WO2018149092A1 - 一种局部视差矢量的导出方法 - Google Patents
一种局部视差矢量的导出方法 Download PDFInfo
- Publication number
- WO2018149092A1 WO2018149092A1 PCT/CN2017/093881 CN2017093881W WO2018149092A1 WO 2018149092 A1 WO2018149092 A1 WO 2018149092A1 CN 2017093881 W CN2017093881 W CN 2017093881W WO 2018149092 A1 WO2018149092 A1 WO 2018149092A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- current
- disparity vector
- block
- disparity
- vector
- Prior art date
- Legal status (The legal status 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 status listed.)
- Ceased
Links
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
- H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
- H04N19/597—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding specially adapted for multi-view video sequence encoding
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N13/00—Stereoscopic video systems; Multi-view video systems; Details thereof
- H04N13/10—Processing, recording or transmission of stereoscopic or multi-view image signals
- H04N13/106—Processing image signals
- H04N13/161—Encoding, multiplexing or demultiplexing different image signal components
Definitions
- the present invention relates to the field of multi-view video coding technologies, and in particular, to a method for deriving a local disparity vector in three-dimensional multi-view video coding.
- 3D multi-view video has been favored for providing richer visual information and more immersive viewing effects. Since the number of viewpoints of three-dimensional multi-view video using two or more cameras is greatly increased, the amount of data is greatly increased compared with the conventional two-dimensional video, and efficient three-dimensional multi-view video compression coding is very important.
- the international video standardization organization MPEG and ITU-T VCEG jointly developed the 3D video compression coding standard 3D-HEVC (High Efficiency Video Coding).
- 3D-HEVC High Efficiency Video Coding
- 3D-AVS Audio Video Coding Standard
- 3D multi-view video coding In addition to spatial redundancy, inter-turn redundancy, and information redundancy, 3D multi-view video coding also has very large inter-view redundancy compared to traditional 2D video: due to the characteristics of multi-view shooting, the same Engraving, multiple cameras shoot the same scene from different angles, the background in the scene and the moving objects are the same, the texture information of the same object between each viewpoint and the motion information of the object are very strong. Relevance.
- some inter-view prediction tools using disparity vectors such as disparity compensation prediction techniques, inter-view motion prediction techniques, and inter-view residual prediction techniques are integrated into 3D.
- 3D-AVS has also integrated parallax compensation prediction technology and inter-view motion prediction technology. The tools that use the disparity vector are described as follows:
- the disparity compensated prediction is introduced.
- the motion compensation prediction is inter prediction performed by referring to reference frames that have been encoded with different engravings at the same viewpoint.
- the disparity compensation prediction is inter-view prediction with reference to the reference frames that have been encoded by the same viewpoint at the same time, as shown in FIG.
- Inter-view motion prediction In order to utilize motion information of reference frames already encoded in a reference view, inter-view motion prediction is introduced. In order to obtain candidate motion information of the current view of the dependent view from the reference view corresponding block, the disparity vector is first obtained for the current block, and then the obtained disparity vector is added to the inter-view reference frame already encoded in the reference view by the position of the current block. Locate the corresponding prediction block of the current block. If the corresponding prediction block adopts the inter prediction mode, the corresponding motion information can be used as the candidate motion information of the current block, as shown in FIG. 2 .
- an inter-view residual prediction is introduced.
- the obtained disparity vector is located to the corresponding block in the reference view, and the residual signal of the current block can be predicted by the residual information of the corresponding block in the reference view, as shown in FIG.
- the disparity vector can be applied to any inter-view prediction technique that needs to locate a corresponding block in the inter-view reference frame for the current block.
- 3D-HEVC and 3D-AVS adopt the following methods of acquiring disparity vectors:
- the disparity vector is derived from the neighborhood neighboring block and several spatial neighboring blocks (see L. Zhang, Y. Chen, M. Karczewicz, "CE5.h: Disparity vector Generation results," JCT2-A0097, July 2012; Jaewon Sung, Moonmo Koo, Sehoon Yea, "3D-CE5.h: Simplification of disparity vector derivation for HEVC -based 3D video coding," JCT2-A0126, July 2012) As shown in FIG.
