采用改进高斯-牛顿法的视频弹性运动估计

运动估计是去除视频时间维冗余的编码技术,而目前通用的平移运动模型无法有效地表示物体的局部非刚性复杂运动.为此,提出一种基于改进高斯-牛顿法的弹性运动估计方法.首先,通过分析初始迭代点对高斯-牛顿迭代结果的影响,采用基于2bit深度像素的均匀搜索预测初始迭代点;其次,通过理论和实验分析发现,不同的迭代步长对弹性运动估计/补偿性能有明显的影响,采用离散余弦变换的低频能量比率估计步长的上限,再利用黄金分割法对步长进行求精.实验结果表明,对于具有不同场景特点的视频序列,该算法始终能够保持较高的估计精度,运动补偿的平均峰值信噪比,比基于块平移模型的全搜索算法和传统弹性运动估计算法分别提高1.73dB和1.42dB.并且,该算法具有更快的收敛速度,一般仅需1~3次迭代就能取得高于传统弹性运动估计和块平移全搜索的峰值信噪比.     

采用自适应缩放系数优化的块匹配运动估计

尽管基于平移模型的快速块匹配运动估计算法在一定程度上解决了高计算量的问题,但却是以牺牲运动补偿质量为代价的,而高阶运动模型尚存在计算量高、收敛不稳定的不足.通过实验统计发现,视频中约有56.21%的块包含缩放运动,进而得出缩放运动是除平移运动外最主要的视频运动形式的结论.进而借助双线性插值,在传统的块平移模型中引进一个缩放系数,将运动补偿误差表示为该缩放系数的一元二次函数,利用韦达定理推导出1D缩放运动下最佳缩放系数的计算方法,并将其进一步推广到2D等比例缩放运动的情况下.在此基础上,提出了一种采用自适应缩放系数优化的快速块匹配运动估计算法.该算法以菱形搜索计算平移矢量,再用自适应缩放系数确定待预测块的最佳匹配块.在33个标准测试视频上的实验结果表明,与基于平移模型的块匹配全搜索和快速菱形搜索相比,该算法的平均运动补偿峰值信噪比（peak signal-to-noise ratio,简称PSNR）分别提高了0.11dB和0.64dB,计算量比全搜索下降了96.02%,略高于菱形搜索;与基于缩放模型的运动估计相比,该算法的平均峰值信噪比较之3D全搜索下降了0.62dB,但是比快速3D菱形搜索提高了0.008dB,而计算量仅分别为两者的0.11%和3.86%,并且无需向解码端传输缩放矢量,能够实现编、解码端的自同步,不会增加边信息的码流开销.此外,该自适应缩放系数计算方法还可与菱形搜索以外的其他快速块匹配运动估计相结合,提高其运动补偿质量.     

结合低位深像素预测起点的小波域运动估计

提出一种用2b深度的像素预测搜索起点的快速小波域运动估计算法.首先,将像素深度的转换形式化为区间分划和区间映射,采用非均匀量化求解区间分划的初始阈值,再用隶属函数计算量化阈值并完成区间映射,从而获得位深度为2b的视频表示;其次,设计了非均匀的搜索起点分布模板,并以此为基础提出一种基于2b深度像素的搜索起点预测算法;最后,以搜索起点为中心,进一步采用改进的低频子带平移运动估计算法MLBSSME在较小的窗口内完成搜索.实验结果表明,对于具有不同场景特点的视频序列,算法始终能保持较高的估计精度,运动补偿的平均峰值信噪比较之低频子带平移运动估计和直接子带运动估计算法高0.41dB和1.43 dB,比空间域全搜索降低0.07 dB.但是,算法的计算量仅相当于空间域全搜索的4.66％、低频子带平移运动估计的4.62％、子带直接运动估计的22.70％.     

