WO2019179096A1 - 一种基于压缩感知的质量可分级快速编码方法 - Google Patents

一种基于压缩感知的质量可分级快速编码方法 Download PDF

Info

Publication number
WO2019179096A1
WO2019179096A1 PCT/CN2018/111537 CN2018111537W WO2019179096A1 WO 2019179096 A1 WO2019179096 A1 WO 2019179096A1 CN 2018111537 W CN2018111537 W CN 2018111537W WO 2019179096 A1 WO2019179096 A1 WO 2019179096A1
Authority
WO
WIPO (PCT)
Prior art keywords
mode
coding
enhancement layer
skip
block
Prior art date
Application number
PCT/CN2018/111537
Other languages
English (en)
French (fr)
Inventor
胡栋
丁健宇
何永洋
Original Assignee
南京邮电大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 南京邮电大学 filed Critical 南京邮电大学
Publication of WO2019179096A1 publication Critical patent/WO2019179096A1/zh

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
    • H04N19/103—Selection of coding mode or of prediction mode
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
    • H04N19/124—Quantisation
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
    • H04N19/132—Sampling, masking or truncation of coding units, e.g. adaptive resampling, frame skipping, frame interpolation or high-frequency transform coefficient masking
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/134—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or criterion affecting or controlling the adaptive coding
    • H04N19/157—Assigned coding mode, i.e. the coding mode being predefined or preselected to be further used for selection of another element or parameter
    • H04N19/159—Prediction type, e.g. intra-frame, inter-frame or bidirectional frame prediction
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
    • H04N19/17—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object
    • H04N19/176—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object the region being a block, e.g. a macroblock
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
    • H04N19/187—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being a scalable video layer
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/30—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using hierarchical techniques, e.g. scalability
    • H04N19/36—Scalability techniques involving formatting the layers as a function of picture distortion after decoding, e.g. signal-to-noise [SNR] scalability
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/42—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals characterised by implementation details or hardware specially adapted for video compression or decompression, e.g. dedicated software implementation
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/46—Embedding additional information in the video signal during the compression process
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/70—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals characterised by syntax aspects related to video coding, e.g. related to compression standards
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/90—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using coding techniques not provided for in groups H04N19/10-H04N19/85, e.g. fractals
    • H04N19/91—Entropy coding, e.g. variable length coding [VLC] or arithmetic coding

