基于块编码特点的压缩视频质量增强算法

针对现有压缩视频质量增强算法未能充分利用压缩视频特点的问题,研究了视频编码与压缩视频质量增强任务之间的本质关系,并针对性地设计了一种基于三维卷积神经网络(3D convolutional neural network, 3D-CNN)的非对齐压缩视频质量增强算法。实验结果表明:相较于高效视频编码(high efficiency video coding, HEVC)标准H.265,所提算法在低延迟(low delay, LD)配置下且量化参数(quantization parameter, QP)为37时,峰值信噪比(peak signal-to-noise ratio, PSNR)提升了0.465 2 dB;相较于数据压缩会议(data compression conference, DCC)中提出的多帧引导的注意力网络(multi-frame guided attention network, MGANet)方法,该算法PSNR的增长量提升了15.1%。

基于HEVC屏幕图像编码的哈希表的优化算法

仿2维匹配算法对屏幕图像中的非连续色调区域有很好的压缩性能,但该算法中哈希表的空间开销较大,不利于硬件实现。为了减小哈希表的空间,通过对原算法优化提出了一种3字节计算哈希值方法,将源数据看作是一个由以YUV三元组为元素组成的数据集合,然后以YUV三元组为单位计算哈希值,这样不但减少了哈希值的计算量,而且使哈希表的存储空间得到很大的节省。实验结果表明,3字节计算哈希值方法使哈希表的存储空间减少为原算法的1/3,所测试屏幕图像的BD-rate性能也有所提高。

An Introduction to High Efficiency Video Coding Range Extensions

High Efficiency Video Coding (HEVC) is the latest international video coding standard, which can provide the similar quality with about half bandwidth compared with its predecessor, H.264/MPEG?4 AVC. To meet the requirement of higher bit depth coding and more chroma sampling formats, range extensions of HEVC were developed. This paper introduces the coding tools in HEVC range extensions and provides experimental results to compare HEVC range extensions with previous video coding standards. Ex?perimental results show that HEVC range extensions improve coding efficiency much over H.264/MPEG?4 AVC High Predictive profile, especially for 4K sequences.

Architecture of Intra Prediction for High Efficiency Video Coding

This paper explains intra prediction method for High Efficiency Video Coding(HEVC).Intra prediction removes correlation of adjacent samples in spatial domain.Intra predictor requires reference images which are stored in external memory.Memory access is required frequently in process of intra prediction.The proposed architecture can reduce external memory access by optimized internal buffer.

Parallelization of intra prediction algorithm based on array processor

For the characteristics of intra prediction algorithms, the data dependence and parallelism between intra prediction models are first analyzed. This paper proposes a parallelization method based on dynamic reconfigurable array processors provided by the project team, and uses data level parallel(DLP) algorithms in multi-core units. The experimental results show that Y-component of peak signal to noise ratio(Y-PSNR) is improved about 10 dB and the time is saved 63% compared with high-efficiency video coding(HEVC) test model HM10.0. This method can effectively reduce codec time of the video and reduce computational complexity.

Reconfigurable design of deblocking filter for variable block sizes

Based on the flexible quadtree partition structure of coding tree units(CTUs),the deblocking filter(DBF)in high efficiency video coding(HEVC)consumes a lot of resources when implemen-ted by hardware.It is difficult to achieve flexible switching between different sizes of coding blocks.Aiming at this problem,a reconfigurable implementation of DBF is proposed.Based on the dynamic programmable reconfigurable video array processor(DPRAP)with context switch reconfiguration mechanism,the runtime flexible switching of two coding block sizes is realized.The experimental results show that the highest work-frequency reaches 151.4 MHz.Compared with the dedicated hardware architecture scheme,the resource consumption can be reduced by 28.1%while realizing the dynamic switching between algorithms of two coding block sizes.Compared with the results of HM16.0,by using a complete I-frame for testing,the average peak signal-to-noise ratio(PSNR)of the reconfigurable implementation proposed in this paper has increased by 3.0508 dB,the coding quality has improved to a certain extent.

High-Efficiency Video Coder in Pruned Environment Using Adaptive Quantization Parameter Selection

The high-efficiency video coder(HEVC)is one of the most advanced techniques used in growing real-time multimedia applications today.However,they require large bandwidth for transmission through bandwidth,and bandwidth varies with different video sequences/formats.This paper proposes an adaptive information-based variable quantization matrix(AIVQM)developed for different video formats having variable energy levels.The quantization method is adapted based on video sequence using statistical analysis,improving bit budget,quality and complexity reduction.Further,to have precise control over bit rate and quality,a multi-constraint prune algorithm is proposed in the second stage of the AI-VQM technique for pre-calculating K numbers of paths.The same should be handy to selfadapt and choose one of the K-path automatically in dynamically changing bandwidth availability as per requirement after extensive testing of the proposed algorithm in the multi-constraint environment for multiple paths and evaluating the performance based on peak signal to noise ratio(PSNR),bit-budget and time complexity for different videos a noticeable improvement in rate-distortion(RD)performance is achieved.Using the proposed AIVQM technique,more feasible and efficient video sequences are achieved with less loss in PSNR than the variable quantization method(VQM)algorithm with approximately a rise of 10%–20%based on different video sequences/formats.

Long-term prediction for hierarchical-B-picture-based coding of video with repeated shots

The latest video coding standard High Efficiency Video Coding （HEVC） can achieve much higher coding efficiency than previous video coding standards. Particularly, by exploiting the hierarchical B-picture prediction structure, temporal redundancy among neighbor frarnes is eliminated remarkably well. In practice, videos available to consumers usually contain many repeated shots, such as TV series, movies, and talk shows. According to our observations, when these videos are encoded by HEVC with the hierarchical B-picture structure, the temporal correlation in each shot is well exploited. However, the long-term correlation between repeated shots has not been used. We propose a long-term prediction （LTP） scheme to use the long-term temporal correlation between correlated shots in a video. The long-term reference （LTR） frames of a source video are chosen by clustering similar shots and extracting the representative frames, and a modified hierarchical B-picture coding structure based on an LTR frame is introduced to support long-term temporal prediction. An adaptive quantization method is further designed for LTR frames to improve the overall video coding efficiency. Experimental results show that up to 22.86% coding gain can be achieved using the new coding scheme.

Learning-based parameter prediction for quality control in three-dimensional medical image compression

Quality control is of vital importance in compressing three-dimensional(3D)medical imaging data.Optimal com-pression parameters need to be determined based on the specific quality requirement.In high efficiency video coding(HEVC),regarded as the state-of-the-art compression tool,the quantization parameter(QP)plays a dominant role in controlling quality.The direct application of a video-based scheme in predicting the ideal parameters for 3D medical image compression cannot guarantee satisfactory results.In this paper we propose a learning-based parameter prediction scheme to achieve efficient quality control.Its kernel is a support vector regression(SVR)based learning model that is capable of predicting the optimal QP from both vid-eo-based and structural image features extracted directly from raw data,avoiding time-consuming processes such as pre-encoding and iteration,which are often needed in existing techniques.Experimental results on several datasets verify that our approach outperforms current video-based quality control methods.