WO2015013850A1 - Hevc视频编码中变换量化方法和视频编码装置 - Google Patents

Hevc视频编码中变换量化方法和视频编码装置 Download PDF

Info

Publication number
WO2015013850A1
WO2015013850A1 PCT/CN2013/080278 CN2013080278W WO2015013850A1 WO 2015013850 A1 WO2015013850 A1 WO 2015013850A1 CN 2013080278 W CN2013080278 W CN 2013080278W WO 2015013850 A1 WO2015013850 A1 WO 2015013850A1
Authority
WO
WIPO (PCT)
Prior art keywords
residual
block
zero
value
residual block
Prior art date
Application number
PCT/CN2013/080278
Other languages
English (en)
French (fr)
Inventor
吕正光
王荣刚
王振宇
董胜富
高文
王文敏
李英
Original Assignee
北京大学深圳研究生院
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 北京大学深圳研究生院 filed Critical 北京大学深圳研究生院
Priority to CN201380003163.9A priority Critical patent/CN104247418A/zh
Priority to PCT/CN2013/080278 priority patent/WO2015013850A1/zh
Publication of WO2015013850A1 publication Critical patent/WO2015013850A1/zh

Links

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/134Methods 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
    • H04N19/154Measured or subjectively estimated visual quality after decoding, e.g. measurement of distortion
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/102Methods 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
    • H04N19/12Selection from among a plurality of transforms or standards, e.g. selection between discrete cosine transform [DCT] and sub-band transform or selection between H.263 and H.264
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/169Methods 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
    • H04N19/17Methods 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
    • H04N19/176Methods 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