Definitions

  • the invention relates to a quality scalable fast coding method based on compressed sensing, and belongs to the technical field of video coding.
  • Video image signal In many practical video compression transmission applications, due to the heterogeneity of the network, different user requirements, different terminal capabilities, and different QoS (Quality of Service) provided by the channel, it is necessary to provide users with different quality and different rates.
  • Video image signal One of the best ways to solve such problems is to use scalable video coding to allow a single encoder to generate multiple levels of compressed code streams. For different levels of code streams, different quality video image signals can be obtained. .
  • the problem of coding complexity and sharp increase in code rate due to fine quantization of one or more enhancement layers in scalable video coding has not been well solved.
  • the study of the theory of compressed sensing provides the possibility to solve this problem.
  • the advantage of the compressed sensing theory is that the amount of projection measurement data of the signal is much smaller than that obtained by the traditional sampling method, which breaks through the bottleneck of the Shannon sampling theorem, making the collection of high-resolution signals possible.
  • the present invention provides a quality scalable fast coding method based on compressed sensing. Firstly, the enhancement layer sub-block candidate mode is quickly selected by using the correlation between the base layer and the enhancement layer, and then According to the experiment, the appropriate sub-blocks are selected, quantized, and transmitted after sparse coding.
  • the present invention is directed to the H.264 and HEVC video coding standards.
  • the compressed sensing theory is added to the finely quantized enhancement layer, and the sparseness of the compressed sensing is selectively treated.
  • the coding sub-block performs coding transmission, which reduces the coding complexity of the original scalable video coding enhancement layer, thereby improving the coding efficiency of the overall coding framework.
  • DCT or DWT is often used as a sparse matrix and Gaussian random matrix or Bernoulli matrix is used as the measurement matrix.
  • an integer DCT transform has been used in the standard video coding reference software JSVM developed by JVT (Joint Video Team), and thus the present invention employs an integer DCT as a sparse matrix and a Gaussian random matrix as a measurement matrix.
  • a quality scalable fast coding method based on compressed sensing comprising: the following steps:
  • Step 1 Initialize the parameters:
  • Step 2 determining whether the current coded frame is an enhancement layer code, if not, indicating that the current coded frame is a base layer code, and encoding according to the original mode;
  • Step 3 Select a mode of the enhancement layer to be coded by using a fast mode; and obtain an optimal sub-block division mode of the current coding unit according to the inter-layer correlation and spatial correlation between the sub-blocks;
  • Step 4 determining whether the residual sub-block transform_size_8x8_flag flag of the enhancement layer is 1, if not step 5, otherwise proceed to step 6;
  • Step 5 Perform the original fine quantization and entropy coding process on the residual sub-block
  • Step 6 Fine-quantize the 8x8-sized residual sub-block, and then use the compressed sensing technology to perform sparse coding;
  • Step 7 determining, at the decoding end, whether the block to be decoded contains the flag bit F m , and if not, performing a normal decoding step;
  • Step 8 Calculate Y and ⁇ by using ⁇ and m obtained by transmission, and reconstruct the original signal according to the orthogonal matching pursuit algorithm.
  • Step 3.1 If the optimal coding mode of the base layer coding block is INTRA4x4, the enhancement layer corresponding position coding block is encoded in the INTRA_BL mode;
  • Step 3.2 If the optimal coding mode of the base layer coding block is INTRA16x16, the candidate mode of the enhancement layer coding block is one of INTRA_BL, MODE_16x16, MODE_SKIP, INTRA16x16, INTRA4x4, and then the optimal one is selected by the rate distortion optimization function.
  • the optimal coding mode as the corresponding position of the enhancement layer;
  • Step 3.3 When the optimal coding mode of the base layer is MODE_SKIP,
  • the optimal coding mode of the coded macroblock on the left, top and left of the corresponding coding position in the enhancement layer includes a combination of MODE_SKIP and MODE_16x16, the candidate modes of the coding position corresponding to the enhancement layer are MODE_SKIP, MODE_16x16, BL_SKIP one of them;
  • the optimal coding mode of the coded macroblock on the left, top and left of the corresponding coding position in the enhancement layer includes a combination of MODE_SKIP, MODE_16x16, MODE_16x8, and MODE_8x16
  • the candidate mode of the coding position corresponding to the enhancement layer is BL_SKIP, One of MODE_SKIP, MODE_16x16, MODE_16x8, MODE_8x16;
  • Step 3.4 When the optimal coding mode of the base layer coding block is MODE_16x16;
  • the optimal coding mode of the coded macroblock on the left, top and left of the corresponding coding position in the enhancement layer includes a combination of MODE_SKIP and MODE_16x16, the candidate mode of the coding position corresponding to the enhancement layer is MODE_SKIP, MODE_16x16, BL_SKIP one of them;
  • the optimal coding mode of the coded macroblock on the left, top and left of the corresponding coding position in the enhancement layer includes a combination of MODE_SKIP, MODE_16x16, MODE_16x8, and MODE_8x16
  • the candidate mode of the coding position corresponding to the enhancement layer is BL_SKIP.
  • the enhancement layer corresponding position coding candidate mode is one of BL_SKIP, MODE_SKIP, MODE_16x16, MODE_16x8, MODE_8x16, MODE_8x8;
  • Step 3.5 When the optimal coding mode of the base layer coding block is MODE_16x8 or MODE_8x16;
  • the optimal coding mode of the left, upper, and upper left encoded macroblocks in the enhancement layer includes the combination of MODE_SKIP, MODE_16x16, MODE_16x8, and MODE_8x16
  • the candidate mode of the enhancement layer corresponding to the coding position is BL_SKIP.
  • the enhancement layer corresponding position coding candidate mode is one of BL_SKIP, MODE_SKIP, MODE_16x16, MODE_16x8, MODE_8x16, MODE_8x8;
  • Step 3.6 If the optimal coding mode of the base layer coding block is MODE_8x8;
  • the optimal coding mode of the left, upper, and upper left encoded macroblocks in the enhancement layer is MODE_8x8 mode
  • the enhancement layer corresponding location candidate mode is one of BL_SKIP mode and MODE_8x8 mode
  • the enhancement layer corresponding position coding candidate mode is one of BL_SKIP, MODE_SKIP, MODE_16x16, and MODE_8x8.
  • the spatial correlation step 3 temper fast mode selection block, when not in effect in step 3.7, proceeds to step 3.8;
  • /Q b +k 2 Q e , where r 1 ,r 2 ,r 3 ,r 4 are the DCT coefficients of z 1 , z 2 , z 3 , z 4 respectively, Q b , Q e are the quantization step sizes of the base layer and the enhancement layer respectively; the calculation formula according to the DCT coefficient r ⁇
  • the present invention provides a quality-scalable fast coding method based on compressed sensing, and proposes a quality scalable fast coding method based on compressed sensing for H.264 and HEVC video coding standards, using sparse sensing theory sparseness.
  • the enhancement layer is selectively sparsely encoded after obtaining the sub-block to be coded, which effectively reduces the code rate of the scalable video coding.
  • the present invention effectively improves the coding efficiency in the case where the signal ratio attenuation is negligible.
  • 1 is a structural block diagram of a scalable video coding encoder
  • FIG. 2 is a schematic diagram of quality scalable coding based on compressed sensing
  • Figure 3 is a flow chart of a mode selection fast algorithm
