之间的相关性等特征,预先排除了超过65%可能性小的Intra_4×4模式,避免了不必要的计算,从而大幅度降低帧内预测的复杂度,同时基本保持了H.264/AVC的编码性能。     

A fast and low complexity approach for H.264/AVC intra mode decision

As one of the main features of H.264/AVC, intra prediction coding technique acts as a basis for encoding performance and efficiency. In official reference software Joint Model (JM), it employs the rate distortion optimization (RDO) technique to get to the best encoding performance. By full searching (FS) all of the candidate modes under the rule of RDO, peak-signal-noise-rate (PSNR) decreases to a very low level, but at the same time, the complexity of calculation increases a lot. Many researchers had devoted to searching the fast algorithm which can decrease the complexity, and had designed so many excellent and intelligent fast algorithms. In this paper we introduced a low complexity and fast approach for H.264/AVC intra prediction algorithm. The new approach is based on reducing the number of candidate modes for further RDO calculation, and decreasing the computational complicacy. It can decide the interpolation direction accurately by calculating the directional pixel-value differences (DPD) of target block, and then do statistic with the obtained values to choose the most probable modes. The experimental results demonstrate that the proposed algorithm can achieve more than 70% time saving than JM, but only a tiny degradation of encoding performance is brought in.     

Low-complexity heterogeneous architecture for H.264/HEVC video transcoding

High efficiency video coding (HEVC) was developed by the Joint Collaborative Team on video coding to replace the current H.264/AVC standard, which has been widely adopted over the last few years. Therefore, there is a lot of legacy content encoded with H.264/AVC, and an efficient conversion to HEVC is needed. This paper presents a hybrid transcoding algorithm which makes use of soft computing techniques as well as parallel processing. On the one hand, a fast quadtree level decision algorithm tries to exploit the information gathered at the H.264/AVC decoder to make faster decisions on coding unit splitting in HEVC using a Naïve–Bayes probabilistic classifier that is determined by a supervised data mining process. On the other hand, a parallel HEVC-encoding algorithm makes use of a heterogeneous platform composed of a multi-core central processing unit plus a graphics processing unit (GPU). In this way, from a coarse point of view, groups of frames or rows of a frame (both options are possible) are divided into threads to be executed on each core (each of which executes one of the aforementioned classifiers) and, from a finer point of view, all these threads work in a collaborative way on a single GPU to perform the motion estimation process on the co-processor. Experimental results show that the proposed transcoder can achieve a good tradeoff between coding efficiency and complexity compared with the anchor transcoder.