Definitions

  • the present invention relates to the field of video coding technologies, and in particular, to a transform quantization method and a video coding apparatus in HEVC video coding.
  • HD and ultra-high resolution video encoding It is a key device to improve the quality of video services.
  • the mainstream video coding standards applied by software-based video coding in the market are mainly H.2.64 and AVS video coding standards, and the next-generation HEVC video coding standard for H.264 has just become an international standard.
  • the more common method for improving the coding efficiency of the H.264 and AVS video coding standards is to use a fast algorithm in the intra or inter prediction process to reduce the number of mode selections to improve the operation speed in the encoding process.
  • HEVC High Efficiency Video Coding
  • H.264 and AVS video coding standards it has higher compression performance than the H.264 and AVS video coding standards, but also because of this, HEVC is relatively computationally complex compared to the older generation coding standards H.264 and AVS video coding standards. It greatly improves the comparison of coding efficiency in the prior art.
  • the commonly used methods can not meet the needs of the HEVC video coding standard, so it is urgent to use an effective method to reduce the computational complexity in the HEVC coding process and improve the coding: effectiveness.
  • the object of the present invention is: In view of the process of performing redundant transform quantization operations on a large number of all zero blocks in the HEVC encoding process, the coding efficiency of HEVC is reduced, and therefore, the video coding transform quantization in the HEVC video coding standard is effectively saved. The time taken by the process improves the coding efficiency, and the present invention provides a transform quantization method and a video coding apparatus in HEVC video coding.
  • the present invention first provides a transform quantization method in HEVC video coding, which includes the following steps:
  • the residual value of the MXM residual block is directly subjected to transform quantization and is stored in the second address of the memory;
  • the residual value of the M X M residual block is detected from the first address and subjected to a transform quantization operation, and the transformed quantized value is stored in the second address.
  • the present invention also provides a HEVC video coding apparatus, comprising:
  • a video signal input an encoding controller, a memory, one or more processors, and a video signal output;
  • the coding controller performs coding control on the original image data input by the video signal input end;
  • the processor is configured to perform operations on the original image data and the inter-predicted image data, and the residual value of the ⁇ M residual block generated during the operation is stored in the memory First address;
  • One or more processors are also used to execute instructions including the following steps:
  • the ⁇ X ⁇ residual block is detected from the first address. And performing a transform quantization operation, and storing the transform quantized value in the second address; encoding the transform quantized value of the non-all zero block stored in the second address and the transform quantized assignment of all zero blocks, and The image data after the encoding operation is output from the video signal output terminal.
  • the beneficial effects of the present invention are: when the video encoding apparatus performs video encoding, by comparing the all-zero block detection threshold T M x M of the M x M residual block with the absolute error and SAD, the residual of the VC ⁇ M in the HEVC encoding process
  • the block performs a pre-decision before performing the change quantization operation to select all zero blocks in the residual block. For the all-zero block that is determined, when the video coding fast transform quantization is implemented, the pre-judgment can be directly skipped to all zeros.
  • the residual block of the block avoids transform quantization operation, thereby reducing the computational complexity of the HEVC encoding process, improving coding efficiency, and saving coding time.
  • FIG. 1 is a flowchart of a transform quantization method in HEVC video coding according to Embodiment 1 of the present invention
  • FIG. 2 is a second flowchart of a transform quantization method in HEVC video coding according to Embodiment 1 of the present invention
  • FIG. 3 is a third flowchart of a transform quantization method in HEVC video coding according to Embodiment 1 of the present invention.
  • FIG. 4 is a fourth flowchart of a transform quantization method in HEVC video coding according to Embodiment 1 of the present invention.
  • FIG. 5 is a flow chart of video coding performed by a HEVC video coding apparatus according to Embodiment 2 of the present invention.
  • a main factor affecting the coding efficiency is: performing a transform quantization operation on a large number of all zero blocks, and thus, for example, in the HEVC interframe or intra coding process
  • the transform quantization operation of these all zero blocks can be skipped, the computation load of HEVC is reduced, the computational complexity in the HEVC encoding process is reduced, and the coding efficiency is improved.
  • the present invention provides an all-zero block pre-decision method and a video coding fast transform quantization implementation method.
  • HEVC High Efficiency Video Coding, is a new video coding standard
  • SAD Sum of Absolute Differences, Absolute Error and
  • SAO Session-At-Once, multi-session burning.
  • CAB AC Context-based Adaptive Binary Arithmetic Coding, Adaptive Binary Arithmetic Coding.
  • this embodiment provides a transform quantization method in HEVC video coding, which includes the following steps:
  • the residual value of the ⁇ X ⁇ residual block is detected from the first address and subjected to a transform quantization operation, and the transformed quantized value is stored in the second address.