基于H.264的自适应快速运动估计算法

提出了一种基于H.264的快速自适应运动估计算法。利用视频序列的统计特性和SAD在空间上的相关性提前结束搜索;根据块的运动量来确定搜索方式;提出了一种新的非对称综合的菱形搜索模式,与快速全搜索算法相比较,该算法的峰值信噪比只下降了0.017 dB,码率只上升了0.45%,而搜索速度提高了15倍;与H.264参考程序中的运动估计算法相比,在图像质量相同的情况下,搜索速度提高了35%。     

基于视频压缩的多分辨率预测菱形搜索算法

为了减小视频压缩编码标准中运动估计算法的计算复杂度和提高运动补偿的准确性,考虑到视频图像质量和算法运行时间两者之间的关系,提出一种多分辨率预测菱形运动估计搜索算法,利用同一视频图像中的相邻宏块运动矢量相似的特征,预测当前块的搜索起点,采用大小不同的搜索匹配宏块,减少搜索范围和搜索点数目。实验结果表明,该算法与经典菱形搜索算法相比,搜索时间平均减少了0.5 ms,信噪比平均提高了0.5 dB。     

一种高效率的快速块匹配运动估计算法

采用菱形搜索算法对各种视频测试序列中运动矢量的研究,基于H.264视频编码标准提出了一种快速块匹配运动估计算法。它是以图像中相邻宏块之间的时空相关性为前提,结合了分布式菱形搜索,预测搜索和中止阈值等一系列技术而提出的,试验结果表明该算法在运算速度方面优于菱形搜索,而获得与全搜索相当的峰值信噪比。     

基于块分类的深度图运动估计及其并行实现

针对三维高效视频编码(3D high efficiency video coding,3D-HEVC)深度图在运动估计过程中没有充分利用平坦区域数据特性而导致的计算量大、编码时间长等问题,提出一种基于块分类的深度图运动估计优化方法.依据编码块像素分布特征,将其分类为平坦或者边缘区域,自适应地为不同类型编码块分配不同的搜索算法,减少整体计算开销.实验结果表明,优化后的算法相比于全搜索算法,平均峰值信噪比(peak signal to noise ratio,PSNR)的损耗仅为0.0870 dB;相比于TZSearch(test zone search)算法,运动搜索次数平均减少12.04％.为进一步提升编码速度,基于视频阵列处理器对优化后的算法设计并行实现方案.实验结果表明,所设计的并行方案串/并平均加速比达到2.8940.     

视频图像运动估计中的一维块匹配算法

运动估计是视频图像压缩和视频图像修复等领域的基础问题,传统的块匹配法搜索质量较好,但搜索速度不够快.针对传统块匹配法搜索速度上的不足,提出一种快速的一维块匹配运动估计算法.首先对运动矢量正交分解,使用特殊权重系数矩阵对二维匹配块做降维处理,得到2组一维特征矩阵;然后选择一维三步搜索法作为搜索策略,最小绝对误差和准则作为匹配准则,使用2组一维特征矩阵搜索匹配运动矢量的2个分量;最后将分量组成完整的运动矢量.通过多组对比实验的结果表明,该算法在保证定量评价PSNR的前提下,显著提升运动估计的搜索速度,视频清晰度越高、匹配块像素尺寸越大,运动估计搜索速度提升越明显.     

结合遗传分割的多分辨率预测匹配算法

为了减小视频压缩算法的计算复杂度和提高恢复图像的准确性,提出一种新的多分辨率预测搜索匹配算法.利用遗传算法首先对视频图像进行分割,划出分界线,然后判断所选择块内是否有边界,若没有,不做运动估计匹配,直接将当前块运动矢量置为零;若有,则进行多分辨率预测估计.预测搜索起点,采用大小不同的搜索匹配宏块,减少搜索点数目.实验结果表明,该算法与经典菱形搜索算法相比,搜索时间平均减少7.6ms,信噪比平均提高1.2dB.     

基于免疫克隆选择的块匹配运动估计

运动估计是视频压缩编码中的关键技术.从运动矢量的特点出发,采用搜索点预测、Gray码编码以及有效的迭代终止准则等策略,提出了基于免疫克隆选择的块匹配运动估计.该方法将块匹配运动估计问题的性质与免疫克隆选择算法所具有的全局搜索特性、解的多样性和不易早熟的特点相融合,在能够获得接近全搜索方法所得到的平均峰值信噪比的前提下,使得平均搜索点数大为降低.仿真实验结果表明,在大多数序列上,该算法都比已有的快速搜索算法具有更高的性能和更少的平均搜索点数.同时,该算法适用面广,对大运动和小运动序列都能得到较好的效果.     