- the spatial neighboring block in addition to the Disparity Compensated Prediction (DCP) block, the spatial neighboring block also considers a Motion Compensated Prediction block using inter-view motion prediction, which is called disparity vector motion compensation prediction. (DV-MCP) block (see J. Kang, Y. Chen, L. Zhang, M. Karczewicz, "3D-CE5.hrelated:Improvements for disparity vector derivation," Document of Joint Collaborative Team on 3D Video CodingExtension Development, JCT3V -B0047,
- the motion information of the DV-MCP block is obtained from the motion information of the block corresponding to the inter-view reference frame, and the position of the corresponding block is obtained by disparity vector positioning, which is one of the sources of the disparity vector DV to be derived.
- disparity vector positioning is one of the sources of the disparity vector DV to be derived.
- the related patents relate to a multi-view video coding method, which uses multiple coding modes for coding units in a secondary view, wherein multiple coding modes include a merge mode, a skip mode, an inter mode, and an intra mode, and the coding unit is allowed to be divided into smaller coding units
- the multi-view video coding method includes: receiving input data related to a current coding unit, where the current coding unit Corresponding to a current texture unit or a current depth unit of a current image in the current secondary view; determining one or more encoding conditions associated with a current coding unit or a set of adjacent units of the current coding unit; and if the one or more coding conditions are satisfied
- the current coding unit is encoded using the merge mode or the skip mode without evaluating the coding performance of at least one of the inter mode and the intra mode.
- the multi-view video coding method described above can greatly reduce the side effects of coding without causing performance loss.
- the object of the present invention is to solve the problem that the source of the disparity vector obtained in the prior art of 3D-HEVC is too small, so that the non-zero disparity vector may not be derived according to the neighboring unit and the disparity derived by the 3D-AVS prior art using the global information is derived.
- the vector is not accurate enough, and a disparity vector derivation scheme for 3D multi-view video coding using local information is proposed.
- a method for deriving a local disparity vector characterized in that: a method for deriving a local disparity vector in multi-view video coding is specifically performed according to the following steps:
- Step 1 Select the left, upper left, upper, and upper right and lower left adjacent coding regions of the current prediction unit PU as five candidate regions, which are respectively recorded as al, b2, bl, b0, and a0, and the size of the region.
- the left side is RxW
- the upper side is RxH
- the right side is RxW
- W and H are the width and height of the current prediction block PU
- R is the area coefficient
- the initial value is 1;
- Step 2 the candidate area is divided into a number of units of lxw size, 1 and w are generally taken as 4;
- Step 3 In the five candidate regions, in a certain order, one by one is determined as follows: determining whether there are available disparity vectors in the lxw units of the current candidate region, and if so, all the disparities The disparity vector obtained by the mean value calculation is used as the local disparity vector of the current block;
- Step 4 According to the disparity vector of the current coding prediction block obtained in step 3, the corresponding block in the inter-view reference frame may be located, and the motion vector of the current block is predicted according to the motion vector of the corresponding block;
- Step 5 The neighboring block of the current block is a motion compensated prediction MCP block or intra prediction Mm block, and the disparity vector obtained in step 3 is used as a disparity vector prediction in the disparity motion compensation process;
- step 3 if all the lxw units of the current candidate region have no disparity vector that can be used, the specific implementation process is as follows: (1) examining the next candidate region; (2) if all candidates under the current R There is no disparity vector that can be used in the region, and the region coefficient R is expanded, that is, the region coefficient is increased by 1, to obtain a new candidate region; whether the candidate region has a disparity vector that can be used according to a predetermined order: if yes, the current candidate region is All of the disparity vectors that can be used are averaged to derive the disparity vector of the current block; if not, the region coefficient R is continued to be expanded until the disparity vector is derived or the neighboring region coefficient R reaches a maximum value RM.