  • the residual block is an all-zero block
  • the residual value is directly subjected to transform and quantization, and the residual value of the residual block determined to be all zero blocks is set to zero, and the transformation of all zero blocks is skipped. Quantify operations.
  • SAD ⁇
  • the step of calculating the all-zero block detection threshold T M x M corresponding to the residual value of the MXM residual block specifically includes the following sub-steps:
  • An all-zero block detection threshold ⁇ ⁇ corresponding to the residual value of the ⁇ residual block is obtained .
  • the 4 ⁇ 4 residual is calculated according to the transform coefficient value of the 4 ⁇ 4 residual block and the transform quantization formula.
  • the residual zero difference detection threshold T 4 corresponding to the residual value of the block, the calculated all zero block detection threshold T 4 satisfies the following relationship:
  • the integer transform in HEVC is defined as:
  • E(x, y) ⁇ C(x, u)e(u, v)C(y, v) , 0 ⁇ x, y ⁇ 2;
  • E(x,y) Q sigii(E(x, y)) x (
  • the absolute value of the transform coefficient has the following relationship:
  • the all-zero block detection threshold ⁇ 4 ⁇ 4 corresponding to the residual value of the 4 ⁇ 4 residual block can be obtained as follows:
  • ⁇ residual block includes 4 x 4 residual block, 8 x 8 residual block, 16 x 16 residual block and 32 X 32 residual block
  • the residual difference of MXM residual block includes 4 x 4 residual
  • the corresponding all-zero block detection threshold T S 2T 4x4 ;
  • the all-zero block detection threshold corresponding to the residual value of the 16 x 16 residual block T 16x l6 4T 4x4 ;
  • the corresponding all-zero block detection threshold T 32 2 8T 4 , wherein the flow of the transform quantization method of the 4 ⁇ 4 residual block, the 8 ⁇ 8 residual block, the 16 ⁇ 16 residual block, and the
  • the transform coding method in video coding in this embodiment is applicable to the inter and inter mode of HEVC coding, and is preferably applied to the inter mode of HEVC coding, and uses the transform coefficient value and the transform quantization formula to calculate the all zero block detection threshold T. 4M , according to the absolute error and the relationship between 8 0 « ⁇ and 8 0 ⁇ 4 , the full zero block detection threshold T MxM of various ⁇ residual blocks can be obtained, such as 4 x 4 residual block, 8 x 8 residual block 16 X 16 residual block and 32 ⁇ 32 residual block. According to the relationship between SAD M ⁇ and ⁇ , all zero blocks in these residual blocks can be pre- decisive before the HEVC code change operation.
  • the transform quantization operation can be skipped directly, and the transform quantization operation performed thereon is omitted, so as to reduce the computational complexity of the HEVC encoding process, improve the coding efficiency, and save the coding time.
  • Embodiment 2
  • the embodiment provides a HEVC video coding apparatus, including:
  • a video signal input an encoding controller, a memory, one or more processors, and a video signal output;
  • the encoding controller performs encoding control on the original image data input by the video signal input end; the processor is configured to perform operations on the original image data and the inter-predicted image data, and the residual value of the MXM residual block generated during the operation is stored in the memory An address
  • One or more processors are also used to execute instructions including the following steps:
  • the residual value of the MXM residual block is directly subjected to transform quantization and is stored in the second address of the memory; Otherwise, if the M x M residual block is a non-all zero block, the residual value of the MXM residual block is detected from the first address and a transform quantization operation is performed, and the transformed quantized value is stored in the second address;
  • the transform quantized value of the non-all zero block stored in the second address and the transform quantized value of all zero blocks are encoded, and the encoded image data is output from the video signal output terminal.
  • e(x, y) is the residual data of the ⁇ residual block, that is, the residual value of the M ⁇ M residual block.
  • the step of calculating the all-zero block detection threshold T MxM corresponding to the residual value of the MXM residual block specifically includes the following sub-steps:
  • the processor first calculates an all-zero block detection threshold T 4x corresponding to the residual value of the 4x4 residual block
  • the video encoding device uses the HEVC video codec standard for video encoding and decoding. After the original image data is input through the video signal input terminal, the video encoding and decoding starts, and the original image data is subjected to codec control by the universal encoding controller. When performing the calculation of the original image data and the inter-predicted image data, a large number of residual blocks are generated, corresponding to the residual block. The residual value is stored in the first address of the memory. Then, the processor performs the transform quantization operation according to the transform quantization method of the first embodiment. The encoding operation flow after the transform quantization is the same as the prior art. Please refer to FIG. 5 . I won't go into details here.
  • the video encoding apparatus in this embodiment performs the all-zero block pre-decision process before the transform quantization process, and does not perform the transform quantization operation on the residual block which is the all-zero block obtained through the pre-decision process.
  • the computational complexity of the HEVC encoding process will undoubtedly be greatly reduced, thereby improving the coding efficiency, saving the coding time, and realizing the fast transform quantization of HEVC video coding.