  • Figure 4 is a schematic diagram of prediction of an enhancement layer sub-block mode
  • Figure 5 is a schematic diagram of the compressed sensing process.
  • Scalable video coding is divided into time scalable, spatial scalable, and quality scalable.
  • this block diagram is a scalable coding scheme of a base layer and an enhancement layer.
  • the enhancement layer may have multiple layers.
  • the inter-layer prediction technique between the base layer and the enhancement layer is removed, and the base layer and the enhancement layer are two independent video codec processes. Since the video sequences of the scalable video coding base layer and the enhancement layer are the same video, only the resolution, the frame rate, or the quality are different or the same, the video resolutions of the base layer and the enhancement layer are the same for the quality scalability.
  • Different quantization step sizes are used in the base layer and the enhancement layer to enable the base layer and the enhancement layer to obtain different quality videos to adapt to different networks and devices.
  • the quantization step size of the base layer is greater than the quantization step size of the enhancement layer. This leads to a sharp increase in the code rate due to fine quantization of the enhancement layer.
  • some studies have dealt with this problem by using compressed sensing, they do not solve the coding complexity caused by compressed sensing reconstruction. Too big a problem.
  • the invention provides a quality scalable fast coding method based on compressed sensing, which utilizes the advantages of compression sensing theory to complete compression when sampling a signal, and the method of the invention selectively and sparsely represents data of an enhancement layer in scalable video coding. Reducing the amount of data of the enhancement layer residual signal does not excessively enhance the computational complexity of the compressed sensing.
  • Step 1 Initialize the parameters:
  • Step 2 Determine whether the current coded frame is an enhancement layer code. If not, it indicates that the current coded frame is a base layer code, and it is coded according to the original mode.
  • Step 3 Select the mode of the enhancement layer to be coded sub-block by using the fast mode. According to the inter-layer correlation and spatial correlation between sub-blocks, the optimal sub-block division mode of the current coding unit is quickly obtained, as shown in FIG. 3, and the specific steps are as follows:
  • Step 3.1 If the optimal coding mode of the base layer coding block is INTRA4x4, the enhancement layer corresponding position coding block is coded in the INTRA_BL mode.
  • Step 3.2 If the optimal coding mode of the base layer coding block is INTRA16x16, the candidate mode of the enhancement layer coding block is one of INTRA_BL, MODE_16x16, MODE_SKIP, INTRA16x16, INTRA4x4, and then the optimal one is selected by the rate distortion optimization function.
  • the optimal coding mode as the corresponding position of the enhancement layer.
  • Step 3.3 When the optimal coding mode of the base layer is MODE_SKIP, as shown in FIG. 4,
  • the optimal coding mode of the coded macroblock on the left, top and left of the corresponding coding position in the enhancement layer includes a combination of MODE_SKIP and MODE_16x16
  • the candidate modes of the coding position corresponding to the enhancement layer are MODE_SKIP, MODE_16x16, BL_SKIP one of them.
  • the candidate mode of the coding position corresponding to the enhancement layer is BL_SKIP, One of MODE_SKIP, MODE_16x16, MODE_16x8, and MODE_8x16.
  • Step 3.4 When the optimal coding mode of the base layer coding block is MODE_16x16.
  • the optimal coding mode of the coded macroblock on the left, top and left of the corresponding coding position in the enhancement layer includes a combination of MODE_SKIP and MODE_16x16
  • the candidate mode of the coding position corresponding to the enhancement layer is MODE_SKIP, MODE_16x16, BL_SKIP one of them.
  • the optimal coding mode of the coded macroblock on the left, top and left of the corresponding coding position in the enhancement layer includes a combination of MODE_SKIP, MODE_16x16, MODE_16x8, and MODE_8x16
  • the candidate mode of the coding position corresponding to the enhancement layer is BL_SKIP.
  • the enhancement layer corresponding position coding candidate mode is one of BL_SKIP, MODE_SKIP, MODE_16x16, MODE_16x8, MODE_8x16, MODE_8x8.
  • Step 3.5 When the optimal coding mode of the base layer coding block is MODE_16x8 or MODE_8x16.
  • the optimal coding mode of the left, upper, and upper left encoded macroblocks in the enhancement layer includes the combination of MODE_SKIP, MODE_16x16, MODE_16x8, and MODE_8x16
  • the candidate mode of the enhancement layer corresponding to the coding position is BL_SKIP.
  • the enhancement layer corresponding position coding candidate mode is one of BL_SKIP, MODE_SKIP, MODE_16x16, MODE_16x8, MODE_8x16, MODE_8x8.
  • Step 3.6 If the optimal coding mode of the base layer coding block is MODE_8x8.
  • the enhancement layer corresponding location candidate mode is one of BL_SKIP mode and MODE_8x8 mode.
  • the enhancement layer corresponding position coding candidate mode is one of BL_SKIP, MODE_SKIP, MODE_16x16, and MODE_8x8.
  • Step 3.7 For the candidate modes involved in steps 3.3 to 3.6, the mode decision is advanced early using the inter-layer correlation degree. And z b z e assume respectively the base layer and the enhancement layer quantized coefficients, the degree of association by the conditions interlayer advancing mode is selected: z e -z b ⁇ k 1, k 1 is considered the threshold value obtained by experiment The best is 2.43.
  • the rate distortion function value and the quantization step size satisfy the condition. Then, the mode selection of the enhancement layer coding block ends, where RD is the rate distortion cost, and RD e and RD b represent the rate distortion cost of the enhancement layer and the base layer, respectively.
  • step 3.7 does not take effect, proceed to step 3.8.
  • Step 3.8 Using spatial correlation to end the mode selection condition:
  • the rate distortion function value and the quantization step size of the base layer and the enhancement layer coding block satisfy the condition Then, the mode selection of the enhancement layer to be coded block ends, where RD is the rate distortion cost, RD 1 and RD 2 are the rate distortion cost of the base layer neighboring block, and RD 3 and RD 4 are the rate distortion of the enhancement layer neighboring block. cost.
  • Step 4 Determine whether the residual sub-block transform_size_8x8_flag flag of the enhancement layer is 1, if not step 5, otherwise proceed to step 6.
  • Step 5 Perform the original fine quantization and entropy coding process on the residual sub-block.
  • Step 6 Fine-quantize the 8x8-sized residual sub-blocks, and then use the compressed sensing technology to perform sparse coding.
  • the specific steps are as follows:
  • Step 6.1 As shown in FIG. 5, firstly, the residual matrix of 8 ⁇ 8 size is changed into a one-dimensional sparse signal ⁇ of length N, and the residual sub-block is sparsely represented by integer DCT transform (sparse basis ,) to obtain a sparse signal.
  • ⁇ the residual matrix of 8 ⁇ 8 size
  • ⁇ the residual sub-block
  • X the residual sub-block
  • Step 6.4 Set the flag F m and entropy encode the data after the measurement value is followed by (64-m) zeros.
  • Step 7 Determine at the decoding end whether the block to be decoded contains the flag bit F m , and if not, perform a normal decoding step.
  • Step 8 Calculate Y and ⁇ by using ⁇ and m obtained by transmission, and reconstruct the original signal according to Orthogonal Matching Pursuit (OMP).
  • OMP Orthogonal Matching Pursuit
  • Step 8.3 Find the element with the largest absolute value in g (n) , ie
  • Step 8.5 Find the approximate solution by least squares method
  • the method of the present invention can reduce the coding rate to some extent while maintaining the code quality attenuation with negligible advancement compared to the reference software.