- a method for deriving a local disparity vector in the multi-view video coding of the present invention has the following beneficial effects:
- the present invention utilizes that the neighboring block of the current block in the current frame is not the block of the disparity compensated prediction, and uses the local disparity vector derived by the current prediction unit as the prediction of the disparity vector in the disparity compensation prediction process; And using the derived disparity vector to locate the corresponding block of the current block in the inter-view reference frame, and predicting the motion vector of the current block according to the motion vector of the corresponding block.
- the derived local disparity vector of the current prediction unit is used as the disparity information corresponding to the inter-view reference frame, and can be used for inter-view motion prediction, disparity compensation prediction, or any other need to locate the corresponding block in the reference view for the current block.
- the invention enables the present invention to improve the multi-view video coding viewpoint 1.96 ⁇ 3 ⁇ 4-2.01 ⁇ 3 ⁇ 4 and the composite viewpoint 1.23%-1.32% under the condition of low delay configuration without increasing the multi-view video coding time. BD-rate value.
- FIG. 1 is a schematic diagram of inter-view compensation prediction according to an embodiment of the present invention.
- FIG. 2 is a schematic diagram of inter-view motion prediction according to an embodiment of the present invention.
- FIG. 3 is a schematic diagram of prediction of inter-view residuals according to an embodiment of the present invention.
- FIG. 4 is a schematic diagram of a ⁇ domain and a spatial neighboring block acquired by a 3D-HEVC disparity vector according to an embodiment of the present invention.
- FIG. 5 is a schematic diagram of a disparity vector motion compensation prediction block for 3D-HEVC disparity vector acquisition according to an embodiment of the present invention
- FIG. 6 is a schematic diagram of a global disparity vector of a 3D-AVS disparity vector acquisition according to an embodiment of the present invention
- FIG. 7 is a schematic diagram of five adjacent candidate regions of a current prediction unit PU according to an embodiment of the present invention.
- a method for obtaining a local disparity vector in the multi-view video coding of the present invention mainly solves the problem that the source of the disparity vector is less available in the current 3D-HEVC, and the zero vector instead of the disparity vector derived from the default is inaccurate.
- a local disparity vector acquisition method in multi-view video coding proposed by using global information to derive a disparity vector is not accurate in 3D-AVS.
- the method mainly divides the left, upper left, upper side, upper right, and lower left five candidate areas that have been encoded by the current prediction unit PU into a plurality of lxw size blocks, and checks whether all the currently selected blocks in the current candidate area are in a certain order.
- the disparity vector is included to determine whether to examine the next candidate region. If there is no disparity vector that can be used in the current candidate region, the adjacent region is expanded to obtain a new candidate region, until the disparity vector or the candidate region arrives in all the divided blocks of the current candidate region. Up to the maximum adjacent area is set, and finally, the mean value of all the disparity vectors in the current candidate area lxw block is used to obtain the local disparity vector of the current prediction unit to achieve the disparity vector acquisition.
- a method for deriving a local disparity vector in the multi-view video coding of the present invention includes:
- Step 1 Select the left, upper left, upper side, and upper right and lower left adjacent coding regions of the current prediction unit PU as five candidate regions, which are respectively recorded as al, b2, bl, b0, and a0, as shown in FIG. 7.
- the size of the area is RxW on the left side, RxH on the upper side, and RxW on the right side; where W and H are the width and height of the current prediction block PU; R is the area coefficient, the initial value is 1; the size of the adjacent area It differs depending on the width and height of the current prediction block PU.
- Step 2 the candidate area is divided into a number of units of lxw size, 1 and w are generally taken as 4;
- Step 3 In the five candidate regions, according to a certain order, one by one determines: whether there is a disparity vector that can be used in the lxw units of the current candidate region, and if so, all the disparities The disparity vector obtained by the mean value calculation is used as the local disparity vector of the current block, and the search is stopped;
- Step 4 If there are no disparity vectors that can be used in all lxw units of the current candidate region, then the next candidate region is examined;
- Step 5 If there is no disparity vector that can be used in all candidate regions under the current R, the region coefficient R is expanded, that is, the region coefficient is increased by 1, and a new candidate region is obtained; [0044] Step six, looping steps three, four, five, until the parallax vector of the current block is derived or the coefficient of the adjacent region reaches a maximum value R M stops;
- Step VII According to the disparity vector of the current coding prediction block obtained in steps 3, 4, 5, and 6, the corresponding block in the inter-view reference frame may be located, and the current block is predicted according to the motion vector of the corresponding block.