Landscapes

  • Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Physics & Mathematics (AREA)
  • Discrete Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Compression Or Coding Systems Of Tv Signals (AREA)

Abstract

本发明涉及HEVC视频编码中变换量化方法和视频编码装置,方法包括:从存储器的第一地址检测原始图像数据与帧间预测图像数据运算时产生的M×M残差块的残差值;计算与该残差值相对应的全零块检测阈值TMxM;将计算的全零块检测阈值TMxM与该残差值的绝对误差和SADMxM比较;若SADMxM<TMxM,则Μ×Μ残差块为全零块,对其残差值直接进行变换量化赋值并存储于存储器的第二地址;否则,Μ×Μ残差块为非全零块,从第一地址检测其残差值进行变换量化操作,并将变换量化值存储于第二地址。HEVC视频编码时,直接跳过为全零块的残差块,不对其进行变换量化操作,降低运算复杂度,提高编码效率,节省编码时间。

Description

HEVC视频编码中变换量化方法和视频编码装置 技术领域
本发明涉及视频编码技术领域, 具体涉及一种 HEVC视频编码中变 换量化方法和视频编码装置。
背景技术
随着视频服务的普及, 高分辨率的视频得到越来越广泛的应用, 尤 其是街前高清分辨率视频以及超高清分辨率视频开始进入网络视频服 务, 因此, 高清及超高清分辩率视频编码器成为提高视频服务质量的关 键设备。
虽然目前已经有一系列基于硬件编码芯片, 或半硬件嵌入式系统如 FPGA和 DSP等高清视频编码器投入市场, 但基于硬件的编码系统存在 开发周期长, 产品定型后不便于修改, 不便于搭建分布式编码云平台等 缺点。 因此, 基千软件的视频编码仍然是目前基于互联网视频服务的一 个主要的编码方案。
目前, 市场上基于软件的视频编码所应用的主流视频编码标准主要 为 H.2.64及 AVS视频编码标准,对于 H.264的下一代 HEVC视频编码标 准则刚刚成为国际标准。 现有技术中, H.264及 AVS视频编码标准提 高编码效率的比较常用的方法是在帧内或者帧间预测过程中利用快速算 法, 降低模式选择的数量, 以提高编码过程中的运算速度。
对于 HEVC视频编码标准, 其相对 H.264及 AVS视频编码标准, 具有更高的压缩性能 但也正因为如此, HEVC相对于老一代编码标准 H.264及 AVS视频编码标准其运算复杂度也相对大大提高, 现有技术中 提高编码效率的比较.常用的方法已经不能满足 HEVC视频编码标准的需 要, 所以迫切需要釆用一种有效的方法来降低 HEVC编码过程中的运算 复杂度, 提高编码:效率。
在利用 HEVC视频编码标准进行编码的过程中, 影响其编码效率的 一个主要因素为变换量化操作过程中大量全零块的存在, 所谓全零块, 指如杲残差块的残差信号经过变换量化后系数为零, 这样的残差块就称 为全零块, 很显然, 对于全零块而言, 变换量化搡作都是多余的, 由于 HEVC本身编码过程的运算复杂性, 再 rp上对全零块所进行的多余的变 换量化操作, 无疑会影响其编码效率。 发明内容
本发明的目的是: 鉴于 HEVC编码过程中, 存在对大量的全零块进 行多余的变换量化操作的过程, 降低了 HEVC的编码效率, 因此, 为有 效的节省 HEVC视频编码标准中视频编码变换量化过程所用的时间, 提 高编码效率, 本发明提供一种 HEVC视频编码中变换量化方法和视频编 码装置。
为实现上述目的, 本发明采用的技术方案如下:
本发明首先提供一种 HEVC视频编码中变换量化方法, 包括如下步 骤:
从存储器的第一地址检测原始图像数据与帧间预测图像数据运算时 产生的 ΜχΜ残差块的残差值;
计算与 ΜχΜ残差块的残差值相对应的全零块检测阈值 ΤΜχΜ; 将计算出的全零块检测阈值!^^与 ΜχΜ残差块的残差值的绝对 误差和 SADMxM比较;
若 SADMxM<TMxM, M X M残差块为全零块, 则对 M X M残差块的 残差值直接进行变换量化赋值, 并存储于存储器的第二地址;
否则, M X M残差块为非全零块 , 则从第一地址检测 M X M残差块 的残差值并进行变换量化操作, 并将变换量化值存储于第二地址。
其次, 本发明还提供一种 HEVC视频编码装置, 包括::
视频信号输入端、 编码控制器、 存储器、 一个或多个处理器及视频 信号输出端;
编码控制器对视频信号输入端输入的原始图像数据进行编码控制; 处理器用于进行原始图像数据与帧间预测图像数据的运算, 运算时 产生的 Μχ M残差块的残差值存储于存储器的第一地址;
一个或多个处理器还用于执行包括以下步骤的指令:
从第一地址检测 M X M残差块的残差值;
计算与 M X M残差块的残差值相对应的全零块检测阈值 TMXM; 将计算出的全零块检测阈值!^^与 ΜχΜ残差块的残差值的绝对 误差和 SADMxM比较;
若 SADMxM<TMxM, ΜχΜ残差块为全零块, 则对 ΜχΜ残差块的 残差值直接进行变换量化赋值, 并存储于存储器的第二地址;
否则, Μ X Μ残差块为非全零块, 则从第一地址检测 Μ X Μ残差块 的残差值并进行变换量化操作, 并将变换量化值存储于第二地址; 对存储于第二地址的非全零块的变换量化值及全零块的变换量化赋 值进行编码运算, 并将编码运算后的图像数据从视频信号输出端输出。
本发明的有益效果是: 视频编码装置进行视频编码时, 通过比较 M x M残差块的全零块检测阈值 TM x M与绝对误差和 SAD , 对 HEVC 编码过程中的 Μ χ M残差块在进行变化量化操作之前先进行预判决, 以 挑选出残差块中的全零块; 对于判决出的全零块, 在实现视频编码快速 变换量化时, 可以直接跳过预判决为全零块的残差块, 避免对其进行变 换量化操作, 从而降低 HEVC编码过程的运算复杂度, 提高编码效率, 节省编码时间。