利用反向投影的flash场景自适应视频编码算法

针对闪光造成的光照变化会导致视频帧之间巨大的强度差异问题,提出利用反向投影的flash场景自适应视频编码算法;根据直方图差异提取闪光和非闪光帧,相应地为每个帧分配适当的编码类型,并在加权预测(WP)参数集确定中采用运动向量导数,通过反向投影保证flash场景的全局一致性;实验结果显示,提出的算法在Lena、Peppers、Building、Baboon、Nestling 5个视频上的峰值信噪比(PSNR)值分别可高达32.31 dB、34.14 dB、34.76 dB、34.94 dB、35.05 dB,非常接近原始图像的PSNR;相比其他几种加权预测算法,提出的算法在PSNR及计算复杂度方面均获得了更加优越的编码性能。     

Efficient video denoising based on dynamic nonlocal means

A video denoising algorithm, which is based on dynamic nonlocal means (DNLM), is developed. Firstly, the standard nonlocal means and Kalman filtering are reviewed briefly. Then, using the idea of nonlocal means and linear minimum variance fusion, a weighted translational motion model without the explicit motion estimation and a weighted translational observation model are proposed to modify the state transition and observation equations. Finally, the overall dynamic denoising algorithm under the Kalman filter framework is presented. The main contribution of our work is a dynamic nonlocal means algorithm that is developed for video denoising under the Kalman filtering framework. In this algorithm, all computations are pixel-wise and it is easy to realize an efficient recursive algorithm for real-time processing. Experimental results for different test videos demonstrate the power of proposed method based on peak signal-to-noise-ratio (PSNR), structural similarity (SSIM) and motion-based video integrity evaluation index (MOVIE). The proposed method performs better than SNLM with the average PSNR gain of 2.33 dB, and outperforms SEQWT, 3DWTF and IFSM with the average SSIM gains of 0.033, 0.0087 and 0.049. It has competitive performance with STA, WRSTF and 3DSWDCT, but needs lower computational cost. Though the proposed DNLM is not competitive with several state-of-the-art video denoising algorithms such as VBM3D, K-SVD, 3D-Patch, and ST-GSM, it may be anyway valuable to readers working in this field as a source of inspiration for their further researches.     

Improved Normalized Partial Distortion Search With Dual-Halfway-Stop for Rapid Block Motion Estimation

Motion estimation is a critical yet computationally intensive task for video encoding. In this paper, we present an enhancement over a normalized partial distortion search (NPDS) algorithm to further reduce block matching motion estimation complexity while retaining video fidelity. The novelty of our algorithm is that, in addition to the halfway-stop technique in NPDS, a dual-halfway-stop (DHS) method, which is based on a dynamic threshold, is proposed, so that block matching is not performed against all matching candidates. An adaptive search range (ASR) mechanism based on inter block distortion further constrains the searching process. Simulation results show that the proposed algorithm has a remarkable computational speedup when compared to that of full search and NPDS algorithms. Particularly, it requires less computation by 92-99% and encounters an average of only 0.08 dB PSNR video degradation when compared to that of full search. The speedup is also very significant when compared to that of fast motion estimation algorithms. This paper describes our work that led to our joint video team (JVT) adopted contribution (included in software JM 10.1 onwards) as well as later enhancements, collectively known as simplified and unified multi-hexagon search (SUMH), a simplified fast motion estimation.     

二值alpha平面辅助的视频对象快速运动估计算法

提出了一种任意形状视频对象的快速运动估计方法.详细分析了alpha平面在视频对象的快速运动估计过程中起到的指导性作用,采用边界扩展和边界掩码技术,提出了一种新的二值alpha平面匹配衡量准则WBAMC (weighted binary alpha-plane matching criterion).结合优先搜索策略,提出了二值alpha平面辅助的视频对象快速运动估计算法BAAME(binary alpha-plane assisted motion estimation),.首先,利用alpha平面和WBAMC准则,将边界宏块的搜索范围缩小至两个搜索起点的单调区域,再采用传统的快速运动估计算法确定其运动向量;然后,用边界宏块的运动向量预测内部宏块的搜索起点;最后,采用快速运动估计算法搜索内部宏块的运动向量.这种方法可与多种空间域和频率域运动估计算法相结合,有效地应用于基于对象的视频编码器中.实验结果表明,对于多种类型的标准测试视频流,BAAME算法始终能够保持较高的估计精度和主观质量,运动补偿的平均PSNR(peak signal-to-noise ratio)较DS(diamosd search)和PSA(pdoety search algorithm)(BAAS(binary alpha-plant assisted search) DS)高出0.1dB～0.SdB,略低于FS(full search),但是其计算复杂度与FS相比降低了20倍.     