- Motion vector
- Step 8 The neighboring block of the current block is a block or intra prediction Mm block of the motion compensation prediction MCP, and the disparity vector obtained by using steps 3, 4, 5, and 6 is used as the disparity vector prediction in the disparity motion compensation process.
- a method for deriving a disparity vector in multi-view video coding according to the present embodiment is specifically obtained according to the following steps:
- Step 1 Select the left-hand, upper-left, upper-side, and upper-right and lower-left encoded regions adjacent to the current prediction unit PU as five candidate regions, which are respectively recorded as al, b2, bl, b0, and a0, as shown in FIG. 7.
- the size of the area is RxW on the left side, RxH on the upper side, and RxW on the right side; where W and H are the width and height of the current prediction block PU; R is the area coefficient, and the initial value is 1, R M Take 4; the size of the adjacent area varies according to the width and height of the current prediction block PU.
- Step 2 dividing the candidate area into a plurality of units of 4 ⁇ 4 size
- Step 3 In the five candidate regions, according to the order of the left side a1, the upper side bl, the upper left b2, the upper right b0, and the lower left a 0, one of the following is determined: determining whether the 4x4 units of the current candidate area are There is a disparity vector that can be used. If there is, the disparity vector obtained by performing the mean calculation of all disparity vectors according to formula (1.1) is used as the local disparity vector of the current block, and the search is stopped;
- ⁇ ⁇ represents the set of all DVs in the candidate region for a given R
- N R represents the set of ⁇ ⁇ Size.
- the maximum operand is fixed.
- the maximum number of current regions N R is 256. If it exceeds 256, the search is stopped according to 256.
- Step 4 If there are no disparity vectors that can be used in all 4x4 units of the current candidate region, then the next candidate region is examined;
- Step 5 If there is no disparity vector that can be used in all candidate regions under the current R, the region coefficient R is expanded, that is, the region coefficient is increased by 1, and a new candidate region is obtained;
- the present embodiment uses the derived disparity vector to locate the corresponding block of the current block in the inter-view reference frame, and predicts the motion vector of the current block according to the motion vector of the corresponding block. And when the neighboring block of the current block is not the block of the parallax compensated prediction; the derived disparity vector is used as the prediction of the disparity vector in the disparity compensated prediction process.
- the derived local disparity vector of the current prediction block is used as the disparity information corresponding to the inter-view reference frame, and can be used for inter-view motion prediction, disparity compensation prediction, or any other tool that needs to locate the corresponding block in the reference view for the current block; Therefore, the present embodiment improves the BD-mte value of the multi-view video coding viewpoint of 1.99% and improves the BD-mte value of the composite view by 1.92% on the basis of not increasing the multi-view video coding time. .
- Embodiment 2 is a diagrammatic representation of Embodiment 1
- a method for deriving a disparity vector in multi-view video coding according to the present embodiment is specifically obtained according to the following steps:
- Step 1 Select the left, upper left, upper, and upper right and lower left adjacent coding regions of the current prediction unit PU as five candidate regions, which are respectively recorded as al, b2, bl, b0, and a0, as shown in FIG. 7.
- the size of the area is RxW on the left side, RxH on the upper side, and RxW on the right side; where W and H are the width and height of the current prediction block PU; R is the area coefficient, and the initial value is 1, R M Take 4; the size of the adjacent area varies according to the width and height of the current prediction block PU.