附图说明
图 1为本发明实施例一中 HEVC视频编码中变换量化方法的流程图 之一;
图 2为本发明实施例一中 HEVC视频编码中变换量化方法的流程图 之二;
图 3为本发明实施例一中 HEVC视频编码中变换量化方法的流程图 之三;
图 4为本发明实施例一中 HEVC视频编码中变换量化方法的流程图 之四;
图 5为本发明实施例二中 HEVC视频编码装置进行视频编码的流程 图。
具体实施方式
下面通过具体实施方式结合附图对本发明作进一步详细说明。
在背景技术中已经介绍, 利用 HEVC视频编码标准进行编码的过程 中, 影响其编码效率的一个主要因素为: 对大量全零块进行变换量化操 作, 因此, 如杲在 HEVC帧间或帧内编码过程时提前检测出残差块中的 全零块, 就可以跳过对这些全零块的变换量化操作, 减少 HEVC的运算 负荷, 降低 HEVC.编码过程中的运算复杂度, 提高编码效率。 基于此种 构思, 本发明提供一种全零块预判决方法和视频编码快速变换量化的实 现方法。
本发明中用到的术语定义如下:
HEVC: High Efficiency Video Coding, 是一种全新的视频编码标准; SAD: Sum of Absolute Differences, 绝对误差和;
SAO: Session-At-Once, 多区段刻录。
CAB AC: Context-based Adaptive Binary Arithmetic Coding , 自适应 二进制算术编码。 实施例一
请参考图 1, 本实施例提供一种 HEVC视频编码中变换量化方法, 包括如下步骤:
从存储器的第一地址检测原始图像数据与帧间预测图像数据运算时 产生的 ΜχΜ残差块的残差值;
计算与 Μ X Μ残差块的残差值相对应的全零块检测阈值 ΤΜΧΜ; 将计算出的全零块检测阈值 1^^与 ΜχΜ残差块的残差值的绝对 误差和 SAD 比较;
若 SADMxM<T , ΜχΜ残差块为全零块, 则对 ΜχΜ残差块的 残差值直接进行变换量化赋值, 并存储于存储器的第二地址;
否则, Μ X Μ残差块为非全零块, 则从第一地址检测 Μ X Μ残差块 的残差值并进行变换量化操作, 并将变换量化值存储于第二地址。
其中, 若残差块为全零块, 则对其残差值直接进行变换量化赋值, 优选将判决为全零块的残差块的残差值置为零, 并跳过全零块的变换量 化操作。
Μχ Μ残差块的绝对误差和 SADMxM满足关系:
M M
SAD^ =^^|e(x,y), 其中, e(x,y)为 ΜχΜ残差块的残差数据, 即 M <M残差块的残差 值。
该实施例中, 请参考图 2, 计算与 M X M残差块的残差值相对应的 全零块检测阈值 TM x M的步骤具体包括如下子步骤:
先计算与 4x4残差块的残差值相对应的全零块检测阈值 T4x4; 再根据 Μ X Μ与 4 X 4变换矩阵之间的关系:
Figure imgf000006_0001
得到与 ΜχΜ残差块的残差值相对应的全零块检测阈值 ΤΜχΜ。 计算与 4 x 4残差块的残差值相对应的全零块检测阈值 I 4的子步 骤中, 根据 4 x 4残差块的变换系数值及变换量化公式, 计算与 4 x 4残 差块的残差值相对应的全零块检测阈值 T4 , 计算的全零块检测阈值 T4 满足如下关系:
丄 2 blt - offset _ ^
^X Q[qreJ 4X4
其中, offset=85«(qbit-9) , qbit=19+QP/6, QP 为量化参数,
Figure imgf000007_0001
上述全零块检测阈值 T4M满足的关系式, 其推导过程如下: 众所周知, 在 HEVC中有如下定义:
B=内部位深度; M=变换块大小;
其中 Q(x)= {26214,23302, 20560, 18396, 16384,
Figure imgf000007_0002
qbit=21+QP/6-M-(B-8);
残差数据 e(x, y):
e(x,y)=fi(x,y)-f2(x,y)。
在 HEVC中整数变换定义为:
E(x, y) =∑∑ C(x, u)e(u, v)C(y, v) , 0<x,y<2; 其中,
Figure imgf000007_0003
L(n) = 1/V2 , 当 n=0; L(n)=l, 其他〈x〉 相当于接近 x的整数。 量化公式为:
E(x,y)Q = sigii(E(x, y)) x (|E(x, y)| x Q[qrcm ] + offset » qbit , 其中, qbit=21+QP/6-M-(B-8)=21+QP/6-2-(8-8)=19+QP/6;
对于帧内块:
offset=171«(qbit-9);
对于帧间块:
offset=85«(qbit-9);
变换系数的绝对值有如下关系: |E(x,y)| <∑∑|e(U;r)||C(x;u)C(y;r)|
υ=ΰ v=0
从以上公式可以推出:
E(x,y), Q < X∑|e(u5v)|C(x,u)Cty,v) xQ eJ+ offset) »qbit
u=0 v=0 其中 Q[qiein]=Q(QP°/i>6), em=QP%6;
如果 |E(X, y)Q|≤ 1 , 我们可以得到变换量化结杲为全零块: 根据
|E(x,y)| <∑∑|e(u,v)|C(x,u)Cty,v)|
i=0 ν=ϋ
我们有:
E(x,y) {u=0,v=0} < Z∑e(u,v) = 13SAD
u=0 v=0
E(x,y) {(u=0,v≠0 J(u≠0,v=0)} < 5 cos(-))^ e(u, v) = 18SAD
8 =0 v=0