使用动态模型的UMHexagonS算法优化

H.264取得了很好的编码效率,但是也具有很高的计算复杂度。对H.264中的非对称十字形多层次六边形格点搜索算法（UMHexagonS）进行了优化,分别对提前终止阈值、搜索窗口大小以及搜索模式提出了3种动态模型,提高了算法的自适应性。对六种不同运动程度的视频序列进行了测试,实验结果表明,优化后的算法相对于原来的UMHexagonS算法平均减少了21.67%的编码时间以及47.49%的运动估计时间,同时只有0.02的峰值信噪比下降以及1.69%的比特率增加。     

Content-Aware Prediction Algorithm With Inter-View Mode Decision for Multiview Video Coding

3-D video will become one of the most significant video technologies in the next-generation television. Due to the ultra high data bandwidth requirement for 3-D video, effective compression technology becomes an essential part in the infrastructure. Thus multiview video coding (MVC) plays a critical role. However, MVC systems require much more memory bandwidth and computational complexity relative to mono-view video coding systems. Therefore, an efficient prediction scheme is necessary for encoding. In this paper, a new fast prediction algorithm, content-aware prediction algorithm (CAPA) with inter-view mode decision, is proposed. By utilizing disparity estimation (DE) to find corresponding blocks between different views, the coding information, such as rate-distortion cost, coding modes, and motion vectors, can be effectively shared and reused from the coded view channel. Therefore, the computation for motion estimation (ME) in most view channels can be greatly reduced. Experimental results show that compared with the full search block matching algorithm (FSBMA) applied to both ME and DE, the proposed algorithm saves 98.4–99.1% computational complexity of ME in most view channels with negligible quality loss of only 0.03–0.06 dB in PSNR.      

A new fast motion estimation algorithm using fast mode decision for high-efficiency video coding standard

High-efficiency video coding is the latest standardization effort of the International Organization for Standardization and the International Telecommunication Union. This new standard adopts an exhaustive algorithm of decision based on a recursive quad-tree structured coding unit, prediction unit, and transform unit. Consequently, an important coding efficiency may be achieved. However, a significant computational complexity is resulted. To speed up the encoding process, efficient algorithms based on fast mode decision and optimized motion estimation were adopted in this paper. The aim was to reduce the complexity of the motion estimation algorithm by modifying its search pattern. Then, it was combined with a new fast mode decision algorithm to further improve the coding efficiency. Experimental results show a significant speedup in terms of encoding time and bit-rate saving with tolerable quality degradation. In fact, the proposed algorithm permits a main reduction that can reach up to 75 % in encoding time. This improvement is accompanied with an average PSNR loss of 0.12 dB and a decrease by 0.5 % in terms of bit-rate.     

DCT-domain coder for digital video applications

In this paper, we present an effective DCT-domain video encoder architecture that decreases the computational complexity of conventional hybrid video encoders by reducing the number of transform operations between the pixel and the DCT domains. The fixed video encoder architecture (such as a fixed DCT block of 8 × 8 size) and a huge number of DCT/IDCT transforms performed during the video encoding process limit the minimum possible computational load of conventional video encoders. In this study, we solve this problem by developing a flexible video encoder architecture, which reduces video encoder computational complexity by performing low-resolution coarse-step motion estimation operations in the DCT domain. When a high level of motion activity is detected, the video encoder slightly increases the computational complexity of the motion estimation operation by computing fine-search block matching for a small-size search window in a reference frame. The proposed DCT-domain video encoder architecture is based on the conventional hybrid coder and on a set of fast integer composition and decomposition DCT transforms. The set of transforms implements a technique for estimation of DCT coefficients of a block that is partitioned by the sub-blocks. Experimental results of this method were compared with the results of the conventional hybrid coder in terms of PSNR quality and computational complexity. This comparison shows that the computational complexity of the proposed encoder is lower by 26.8% with respect to the conventional hybrid video coder for the same objective PSNR quality.