基于APBT的类JPEG图像压缩以及HEVC视频编码与水印算法研究

发布时间：2019-05-07 07:58

【摘要】：在多媒体技术与网络通信技术快速发展的今天,人们可以通过互联网极其便利地获取高清的图片与视频信息。多媒体数据的传输对CPU和网络的压力也越来越大,同时网络的传播也带来了图像、视频等相关数字产品的版权保护与认证问题。针对以上问题,本文首先对图像、视频压缩相关标准、图形处理器(Graphics Processing Unit,GPU)并行以及视频水印相关技术的研究背景、意义以及国内外现状进行了阐述。之后研究了基于GPU并行的图像压缩编码以及目前最新的视频编码标准HEVC(High Efficiency Video Coding),并对HEVC中涉及到的相关技术进行了深入剖析,同时对其性能的改善做了相关探索。最后结合新一代视频压缩标准HEVC,对水印的预处理方案和视频水印的嵌入方案做了深入研究。本论文研究的主要内容与主要创新点如下:(1)深入分析了 JPEG图像压缩标准,针对JPEG图像编码速度较慢的问题,提出了一种基于GPU并行的JPEG图像编码方案,相较于传统的JPEG图像编码方案,可以提升近百倍的效率。(2)针对传统JPEG编码中分块离散余弦变换(Discrete Cosine Transform,DCT)在低码率时存在的块效应问题,借鉴了全相位双正交变换(All Phase Biorthogonal Transform,APBT)的构造过程,结合离散正弦变换(Discrete Sine Transform,DST)和全相位数字滤波(All Phase Digital Filtering,APDF)理论,提出了全相位离散正弦双正交变换(All Phase Discrete Sine Biorthogonal Transform,APDSBT),进一步完善了全相位双正交变换理论,并将其应用在JPEG编码中代替传统的DCT变换,实现了基于并行APDSBT的类JPEG编码。并行的APDSBT-JPEG图像编码算法,在解决了低码率下传统JPEG编码带来的块效应的同时提升了算法效率,相较于传统JPEG编码在主观和客观性能上带来了提升。(3)深入分析了 HEVC视频编码框架,提出了基于APBT变换的HEVC视频编码算法,使用APBT变换代替原有标准中的DCT变换。通过分析APBT变换对不同编码单元(Coding Unit,CU)下HEVC视频编码性能的影响,提出了一套最优的替换方案,并推导出了整数的APBT变换,研究了 APBT与DCT之间的联系,在不增加算法复杂度的前提下,提升了 HEVC的编码性能。(4)提出了一种基于HEVC压缩域的视频水印算法,该方案在量化后的系数中嵌入水印信息,可以很好地解决由水印嵌入带来的累积误差,并在数字水印预处理阶段采用APBT变换对水印数据进行简单压缩,优化了水印嵌入方案,提高了视频水印的嵌入容量。由于HEVC视频编码标准的自身特点,基于压缩域的HEVC视频水印无法抵抗重压缩攻击,因此本文还提出了一种基于APBT与奇异值分解(Singular Value Decomposition,SVD)的视频水印算法,该算法可以很好地抵抗HEVC压缩,表现出了极好的稳定性。
[Abstract]:With the rapid development of multimedia technology and network communication technology, people can obtain high-definition picture and video information very conveniently through the Internet. The transmission of multimedia data brings more and more pressure on CPU and network. At the same time, the transmission of multimedia data also brings about copyright protection and authentication of digital products such as image and video. In order to solve the above problems, the research background, significance and current situation of image and video compression standards, graphics processor (Graphics Processing Unit,GPU (graphics processor) parallelism and video watermarking technology are described in this paper. Then the GPU-based parallel image compression coding and the latest video coding standard HEVC (High Efficiency Video Coding), are studied and the related technologies in HEVC are analyzed deeply. At the same time the improvement of its performance is explored. Finally, based on the new generation of video compression standard HEVC, the watermarking preprocessing scheme and the video watermarking embedding scheme are deeply studied. The main contents and innovations of this thesis are as follows: (1) the JPEG image compression standard is analyzed in depth. Aiming at the slow speed of JPEG image coding, a parallel JPEG image coding scheme based on GPU is proposed. Compared with the traditional JPEG image coding scheme, it can improve the efficiency by nearly a hundred times. (2) aiming at the block effect problem of the block discrete cosine transform (Discrete Cosine Transform,DCT) in the traditional JPEG coding at low bit rate, The construction process of all-phase biorthogonal transform (All Phase Biorthogonal Transform,APBT) is used for reference, and the theory of discrete sine transform (Discrete Sine Transform,DST) and all-phase digital filtering (All Phase Digital Filtering,APDF) is combined. In this paper, all-phase discrete sinusoidal biorthogonal transform (All Phase Discrete Sine Biorthogonal Transform,APDSBT) is proposed, and the theory of all-phase biorthogonal transform is further perfected. The JPEG-like coding based on parallel APDSBT is realized by applying it to JPEG coding instead of traditional DCT transform. The parallel APDSBT-JPEG image coding algorithm not only solves the block effect caused by the traditional JPEG coding at low bit rate, but also improves the efficiency of the algorithm. Compared with the traditional JPEG coding, the subjective and objective performance of the HEVC video coding is improved. (3) the HEVC video coding framework is deeply analyzed, and the HEVC video coding algorithm based on APBT transform is proposed. The APBT transform is used to replace the DCT transform in the original standard. By analyzing the effect of APBT transform on the performance of HEVC video coding under different coding units (Coding Unit,CU), an optimal replacement scheme is proposed, and the integer APBT transform is derived, and the relationship between APBT and DCT is studied. Without increasing the complexity of the algorithm, the coding performance of HEVC is improved. (4) A video watermarking algorithm based on HEVC compressed domain is proposed, which embeds watermark information in the quantized coefficients. The accumulated error caused by watermark embedding can be solved well. In the pre-processing stage of digital watermarking, the watermark data is simply compressed by APBT transform, and the watermark embedding scheme is optimized, and the embedding capacity of video watermarking is improved. Because of the characteristics of HEVC video coding standard, HEVC video watermarking based on compressed domain can not resist the attack of re-compression. Therefore, a video watermarking algorithm based on APBT and singular value decomposition (Singular Value Decomposition,SVD) is proposed in this paper. This algorithm can resist HEVC compression very well and shows excellent stability.
【学位授予单位】：山东大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN919.81

【相似文献】