E(x, y)L {u≠。0,v≠。0}}≤ ( 55φ ){ 5 e(uv) = 25SAD
Figure imgf000008_0001
通过计算有:
25 i (u'v)lxQ[q腿] < 2 b11 - offset
u=(J v=0
从而可以得到如下与 4χ4 残差块的残差值相对应的全零块检测阈 值 Τ4χ4:
丄 2 b" - offset
X Qk 4X4
对于 M X M残差块,根据不同大小的变换块变换矩阵之间的关系和 DCT变换公式, 我们可以知道, 针对于不同的 ΜχΜ残差块, 对应的全 块检测阈值 ΤΜ < Μ公式如下:
ΤΜ,Μ= χτ4χ4, 其中, Μ=32, 16,8, 4。 即: ΜχΜ残差块包括 4 x 4残差块、 8 x 8残差块、 16 x 16残差块 和 32 X 32残差块而, M X M残差块的残差值对应包括 4 x 4残差块的残 差值、 8 x 8残差块的残差值、 16 X 16残差块的残差值和 32 X 32残差块 的残差值, 8 x 8残差块的残差值所对应的全零块检测阈值 TS =2T4x4; 16 x 16残差块的残差值所对应的全零块检测阈值 T16x l6=4T4x4; 32 χ 32 残差块的残差值所对应的全零块检测阈值 T32 2=8T4 , 其中, 4 x 4残 差块、 8 x 8残差块、 16 X 16残差块和 32 x 32残差块的变换量化方法的 流程请参考图 3和图 4。
本实施例中的视频编码中变换量化方法适用于 HEVC编码的帧间和 帧内模式, 优选适用于 HEVC编码的帧间模式中, 利用变换系数值及变 换量化公式, 计算全零块检测阈值 T4M, 根据绝对误差和 8 0«^与 8 0^4的关系, 能得出多种 ΜχΜ残差块的全零块检测阈值 TMxM, 如 4 x 4残差块、 8 x 8残差块、 16 X 16残差块和 32 χ 32残差块,根据 SADM χΜ与 ΤΜχΜ的关系比较, 可以在 HEVC编码变化量换操作前对这些残差 块中的全零块进行预判决, 对于为全零块的残差块, 可以直接跳过其变 换量化操作, 进而省去对其进行的变换量化操作, 以降低 HEVC编码过 程的运算复杂度, 提高编码效率, 节省编码时间。 实施例二
本实施例提供一种 HEVC视频编码装置, 包括:
视频信号输入端、 编码控制器、 存储器、 一个或多个处理器及视频 信号输出端;
编码控制器对视频信号输入端输入的原始图像数据进行编码控制; 处理器用于进行原始图像数据与帧间预测图像数据的运算, 运算时 产生的 M X M残差块的残差值存储于存储器的第一地址;
一个或多个处理器还用于执行包括以下步骤的指令:
从第一地址检测 M X M残差块的残差值;
计算与 M X M残差块的残差值相对应的全零块检测阈值 TMxM; 将计算出的全零块检测阈值!^^与 ΜχΜ残差块的残差值的绝对 误差和 SADMxM比较;
若 SADMXM<TM M, M X M残差块为全零块, 则对 M X M残差块的 残差值直接进行变换量化赋值, 并存储于存储器的第二地址; 否则, M x M残差块为非全零块 , 则从第一地址检测 M X M残差块 的残差值并进行变换量化操作, 并将变换量化值存储于第二地址;
对存储于第二地址的非全零块的变换量化值及全零块的变换量化赋 值进行编码运算, 并将编码运算后的图像数据从视频信号输出端输出。
其中, Μχ M残差块的残差值的绝对误差和 SADMxM满足关系:
Figure imgf000010_0001
x=0 y=0
其中, e(x,y)为 ΜχΜ残差块的残差数据, 即 M <M残差块的残差 值。
该实施例中, 计算与 M X M残差块的残差值相对应的全零块检测阈 值 TMxM的步骤具体包括如下子步骒:
处理器先计算与 4x4残差块的残差值相对应的全零块检测阈值 T4x
4;
再根据 Μ X Μ与 4 X 4变换矩阵之间的关系:
得到与 Μ X Μ残差块的残差值相对应的全零块检测阈值 ΤΜχΜ 计算与 4x4残差块的残差值相对应的全零块检测阈值 Τ4χ4的子步 骤中, 根据 4x4残差块的变换系数值及变换量化公式, 计算与 4x4残 差块的残差值相对应的全零块检测阈值 Τ, 计算的全零块检测阈值 Τ4 满足如下关系:
丄 2 b" - offset
X Qk 4X4 ,
其中, offset=85«(qbit-9) , qbit=19+QP/6 QP 为量化参数, q =QP%6
全零块检测阈值 T4 满足的关系推导过程与实施例一相同, 这里不 再赘述。
该实施例中, 视频编码装置采用 HEVC视频编解码标准进行视频编 解码, 原始图像数据通过视频信号输入端输入后, 视频编解码开始, 原 始图像数据经过通用编码控制器进行编解码控制, 处理器进行原始图像 数据与帧间预测图像数据的运算时会产生大量的残差块, 残差块所对应 的残差值则存储在存储器的第一地址, 之后, 处理器按照实施例一的变 换量化方法进行变换量化操作, 变换量化搡作后的编码运算流程与现有 技术相同, 请参考图 5 , 这里不再赘述。
该实施例中的视频编码装置, 由于在变换量化过程前先进行全零块 预判决过程, 而对于经过预判决过程所得到的为全零块的残差块, 不再 进行变换量化操作, 这对于有大量全零块存在的 HEVC编码过程来说, 无疑会很大程度的降低 HEVC编码过程的运算复杂度, 从而提高编码效 率, 节省编码时间, 实现 HEVC视频编码的快速变换量化。 本领域技术人员可以理解, 上述实施方式中各种方法的全部或部分 步骤可以通过程序来指令相关硬件完成, 读程序可以存储于一计算机可 读存储介质中, 存储介质可以包括: 只读存储器、 随机存储器、 磁盘或 光盘等。 以上所述仅为本发明的较佳实施例, 只是用于帮助理解本发明并不 用以限制本发明。 对于本领域的一般技术人员, 依据本发明的思想, 可 以对上述具体实施方式进行变化。