- Step 2 dividing the candidate area into a plurality of units of 4 ⁇ 4 size
- Step 3 in the five candidate regions, according to the left side al, the upper side bl, the upper right side b0, the lower left a0, the upper left b
- the order of 2 is judged one by one: Determine whether or not these 4x4 units of the current candidate area are There is a disparity vector that can be used, and if so, the disparity vector obtained by performing the mean calculation of all the disparity vectors according to the formula (1.1) in the first embodiment is used as the local disparity vector of the current block, and the search is stopped;
- Step 4 If there are no disparity vectors that can be used in all 4x4 units of the current candidate region, then the next candidate region is examined;
- Step 5 If there is no disparity vector that can be used in all candidate regions under the current R, the region coefficient R is expanded, that is, the region coefficient is increased by 1, and a new candidate region is obtained;
- This embodiment is the same as the first embodiment, and the derived partial disparity vector of the current prediction block is used as the disparity information corresponding to the inter-view reference frame, and can be used for inter-view motion prediction, disparity compensation prediction, or any other need for current
- the tool for positioning the corresponding block in the reference view so that the present embodiment improves the BD-rate value of the multi-view video coding viewpoint by 1.96% under the condition of low delay configuration without increasing the multi-view video coding time.
- the 1.23% BD-rate value of the synthetic viewpoint is increased.
- Embodiment 3 is a diagrammatic representation of Embodiment 3
- a method for deriving a disparity vector in multi-view video coding according to the present embodiment is specifically obtained according to the following steps:
- Step 1 Select the left, upper left, upper side, and upper right and lower left adjacent coding regions of the current prediction unit PU as five candidate regions, which are respectively recorded as al, b2, bl, b0, and a0, as shown in FIG. 7.
- the size of the area is RxW on the left side, RxH on the upper side, and RxW on the right side; where W and H are the width and height of the current prediction block PU; R is the area coefficient, and the initial value is 1, R M Take 4; the size of the adjacent area varies according to the width and height of the current prediction block PU.
- Step 2 dividing the candidate area into a plurality of units of 4 ⁇ 4 size
- Step 3 In the five candidate regions, according to the order of the left side a1, the upper side bl, the upper left b2, the lower left a0, and the upper right b 0, one of the following is determined: determining whether the 4x4 units of the current candidate area are There is a disparity vector that can be used. If there is, the disparity vector obtained by performing the mean calculation of all the disparity vectors according to the formula (1.1) in the first embodiment is used as the local disparity vector of the current block, and stops. Stop searching
- Step 4 If there are no disparity vectors that can be used in all 4x4 units of the current candidate region, then the next candidate region is examined;
- Step 5 If there is no disparity vector that can be used in all candidate regions under the current R, the region coefficient R is expanded, that is, the region coefficient is increased by 1, and a new candidate region is obtained;
- This embodiment is the same as the first embodiment, and the derived partial disparity vector of the current prediction block is used as the disparity information corresponding to the inter-view reference frame, and can be used for inter-view motion prediction, disparity compensation prediction, or any other need for current
- the tool for locating the corresponding block in the reference view so that the BD-rate value of the multi-view video coding viewpoint 2.01 ⁇ 3 ⁇ 4 is improved under the low delay configuration condition without increasing the multi-view video coding time. , increased the BD-rate value of 1.28% of the synthetic viewpoint.