Landscapes

  • Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Compression Or Coding Systems Of Tv Signals (AREA)
  • Compression, Expansion, Code Conversion, And Decoders (AREA)

Abstract

本发明公开了一种基于压缩感知的质量可分级快速编码方法，属于视频编码技术领域。本发明方法利用压缩感知理论的稀疏性，在对质量可分级增强层进行编码时对残差块尺寸为8x8的子块进行稀疏表示，编码时为了满足标准编码结构提出补0操作再进行熵编码。本发明还利用了基本层和增强层之间的层间相关性来快速选择子块编码模式以进一步降低编码算法的计算复杂度。相比现有技术，本发明方法能够在保持编码后图像质量的前提下，有效地降低编码端的码率，提高编码器的编码效率。

Description

一种基于压缩感知的质量可分级快速编码方法 技术领域
本发明涉及一种基于压缩感知的质量可分级快速编码方法，属于视频编码技术领域。
背景技术
在许多实际的视频压缩传输应用中，由于网络的异构、用户要求不同、终端能力不同、信道所能提供的Qos(Quality of Service)不同等因素的存在，需要为用户提供不同质量、不同速率的视频图像信号。而解决此类问题的最好方法之一便是采用可分级视频编码，让单个编码器产生多个层次的压缩码流，对于不同层次的码流进行解码，便可以获得不同质量的视频图像信号。然而，可分级视频编码中由于一个甚至多个增强层的精细量化而导致的编码复杂度以及码率急剧增加的问题始终未能很好地解决。压缩感知理论的研究为这一问题的解决提供了可能。压缩感知理论的优点在于信号的投影测量数据量远远小于传统采样方法所获的数据量，突破了香农采样定理的瓶颈，使得高分辨率信号的采集成为了可能。
近年来，已有一些利用压缩感知来改进可分级视频编码的方法。例如，Siyuan Xiang和Lin Cai于2011年提出一种应用于无线网络环境下的基于压缩感知的可分级视频编码框架，该编码框架在主要通过不使用运动估计、运动补偿以及帧间预测时只将部分I帧的变换系数进行重构后作为参考帧来降低编码的计算复杂度，但其编码方法总体效果并不理想。此外，S.N. Karishma等人也在2016年提出一种适用于空间应用的压缩感知可分级编码框架。以上两种方法虽然利用了压缩感知达到了降低编码复杂度的目的，但是并没有很好地解决由于多层编码以及压缩感知重构所带来的时延问题，因此总体编码时间有再次被减少的空间。
发明内容
目的：为了克服现有技术中存在的不足，本发明提供一种基于压缩感知的质量可分级快速编码方法，首先利用基本层和增强层之间的相关性快速选择增强层子块候选模式，之后根据实验选择适合的子块经过量化，稀疏编码后传输。
本发明面向H.264和HEVC视频编码标准，在质量可分级视频编码的基本层保持不变的情况下，对精细量化后的增强层加入压缩感知理论，结合压缩感知的稀疏性有选择地对待编码子块进行编码传输，降低原有可分级视频编码增强层编码复杂度，从而提高整体编码框架的编码效率。
由于压缩感知中所用到的稀疏矩阵和测量矩阵需要满足有限等距性(Restricted Isometry Property,RIP)原则，实际实验中常用DCT或DWT作为稀疏矩阵而采用高斯随机矩阵或贝努利矩阵作为测量矩阵，而在JVT(Joint Video Team)开发的标准视频编码参考软件JSVM中已使用到整数DCT变换，因而本发明采用整数DCT作为稀疏矩阵，高斯随机矩阵为测量矩阵。且通过实验总结分析，当增强层残差块的大小为8x8时，对其进行压缩感知获得的实验效果明显优于对其他尺寸残差块进行稀疏处理的结果，故本发明只对增强层8x8尺寸大小的子块进行稀疏表示。
技术方案：为解决上述技术问题，本发明采用的技术方案为：
一种基于压缩感知的质量可分级快速编码方法，其特征在于：包括如下步骤：
步骤1：初始化参数：
1.1：利用高斯随机函数生成大小为64x64高斯随机矩阵Φ；
1.2：设置可分级视频编码层数为2；
步骤2：判断当前编码帧是否是增强层编码，若不是，表示当前编码帧是基本层编码，对其按照原先方式进行编码；
步骤3：用快速模式选择得到增强层待编码子块的模式；根据子块之间的层间相关性和空间相关性，快速得到当前编码单元的最佳子块划分模式；
步骤4：判断增强层的残差子块transform_size_8x8_flag标志位是否为1，若不是进行步骤5，否则进行步骤6；
步骤5：对残差子块进行原有的细量化和熵编码过程；
步骤6：对8x8大小的残差子块进行细量化，之后利用压缩感知技术对其进行稀疏编码；
步骤7：在解码端判断待解码块是否含有标志位F m，若没有，进行正常的解码步骤；
步骤8：利用传输得到的Φ以及m计算出Y以及φ，再根据正交匹配追踪算法重构得到原信号。
作为优选方案：步骤3.1：若基本层编码块的最优编码模式为INTRA4x4，则增强层对应位置编码块采用INTRA_BL模式进行编码；
步骤3.2：若基本层编码块的最优编码模式为INTRA16x16，则增强层编码块的候选模式为INTRA_BL、MODE_16x16、MODE_SKIP、INTRA16x16、INTRA4x4其中一种，之后通过率失真优化函数选择其中最优的一种作为增强层对应位置的最优编码模式；
步骤3.3：当基本层的最优编码模式为MODE_SKIP时，
3.3.1：若增强层中对应编码位置的左面、上面、左上面已编码的宏块的最优编码模式均为MODE_SKIP，则此增强层对应编码位置采用MODE_SKIP模式进行编码；
3.3.2：若增强层中对应编码位置的左面、上面、左上面已编码的宏块的最优编码模式包含MODE_SKIP和MODE_16x16的组合，则增强层对应编码位置的候选模式为MODE_SKIP、MODE_16x16、BL_SKIP其中一种；
3.3.3若增强层中对应编码位置的左面、上面、左上面已编码的宏块的最优编码模式包含MODE_SKIP、MODE_16x16、MODE_16x8、MODE_8x16的组合，则增强层对应编码位置的候选模式为BL_SKIP、MODE_SKIP、MODE_16x16、MODE_16x8、MODE_8x16其中一种；
步骤3.4：当基本层编码块的最优编码模式为MODE_16x16时；
3.4.1：若增强层中对应编码位置的左面、上面、左上面已编码的宏块的最优编码模包含MODE_SKIP和MODE_16x16的组合，则增强层对应编码位置的候选模式为MODE_SKIP、MODE_16x16、BL_SKIP其中一种；
3.4.2：若增强层中对应编码位置的左面、上面、左上面已编码的宏块的最优编码模式包含MODE_SKIP、MODE_16x16、MODE_16x8、MODE_8x16的 组合，则增强层对应编码位置的候选模式为BL_SKIP、MODE_SKIP、MODE_16x16、MODE_16x8、MODE_8x16其中一种；
3.4.3：否则，增强层对应位置编码候选模式为BL_SKIP、MODE_SKIP、MODE_16x16、MODE_16x8、MODE_8x16、MODE_8x8其中一种；
步骤3.5：当基本层编码块的最优编码模式为MODE_16x8或MODE_8x16时；
3.5.1：若增强层中对应编码位置的左面、上面、左上面已编码的宏块的最优编码模式包含MODE_SKIP、MODE_16x16、MODE_16x8、MODE_8x16的组合，则增强层对应编码位置的候选模式为BL_SKIP、MODE_SKIP、MODE_16x16、MODE_16x8、MODE_8x16其中一种；