Claims

权 利 要 求
1.一种 HEVC视频编码中变换量化方法, 包括如下步骤:
从存储器的第一地址检测原始图像数据与帧间预测图像数据运算时 产生的 ΜχΜ残差块的残差值;
计算与 Μ X Μ残差块的残差值相对应的全零块检测阈值 ΤΜΧΜ; 将计算出的全零块检测阈值 1^^与 ΜχΜ残差块的残差值的绝对 误差和 SADMxM比较;
若 SADMxM<TMxM, ΜχΜ残差块为全零块, 则对 Μχ Μ残差块的 残差值直接进行变换量化赋值, 并存储于存储器的第二地址;
否则, M X M残差块为非全零块, 则从第一地址检测 M X M残差块 的残差值并进行变换量化操作, 并将变换量化值存储于第二地址。
2. 如权利要求 1 所述的 HEVC视频编码中变换量化方法, 其特征 在于, 所述计算与 M X M残差块的残差值相对应的全 I块检测阈值 TMx M的步骤包括如下子步骤:
先计算与 4x4残差块的残差值相对应的全零块检测阈值 T4x4;
再根据 Μ X Μ与 4 X 4变换矩阵之间的关系:
Figure imgf000012_0001
得到与 ΜχΜ残差块的残差值相对应的全零块检测阈值 ΤΜχΜ
3. 如利要求 2所述的 HEVC视频编码中变换量化方法, 其特征在 于, 所述计算与 4x4残差块的残差值相对应的全零块检测阈值 T4 的 子步骤中, 根据 4x4残差块的变换系数值及变换量化公式, 计算与 4χ 4残差块的残差值相对应的全零块检测阈值 Τ, 计算的全零块检测阈 值!^4满足如下关系:
丄 2 b" - offset
X Qk 4X4 ,
其中, offset=85«(qbit-9) , qbit=19+QP/6, QP 为量化参数,
Figure imgf000012_0002
4.如权利要求 1-3任一所述的 HEVC视频编码中变换量化方法, 其 特征在于, ΜχΜ残差块的残差值包括 4x4残差块的残差值、 8x8残 差块的残差值、 16x 16残差块的残差值和 32 x 32残差块的残差值, 8χ 8残差块的残差值所对应的全零块检测阈值 TSxS=2T4 ; 16 X 16残差块 的残差值所对应的全零块检测阈值 T16 6=4T4M; 32 X 32 残差块的残差 值所对应的全零块检测阈值 Τ32 χ 32=8Τ4 χ 4。
5.如权利要求 1-3任一所述的 HEVC视频编码中变换量化方法, 其 特征在于, Μχ Μ残差块的残差值的绝对误差和 SADMxM满足关系:
M M
SAD = |e(x,y),
x=0 y=0
其中, e(x,y)为 ΜχΜ残差块的残差值。
6. 一种 HEVC视频编码装置, 包括:
视频信号输入端、 编码控制器、 存储器、 一个或多个处理器及视频 信号输出端;
编码控制器对视频信号输入端输入的原始图像数据进行编码控制; 处理器用于进行原始图像数据与帧间预测图像数据的运算, 运算时 产生的 ΜχΜ残差块的残差值存储于存储器的第一地址;
一个或多个处理器还用于执行包括以下步骤的指令:
从第一地址检测 M X M残差块的残差值;
计算与 ΜχΜ残差块的残差值相对应的全零块检测阈值 ΤΜχΜ; 将计算出的全零块检测阈值!^^与 ΜχΜ残差块的残差值的绝对 误差和 SADMxM比较;
若 SADMxM<T , M X M残差块为全零块, 则对 M X M残差块的 残差值直接进行变换量化赋值, 并存储于存储器的第二地址;
否则, M X M残差块为非全零块 , 则从第一地址检测 M X M残差块 的残差值并进行变换量化操作, 并将变换量化值存储于第二地址;
对存储于第二地址的非全零块的变换量化值及全零块的变换量化 赋值进行编码运算,并将编码运算后的图像数据从视频信号输出端输出。
PCT/CN2013/080278 2013-07-29 2013-07-29 Hevc视频编码中变换量化方法和视频编码装置 WO2015013850A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
CN201380003163.9A CN104247418A (zh) 2013-07-29 2013-07-29 Hevc视频编码中变换量化方法和视频编码装置
PCT/CN2013/080278 WO2015013850A1 (zh) 2013-07-29 2013-07-29 Hevc视频编码中变换量化方法和视频编码装置