3.5.2：否则，增强层对应位置编码候选模式为BL_SKIP、MODE_SKIP、MODE_16x16、MODE_16x8、MODE_8x16、MODE_8x8其中一种；
步骤3.6：若基本层编码块的最优编码模式为MODE_8x8时；
3.6.1：若增强层中对应编码位置的左面、上面、左上面已编码的宏块的最优编码模式均为MODE_8x8模式，则增强层对应位置候选模式为BL_SKIP模式、MODE_8x8模式其中一种；
3.6.2：否则，增强层对应位置编码候选模式为BL_SKIP、MODE_SKIP、MODE_16x16、MODE_8x8其中一种。
作为优选方案：步骤3.7：对于步骤3.3至3.6中涉及的候选模式，利用层间关联度提前结束模式决策；假设z e和z b分别为基本层和增强层量化后的系数，则通过层间关联度提前模式选择的条件为：z e-z b≤k 1，k 1为通过实 验综合考虑所得的阈值；可重写为r e≤Q er b/Q b+k 1Q e，其中Q b，Q e分别为基本层和增强层的量化步长；r b，r e分别为基本层和增强层的DCT系数，DCT系数的计算公式为r＝∑∑d iux uvd jv，其中d iu为整数DCT变换中(i,u)位置所对应的值，x uv为残差信号值，由于d iu的取值小于
Figure PCTCN2018111537-appb-000001
所以
Figure PCTCN2018111537-appb-000002
因而可得
Figure PCTCN2018111537-appb-000003
其中SAD为绝对残差和，SAD e和SAD b分别代表增强层和基本层的绝对残差和；于是当基本层和增强层对应编码块的率失真函数值和量化步长满足条件
Figure PCTCN2018111537-appb-000004
时，则增强层编码块的模式选择结束，其中RD为率失真代价，RD e和RD b分别代表增强层和基本层的率失真代价。
作为优选方案：所述步骤3中空间相关性子块快速模式选择，当步骤3.7没有生效时，进入步骤3.8；所述步骤3.8：利用空间相关性提前结束模式选择的条件为：|z 1-z 2|-|z 3-z 4|≤k 2，其中z 1，z 2为增强层两相邻子块的量化系数，z 3，z 4为基本层两相邻子块的量化系数，k 2为通过实验所得阈值；该条件可重写为|r 1-r 2|≤Q e|r 3-r 4|/Q b+k 2Q e，其中r 1，r 2，r 3，r 4分别为z 1，z 2，z 3，z 4的DCT系数，Q b，Q e分别为基本层和增强层的量化步长；根据DCT系数的计算公式r＝∑∑d iux uvd jv，可以得到
Figure PCTCN2018111537-appb-000005
其中SAD为绝对残差和，SAD 1，SAD 2为基本层相邻块绝对残差和，SAD 3和SAD 4增强层相邻块绝对残差和；因此，当基本层和增强层编码块的率失真函数值和量化步长满足条件
Figure PCTCN2018111537-appb-000006
时，则增强层待编码块的模式选择结束，其中RD为率失真代价，RD 1和RD 2为基本层相邻块的率失真代价，RD 3和RD 4 为增强层相邻块的率失真代价。
有益效果：本发明提供的一种基于压缩感知的质量可分级快速编码方法，面向H.264和HEVC视频编码标准，提出一种基于压缩感知的质量可分级快速编码方法，利用压缩感知理论的稀疏性，在快速得到增强层待编码子块之后对其有选择地进行稀疏编码，有效地降低了可分级视频编码的码率。相比于原有的质量可分级编码，本发明在信号比衰减可忽略的情况下，有效地提高了编码效率。
附图说明
图1为可分级视频编码编码器的结构框图；
图2基于压缩感知的质量可分级编码示意图；
图3模式选择快速算法流程图；
图4增强层子块模式预测示意图；
图5压缩感知过程示意图。
具体实施方式
下面结合附图对本发明作更进一步的说明。
可分级视频编码分为时间可分级、空间可分级和质量可分级。如图1所示，此框图为一个基本层和一个增强层的可分级编码情形，在此需要指出的是增强层可以有多层。从图1可知，除去基本层和增强层之间的层间预测技术，基本层和增强层是两个独立的视频编解码过程。由于可分级视频编码基层和增强层的视频序列是同一个视频，只是分辨率、帧率、或者质量不同或者相同而已，对于质量可分级而言，基本层和增强层的视频分辨率是相同的，在基本层和增强层中使用了不同的量化步长来使基本层和 增强层得到不同质量的视频以适应不同网络及设备，通常基本层的量化步长要大于增强层的量化步长，这就会导致增强层由于精细量化带来码率的急剧增加，尽管当前已有一些研究通过利用压缩感知来处理该问题，但是它们并没有很好地解决压缩感知重构带来的编码复杂度过大的问题。
本发明提出一种基于压缩感知的质量可分级快速编码方法，利用压缩感知理论对信号采样时完成压缩的优点，本发明方法对可分级视频编码中的增强层的数据进行有选择地稀疏表示，减少增强层残差信号的数据量的同时也不会过多的增强压缩感知的计算复杂度。
如图2所示，一种基于压缩感知的质量可分级快速编码方法，从此流程图上可以看出本发明的改进算法对基本层和增强层有不同的算法。
步骤1：初始化参数：
1.1：利用高斯随机函数生成大小为64x64高斯随机矩阵Φ；
1.2：设置可分级视频编码层数为2。
步骤2：判断当前编码帧是否是增强层编码，若不是，表示当前编码帧是基本层编码，对其按照原先方式进行编码。
步骤3：用快速模式选择得到增强层待编码子块的模式。根据子块之间的层间相关性和空间相关性，快速得到当前编码单元的最佳子块划分模式，如图3所示，具体步骤如下：
步骤3.1：若基本层编码块的最优编码模式为INTRA4x4，则增强层对应位置编码块采用INTRA_BL模式进行编码。
步骤3.2：若基本层编码块的最优编码模式为INTRA16x16，则增强层 编码块的候选模式为INTRA_BL、MODE_16x16、MODE_SKIP、INTRA16x16、INTRA4x4其中一种，之后通过率失真优化函数选择其中最优的一种作为增强层对应位置的最优编码模式。
步骤3.3：当基本层的最优编码模式为MODE_SKIP时，如图4所示，
3.3.1：若增强层中对应编码位置的左面、上面、左上面已编码的宏块的最优编码模式均为MODE_SKIP，则此增强层对应编码位置采用MODE_SKIP模式进行编码。
3.3.2：若增强层中对应编码位置的左面、上面、左上面已编码的宏块的最优编码模式包含MODE_SKIP和MODE_16x16的组合，则增强层对应编码位置的候选模式为MODE_SKIP、MODE_16x16、BL_SKIP其中一种。
3.3.3若增强层中对应编码位置的左面、上面、左上面已编码的宏块的最优编码模式包含MODE_SKIP、MODE_16x16、MODE_16x8、MODE_8x16的组合，则增强层对应编码位置的候选模式为BL_SKIP、MODE_SKIP、MODE_16x16、MODE_16x8、MODE_8x16其中一种。
步骤3.4：当基本层编码块的最优编码模式为MODE_16x16时。
3.4.1：若增强层中对应编码位置的左面、上面、左上面已编码的宏块的最优编码模包含MODE_SKIP和MODE_16x16的组合，则增强层对应编码位置的候选模式为MODE_SKIP、MODE_16x16、BL_SKIP其中一种。
3.4.2：若增强层中对应编码位置的左面、上面、左上面已编码的宏块的最优编码模式包含MODE_SKIP、MODE_16x16、MODE_16x8、MODE_8x16的组合，则增强层对应编码位置的候选模式为BL_SKIP、MODE_SKIP、 MODE_16x16、MODE_16x8、MODE_8x16其中一种。