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2013/080278 WO2015013850A1 (zh) 2013-07-29 2013-07-29 Hevc视频编码中变换量化方法和视频编码装置

Publications (1)

Publication Number Publication Date
WO2015013850A1 true WO2015013850A1 (zh) 2015-02-05

Family

ID=52231762

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2013/080278 WO2015013850A1 (zh) 2013-07-29 2013-07-29 Hevc视频编码中变换量化方法和视频编码装置

Country Status (2)

Country Link
CN (1) CN104247418A (zh)
WO (1) WO2015013850A1 (zh)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10511859B2 (en) * 2016-02-04 2019-12-17 Mediatek Inc. Method and apparatus for image compression without residue signaling
CN107613292B (zh) * 2016-07-11 2020-04-07 合肥君正科技有限公司 一种图像复杂度评估方法及装置
CN107566846B (zh) * 2017-09-20 2020-04-14 北京字节跳动网络技术有限公司 视频编码跳过模式决策方法、装置、设备及存储介质
CN115174908B (zh) * 2022-06-30 2023-09-15 北京百度网讯科技有限公司 视频编码的变换量化方法、装置、设备以及存储介质

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1770863A (zh) * 2004-11-02 2006-05-10 华为技术有限公司 零块预判条件的确定方法及零块预先判定方法
CN101795404A (zh) * 2009-10-21 2010-08-04 深圳市融创天下科技发展有限公司 基于量化步长的零块阈值确定方法及零块预先判断方法
CN102447907A (zh) * 2012-01-31 2012-05-09 北京工业大学 一种针对hevc的视频序列的编码方法
US20120189052A1 (en) * 2011-01-24 2012-07-26 Qualcomm Incorporated Signaling quantization parameter changes for coded units in high efficiency video coding (hevc)
CN102685499A (zh) * 2012-05-07 2012-09-19 西安交通大学 一种基于能量守恒的全零块检测方法

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1770863A (zh) * 2004-11-02 2006-05-10 华为技术有限公司 零块预判条件的确定方法及零块预先判定方法
CN101795404A (zh) * 2009-10-21 2010-08-04 深圳市融创天下科技发展有限公司 基于量化步长的零块阈值确定方法及零块预先判断方法
US20120189052A1 (en) * 2011-01-24 2012-07-26 Qualcomm Incorporated Signaling quantization parameter changes for coded units in high efficiency video coding (hevc)
CN102447907A (zh) * 2012-01-31 2012-05-09 北京工业大学 一种针对hevc的视频序列的编码方法
CN102685499A (zh) * 2012-05-07 2012-09-19 西安交通大学 一种基于能量守恒的全零块检测方法

Also Published As

Publication number Publication date
CN104247418A (zh) 2014-12-24

Similar Documents

Publication Publication Date Title
US10841583B2 (en) Coding unit depth determining method and apparatus
WO2016138779A1 (zh) 帧内编解码方法、编码器和解码器
TWI344793B (en) Power aware method and apparatus of video decoder on a multi-core platform
WO2016131229A1 (zh) 用于视频图像编码和解码的方法、编码设备和解码设备
WO2015013850A1 (zh) Hevc视频编码中变换量化方法和视频编码装置
Liao et al. Double H. 264/AVC compression detection using quantized nonzero AC coefficients
WO2022078163A1 (zh) 视频解码方法、视频编码方法及相关装置
WO2016180129A1 (zh) 预测模式选择方法、装置及设备
TW200942045A (en) Method for video coding
WO2021203394A1 (zh) 环路滤波的方法与装置
WO2009097809A1 (zh) 基于自适应块变换的帧内预测方法及装置
WO2020172907A1 (zh) 一种量化系数结束标志位的上下文模型选取方法及装置
WO2023045420A1 (zh) 图像处理方法及装置、电子设备和存储介质
WO2023131059A1 (zh) 图像编码方法、图像编码装置、电子设备和可读存储介质
JP7482536B2 (ja) 適応的な数の領域を伴う幾何学的分割のための形状適応離散コサイン変換
WO2011047520A1 (zh) 基于量化步长的零块阈值确定方法及零块预先判断方法
WO2018014301A1 (zh) 视频编码方法及装置
WO2006047936A1 (en) A method for determining the condition in zero block prejudgment and for prejudging zero block
TWI249291B (en) Method of video compression that accommodates scene changes
WO2022116246A1 (zh) 帧间预测方法、视频编解码方法、装置及介质
Zhang et al. Globally variance-constrained sparse representation and its application in image set coding
WO2023231414A1 (zh) 率失真优化量化方法及装置
WO2023173809A1 (zh) 视频编解码方法、装置、存储介质、电子设备及计算机程序产品
CN116095341A (zh) 一种水印嵌入方法、装置、设备以及存储介质
WO2018205781A1 (zh) 一种实现运动估计的方法及电子设备

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: 13890463

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: 13890463

Country of ref document: EP

Kind code of ref document: A1