3.4.3：否则，增强层对应位置编码候选模式为BL_SKIP、MODE_SKIP、MODE_16x16、MODE_16x8、MODE_8x16、MODE_8x8其中一种。
步骤3.5：当基本层编码块的最优编码模式为MODE_16x8或MODE_8x16时。
3.5.1：若增强层中对应编码位置的左面、上面、左上面已编码的宏块的最优编码模式包含MODE_SKIP、MODE_16x16、MODE_16x8、MODE_8x16的组合，则增强层对应编码位置的候选模式为BL_SKIP、MODE_SKIP、MODE_16x16、MODE_16x8、MODE_8x16其中一种。
3.5.2：否则，增强层对应位置编码候选模式为BL_SKIP、MODE_SKIP、MODE_16x16、MODE_16x8、MODE_8x16、MODE_8x8其中一种。
步骤3.6：若基本层编码块的最优编码模式为MODE_8x8时。
3.6.1：若增强层中对应编码位置的左面、上面、左上面已编码的宏块的最优编码模式均为MODE_8x8模式，则增强层对应位置候选模式为BL_SKIP模式、MODE_8x8模式其中一种。
3.6.2：否则，增强层对应位置编码候选模式为BL_SKIP、MODE_SKIP、MODE_16x16、MODE_8x8其中一种。
步骤3.7：对于步骤3.3至3.6中涉及的候选模式，利用层间关联度提前结束模式决策。假设z e和z b分别为基本层和增强层量化后的系数，则通过层间关联度提前模式选择的条件为：z e-z b≤k 1，k 1为通过实验综合考虑所得的阈值，最佳为2.43。可重写为r e≤Q er b/Q b+k 1Q e，其中Q b，Q e分别为基本层 和增强层的量化步长；r b，r e分别为基本层和增强层的DCT系数，DCT系数的计算公式为r＝∑∑d iux uvd jv，其中d iu为整数DCT变换中(i,u)位置所对应的值，x uv为残差信号值，由于d iu的取值小于
Figure PCTCN2018111537-appb-000007
所以
Figure PCTCN2018111537-appb-000008
因而可得
Figure PCTCN2018111537-appb-000009
其中SAD为绝对残差和，SAD e和SAD b分别代表增强层和基本层的绝对残差和。于是当基本层和增强层对应编码块的率失真函数值和量化步长满足条件
Figure PCTCN2018111537-appb-000010
时，则增强层编码块的模式选择结束，其中RD为率失真代价，RD e和RD b分别代表增强层和基本层的率失真代价。当步骤3.7没有生效时，进入步骤3.8。
步骤3.8：利用空间相关性提前结束模式选择的条件为：|z 1-z 2|-|z 3-z 4|≤k 2，其中z 1，z 2为增强层两相邻子块的量化系数，z 3，z 4为基本层两相邻子块的量化系数，k 2为通过实验所得阈值，最佳设定为4.31。该条件可重写为|r 1-r 2|≤Q e|r 3-r 4|/Q b+k 2Q e，其中r 1，r 2，r 3，r 4分别为z 1，z 2，z 3，z 4的DCT系数，Q b，Q e分别为基本层和增强层的量化步长。根据DCT系数的计算公式r＝∑∑d iux uvd jv，可以得到
Figure PCTCN2018111537-appb-000011
其中SAD为绝对残差和，SAD 1，SAD 2为基本层相邻块绝对残差和，SAD 3和SAD 4增强层相邻块绝对残差和。因此，当基本层和增强层编码块的率失真函数值和量化步长满足条件
Figure PCTCN2018111537-appb-000012
时，则增强层待编码块的模式选择结束，其中RD为率失真代价，RD 1和RD 2为基本层相邻块的率失真代价，RD 3和RD 4为增强层相邻块的率失真代价。
步骤4：判断增强层的残差子块transform_size_8x8_flag标志位是否 为1，若不是进行步骤5，否则进行步骤6。
步骤5：对残差子块进行原有的细量化和熵编码过程。
步骤6：对8x8大小的残差子块进行细量化，之后利用压缩感知技术对其进行稀疏编码。具体步骤如下所示：
步骤6.1：如图5所示，首先将8x8大小的残差矩阵变为长度为N的一维稀疏信号Θ，利用整数DCT变换(稀疏基ψ)对残差子块进行稀疏表示，得到稀疏信号X。
步骤6.2：选用一个与稀疏基ψ满足RIP原则、大小为mx64高斯随机测量矩阵φ，其中m的计算公式为：m＝klog 2(N/k)，其中k为稀疏信号中的稀疏度，即不为0的个数。
步骤6.3：将稀疏信号X投影到测量矩阵φ上，得到信号Y，计算公式为Y＝φ·X。
步骤6.4：设立标志位F m并对测量值后面补上(64-m)个0后的数据进行熵编码。
步骤7：在解码端判断待解码块是否含有标志位F m，若没有，进行正常的解码步骤。
步骤8：利用传输得到的Φ以及m计算出Y以及φ，再根据正交匹配追踪算法(Orthogonal Matching Pursuit,OMP)重构得到原信号，具体重构步骤如下所示：
步骤8.1：初始化参数设置：残差r (0)＝y，重建信号x (0)＝0，信号的索引集为Γ (0)＝φ，迭代次数为n＝0，停止迭代判决误差ε>0。
步骤8.2：计算残差和观测矩阵的每行内积g (n)＝φ·r (n-1)。
步骤8.3：找出g (n)中绝对值最大的元素，即
Figure PCTCN2018111537-appb-000013
步骤8.4：更新索引集Γ (n)＝Γ (n-1)U{k}，及原子集合
Figure PCTCN2018111537-appb-000014
步骤8.5：利用最小二乘法求得近似解
Figure PCTCN2018111537-appb-000015
步骤8.6：更新残差r (n)＝y-x (n)。
步骤8.7：判断是否满足迭代停止条件，若满足则停止，令x＝x (n)，输出x，否则n＝n+1，返回步骤8.1。
为验证本发明方法相比于原标准中方法所取得的有益效果，进行以下验证实验：选取三段不同的视频序列，利用本发明的方法进行编码，三段视频序列(PartyScene，FlowerVase，ParkRunner)均为分辨率为1280x720，帧率为30。本发明编码方法在H.264的可分级参考软件JSVM9.18上实现并与参考软件进行对比实验一，与另一篇在IEEE International Conference on Communications发表的结合压缩感知的可分级视频编码框架(Scalable Video Coding with Compressive Sensing for Wireless Videocast)进行对比试验二。
所得得到实验数据如下图的表1，表2所示：
表1
Figure PCTCN2018111537-appb-000016
Figure PCTCN2018111537-appb-000017
表2
Figure PCTCN2018111537-appb-000018
从表1的实验数据可以看出相比参考软件本发明的方法能够在维持编码质量衰减在可忽略的提前下一定程度地降低编码码率。
从表2的实验数据可以看出相比与比较算法，本文的算法在视频图像质量衰减效果可忽略的情况下，大幅度的降低了编码时间，提高算法的编码效率。
以上所述仅是本发明的优选实施方式，应当指出：对于本技术领域的普通技术人员来说，在不脱离本发明原理的前提下，还可以做出若干改进和润饰，这些改进和润饰也应视为本发明的保护范围。

Claims (8)

  1. 一种基于压缩感知的质量可分级快速编码方法，其特征在于：包括如下步骤：
    步骤1：初始化参数：
    1.1：利用高斯随机函数生成大小为64x64高斯随机矩阵Φ；
    1.2：设置可分级视频编码层数为2；
    步骤2：判断当前编码帧是否是增强层编码，若不是，表示当前编码帧是基本层编码，对其按照原先方式进行编码；
    步骤3：用快速模式选择得到增强层待编码子块的模式；根据子块之间的层间相关性和空间相关性，快速得到当前编码单元的最佳子块划分模式；
    步骤4：判断增强层的残差子块transform_size_8x8_flag标志位是否为1，若不是进行步骤5，否则进行步骤6；
    步骤5：对残差子块进行原有的细量化和熵编码过程；
    步骤6：对8x8大小的残差子块进行细量化，之后利用压缩感知技术对其进行稀疏编码；
    步骤7：在解码端判断待解码块是否含有标志位F m，若没有，进行正常的解码步骤；
    步骤8：利用传输得到的Φ以及m计算出Y以及φ，再根据正交匹配追踪算法重构得到原信号。
  2. 根据权利要求1所述的一种基于压缩感知的质量可分级快速编码方法，其特征在于：所述步骤3中快速模式选择得到增强层待编码子块的模式，具体步骤如下：
    步骤3.1：若基本层编码块的最优编码模式为INTRA4x4，则增强层对应位置编码块采用INTRA_BL模式进行编码；
    步骤3.2：若基本层编码块的最优编码模式为INTRA16x16，则增强层编码块的候选模式为INTRA_BL、MODE_16x16、MODE_SKIP、INTRA16x16、INTRA4x4其中一种，之后通过率失真优化函数选择其中最优的一种作为增强层对应位置的最优编码模式；
    步骤3.3：当基本层的最优编码模式为MODE_SKIP时，
    3.3.1：若增强层中对应编码位置的左面、上面、左上面已编码的宏块的最优编码模式均为MODE_SKIP，则此增强层对应编码位置采用MODE_SKIP模式进行编码；
    3.3.2：若增强层中对应编码位置的左面、上面、左上面已编码的宏块的最优编码模式包含MODE_SKIP和MODE_16x16的组合，则增强层对应编码位置的候选模式为MODE_SKIP、MODE_16x16、BL_SKIP其中一种；
    3.3.3若增强层中对应编码位置的左面、上面、左上面已编码的宏块的最优编码模式包含MODE_SKIP、MODE_16x16、MODE_16x8、MODE_8x16的组合，则增强层对应编码位置的候选模式为BL_SKIP、MODE_SKIP、MODE_16x16、MODE_16x8、MODE_8x16其中一种；
    步骤3.4：当基本层编码块的最优编码模式为MODE_16x16时；
    3.4.1：若增强层中对应编码位置的左面、上面、左上面已编码的宏块的最优编码模包含MODE_SKIP和MODE_16x16的组合，则增强层对应编码位置的候选模式为MODE_SKIP、MODE_16x16、BL_SKIP其中一种；
    3.4.2：若增强层中对应编码位置的左面、上面、左上面已编码的宏块的最优编码模式包含MODE_SKIP、MODE_16x16、MODE_16x8、MODE_8x16的组合，则增强层对应编码位置的候选模式为BL_SKIP、MODE_SKIP、MODE_16x16、MODE_16x8、MODE_8x16其中一种；
    3.4.3：否则，增强层对应位置编码候选模式为BL_SKIP、MODE_SKIP、MODE_16x16、MODE_16x8、MODE_8x16、MODE_8x8其中一种；
    步骤3.5：当基本层编码块的最优编码模式为MODE_16x8或MODE_8x16时；
    3.5.1：若增强层中对应编码位置的左面、上面、左上面已编码的宏块的最优编码模式包含MODE_SKIP、MODE_16x16、MODE_16x8、MODE_8x16的组合，则增强层对应编码位置的候选模式为BL_SKIP、MODE_SKIP、MODE_16x16、MODE_16x8、MODE_8x16其中一种；
    3.5.2：否则，增强层对应位置编码候选模式为BL_SKIP、MODE_SKIP、MODE_16x16、MODE_16x8、MODE_8x16、MODE_8x8其中一种；
    步骤3.6：若基本层编码块的最优编码模式为MODE_8x8时；
    3.6.1：若增强层中对应编码位置的左面、上面、左上面已编码的宏块的最优编码模式均为MODE_8x8模式，则增强层对应位置候选模式为BL_SKIP模式、MODE_8x8模式其中一种；
    3.6.2：否则，增强层对应位置编码候选模式为BL_SKIP、MODE_SKIP、MODE_16x16、MODE_8x8其中一种。
  3. 根据权利要求2所述的一种基于压缩感知的质量可分级快速编码方法， 其特征在于：所述步骤3中层间相关性子块快速模式选择，步骤如下：
    步骤3.7：对于步骤3.3至3.6中涉及的候选模式，利用层间关联度提前结束模式决策；假设z e和z b分别为基本层和增强层量化后的系数，则通过层间关联度提前模式选择的条件为：z e-z b≤k 1，k 1为通过实验综合考虑所得的阈值；可重写为r e≤Q er b/Q b+k 1Q e，其中Q b，Q e分别为基本层和增强层的量化步长；r b，r e分别为基本层和增强层的DCT系数，DCT系数的计算公式为r＝∑∑d iux uvd jv，其中d iu为整数DCT变换中(i,u)位置所对应的值，x uv为残差信号值，由于diu的取值小于
    Figure PCTCN2018111537-appb-100001
    所以
    Figure PCTCN2018111537-appb-100002
    因而可得
    Figure PCTCN2018111537-appb-100003
    其中SAD为绝对残差和，SAD e和SAD b分别代表增强层和基本层的绝对残差和；于是当基本层和增强层对应编码块的率失真函数值和量化步长满足条件
    Figure PCTCN2018111537-appb-100004
    时，则增强层编码块的模式选择结束，其中RD为率失真代价，RD e和RD b分别代表增强层和基本层的率失真代价。
  4. 根据权利要求1所述的一种基于压缩感知的质量可分级快速编码方法，其特征在于：所述步骤3中空间相关性子块快速模式选择，当步骤3.7没有生效时，进入步骤3.8；所述步骤3.8：利用空间相关性提前结束模式选择的条件为：|z 1-z 2|-|z 3-z 4|≤k 2，其中z 1，z 2为增强层两相邻子块的量化系数，z 3，z 4为基本层两相邻子块的量化系数，k 2为通过实验所得阈值；该条件可重写为|r 1-r 2|≤Q e|r 3-r 4|/Q b+k 2Q e，其中r 1，r 2，r 3，r 4分别为z 1，z 2，z 3，z 4的DCT系数，Q b，Q e分别为基本层和增强层的量化步长；根据DCT系数的计算公式r＝∑∑d iux uvd jv，可以得到
    Figure PCTCN2018111537-appb-100005
    其中 SAD为绝对残差和，SAD 1，SAD 2为基本层相邻块绝对残差和，SAD 3和SAD 4增强层相邻块绝对残差和；因此，当基本层和增强层编码块的率失真函数值和量化步长满足条件
    Figure PCTCN2018111537-appb-100006
    时，则增强层待编码块的模式选择结束，其中RD为率失真代价，RD 1和RD 2为基本层相邻块的率失真代价，RD 3和RD 4为增强层相邻块的率失真代价。
  5. 根据权利要求1所述的一种基于压缩感知的质量可分级快速编码方法，其特征在于：所述步骤6具体步骤如下：
    所述步骤6.1：首先将8x8大小的残差矩阵变为长度为N的一维稀疏信号Θ，利用整数DCT变换，稀疏基ψ，对残差子块进行稀疏表示，得到稀疏信号X；
    步骤6.2：选用一个与稀疏基ψ满足RIP原则、大小为mx64高斯随机测量矩阵φ，其中m的计算公式为：m＝klog 2(N/k)，其中k为稀疏信号中的稀疏度，即不为0的个数；
    步骤6.3：将稀疏信号X投影到测量矩阵φ上，得到信号Y，计算公式为Y＝φ·X；
    步骤6.4：设立标志位F m并对测量值后面补上(64-m)个0后的数据进行熵编码；
  6. 根据权利要求1所述的一种基于压缩感知的质量可分级快速编码方法，其特征在于：所述步骤8中具体重构步骤如下：
    步骤8.1：初始化参数设置：残差r (0)＝y，重建信号x (0)＝0，信号的索引集为Γ (0)＝φ，迭代次数为n＝0，停止迭代判决误差ε>0；
    步骤8.2：计算残差和观测矩阵的每行内积g (n)＝φ·r (n-1)；
    步骤8.3：找出g (n)中绝对值最大的元素，即
    Figure PCTCN2018111537-appb-100007
    步骤8.4：更新索引集Γ (n)＝Γ (n-1)U{k}，及原子集合
    Figure PCTCN2018111537-appb-100008
    步骤8.5：利用最小二乘法求得近似解
    Figure PCTCN2018111537-appb-100009
    步骤8.6：更新残差r (n)＝y-x (n)；
    步骤8.7：判断是否满足迭代停止条件，若满足则停止，令x＝x (n)，输出x，否则n＝n+1，返回步骤8.1。
  7. 根据权利要求3所述的一种基于压缩感知的质量可分级快速编码方法，其特征在于：所述k 1最佳设定为2.43。
  8. 根据权利要求4所述的一种基于压缩感知的质量可分级快速编码方法，其特征在于：所述k 2最佳设定为4.31。
PCT/CN2018/111537 2018-03-22 2018-10-24 一种基于压缩感知的质量可分级快速编码方法 WO2019179096A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201810242647.3A CN108471531B (zh) 2018-03-22 2018-03-22 一种基于压缩感知的质量可分级快速编码方法
CN201810242647.3 2018-03-22

Publications (1)

Publication Number Publication Date
WO2019179096A1 true WO2019179096A1 (zh) 2019-09-26

Family

ID=63265780

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2018/111537 WO2019179096A1 (zh) 2018-03-22 2018-10-24 一种基于压缩感知的质量可分级快速编码方法

Country Status (2)

Country Link
CN (1) CN108471531B (zh)
WO (1) WO2019179096A1 (zh)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108471531B (zh) * 2018-03-22 2020-02-07 南京邮电大学 一种基于压缩感知的质量可分级快速编码方法
CN109547961B (zh) * 2018-11-29 2020-06-09 北京理工大学 一种无线传感网络中大数据量压缩感知编解码方法
CN109743571B (zh) * 2018-12-26 2020-04-28 西安交通大学 一种基于并行压缩感知多层残差系数的图像编码方法
CN109819258A (zh) * 2019-01-30 2019-05-28 东华大学 一种基于螺旋式扫描的分块压缩感知方向预测编码方法

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6363119B1 (en) * 1998-03-05 2002-03-26 Nec Corporation Device and method for hierarchically coding/decoding images reversibly and with improved coding efficiency
CN101272489A (zh) * 2007-03-21 2008-09-24 中兴通讯股份有限公司 视频图像质量增强的编解码装置与编解码方法
CN102769747A (zh) * 2012-06-29 2012-11-07 中山大学 一种基于并行迭代的分级分布式视频编解码方法及系统
CN102833542A (zh) * 2012-08-09 2012-12-19 芯原微电子(北京)有限公司 一种提高可伸缩视频编码质量增强层编码速度装置和方法
CN108471531A (zh) * 2018-03-22 2018-08-31 南京邮电大学 一种基于压缩感知的质量可分级快速编码方法

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6363119B1 (en) * 1998-03-05 2002-03-26 Nec Corporation Device and method for hierarchically coding/decoding images reversibly and with improved coding efficiency
CN101272489A (zh) * 2007-03-21 2008-09-24 中兴通讯股份有限公司 视频图像质量增强的编解码装置与编解码方法
CN102769747A (zh) * 2012-06-29 2012-11-07 中山大学 一种基于并行迭代的分级分布式视频编解码方法及系统
CN102833542A (zh) * 2012-08-09 2012-12-19 芯原微电子(北京)有限公司 一种提高可伸缩视频编码质量增强层编码速度装置和方法
CN108471531A (zh) * 2018-03-22 2018-08-31 南京邮电大学 一种基于压缩感知的质量可分级快速编码方法

Also Published As

Publication number Publication date
CN108471531B (zh) 2020-02-07
CN108471531A (zh) 2018-08-31

Similar Documents

Publication Publication Date Title
WO2019179096A1 (zh) 一种基于压缩感知的质量可分级快速编码方法
Chen et al. Efficient and robust image coding and transmission based on scrambled block compressive sensing
Aaron et al. Wyner-Ziv residual coding of video
JP5384694B2 (ja) 多層ビデオ設計のためのレート制御
JP6210948B2 (ja) 画像推定方法
KR101166732B1 (ko) 추정된 코딩 비용을 이용하는 비디오 코딩 모드 선택
TW201301900A (zh) 視訊解碼器之編碼成位元串之解碼方法
TW201906412A (zh) 執行具有切片邊界預測之固定尺寸切片編碼之方法與裝置
Schroeder et al. Block structure reuse for multi-rate high efficiency video coding
CN100508608C (zh) 一种无预测循环的抗误码视频编解码方法
CN100551060C (zh) 一种视频编解码方法
CN101115200A (zh) 一种有效的运动矢量可伸缩编码方法
KR100801967B1 (ko) 문맥 기반 적응적 가변 길이 부호화 인코더 및 디코더,문맥 기반 적응성 가변 길이 부호화하는 방법과 복호화하는방법 및 이를 이용한 동영상 전송 시스템.
JP6210949B2 (ja) 画像推定方法
Milani et al. Distributed video coding based on lossy syndromes generated in hybrid pixel/transform domain
CN111355952B (zh) 一种编码系数消除方法及改进型hevc编码方法
CN109495745B (zh) 一种基于逆量化/逆变换的无损压缩解码方法
Taieb et al. Spatial correlation-based side information refinement for distributed video coding
KR102335184B1 (ko) 복합 모션-보상 예측
CN1514651A (zh) 低码率下的高效小波视频编解码方法
Belyaev Fast Decoding and Parameters Selection for CS-JPEG Video Codec
Letaief et al. Real time HEVC video compression using Hybrid DWT-DCT algorithm on an FPGA
Sun et al. A fast quality scalable video coding method based on compressed sensing
Gan et al. Wyner-Ziv coding of video using compressive sensing without feedback channel
Xu et al. Sub-sampling framework of distributed video coding

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 18910610

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 18910610

Country of ref document: EP

Kind code of ref document: A1