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
一种多视点视频编码中的局部视差矢量获取方法，将当前预测单元PU相邻已经编码的左边、左上、上边、右上、左下五个候选区域划分为若干个l×w大小的块，按照一定的顺序考察当前候选区域所有划分的块中是否含有视差矢量来决定是否考察下一候选区域，如果当前候选区域均没有可以使用的视差矢量则扩大相邻区域得到新的候选区域，直到当前候选区域所有划分的块中存在视差矢量或候选区域到达设置的最大相邻区域为止，最后利用当前候选区域l×w块中的所有视差矢量求均值获得当前预测单元的局部视差矢量。本方法解决了当前3D-HEVC中获取视差矢量的来源较少，易由默认导出的零矢量代替视差矢量导致不准确，3D-AVS中利用全局信息导出视差矢量不够准确的问题。
Description
一种局部视差矢量的导出方法
技术领域
[0001] 本发明涉及多视点视频编码技术领域， 尤其涉及三维多视点视频编码中的局部 视差矢量的导出方法。
背景技术
[0002] 近几年来， 三维多视点视频因能提供更为丰富的视觉信息和更具沉浸感的观看 效果， 被人们所青睐。 由于三维多视点视频利用两个或多个摄像机进行拍摄视 点数大大增加， 其数据量和传统的二维视频相比也大大增加， 高效的三维多视 点视频压缩编码的研究十分重要。 目前， 国际视频标准化组织 MPEG和 ITU-T VCEG联合制定 3D视频压缩编码标准 3D-HEVC(High Efficiency Video Coding)。 同吋中国数字音视频编解码技术标准工作组也正在制定具有中国自主知识产权 的 3D多视点视频编码标准 3D-AVS(Audio Video Coding Standard)^
[0003] 与传统二维视频相比， 除了空间冗余、 吋间冗余和信息冗余外， 三维多视点视 频编码还存在着非常大的视点间冗余： 由于多视点拍摄的特点， 同一吋刻， 多 个摄像机从不同的角度对同一个场景进行拍摄， 场景中的背景以及运动的物体 都是相同的， 各视点之间的同一个物体的纹理信息以及物体的运动信息都具有 很强的相关性。 为了利用视点间的这种相关性， 从而消除视点间的冗余， 一些 利用视差矢量的视间预测工具如视差补偿预测技术、 视间运动预测技术和视间 残差预测技术等被集成到 3D-HEVC中， 目前 3D-AVS也已经集成了视差补偿预测 技术和视间运动预测技术。 利用视差矢量的工具描述如下：
[0004] 视差补偿预测
[0005] 为了利用参考视点中已经编码的参考帧的纹理信息， 与运动补偿预测相似， 引 入了视差补偿预测。 运动补偿预测是参考相同视点不同吋刻已经编码的参考帧 而进行的帧间预测， 视差补偿预测是参考不同视点同一吋刻已经编码的参考帧 而进行的视间预测， 如图 1所示。
[0006] 视间运动预测
[0007] 为了利用参考视点中已经编码的参考帧的运动信息， 引入了视间运动预测。 为 了从参考视点对应块获得依赖视点当前块的候选运动信息， 首先要为当前块获 取视差矢量， 然后由当前块的位置加上获取得到的视差矢量在参考视点中已经 编码的视间参考帧中定位到当前块的对应预测块。 如果对应预测块采用的是帧 间预测模式， 那么相应的运动信息就可以作为当前块的候选运动信息， 如图 2所 示。
[0008] 视间残差预测
[0009] 为了利用参考视点中的残差信息， 引入了视间残差预测。 由获取得到的视差矢 量定位到参考视点中的对应块， 当前块的残差信号可以由参考视点中对应块的 残差信息进行预测， 如图 3所示。
[0010] 根据以上描述， 可以发现视差矢量可以应用到任何需要为当前块在视间参考帧 中定位到对应块的任何视间预测技术。 目前 3D-HEVC和 3D-AVS分别采用以下的 视差矢量的获取方式：
[0011] 3D-HEVC中的视差矢量获取
[0012] 在目前的 3D-HEVC测试模型中， 视差矢量是通过吋域邻近块和几个空域邻近 块导出的 (参见 L. Zhang, Y. Chen, M. Karczewicz, "CE5.h: Disparity vector generation results," JCT2-A0097, July 2012； Jaewon Sung, Moonmo Koo, Sehoon Yea, "3D-CE5.h: Simplification of disparity vector derivation for HEVC -based 3D video coding," JCT2-A0126, July 2012) ， 如图 4所示， 其中空域邻近块除了做视 间补偿预测 （Disparity Compensated Prediction, DCP) 块以外， 还考虑使用视间 运动预测的运动补偿预测 （Motion Compensated Prediction) 块， 被叫做视差矢 量运动补偿预测 （DV-MCP) 块 （参见 J. Kang, Y. Chen, L. Zhang, M. Karczewicz, "3D-CE5.hrelated:Improvements for disparity vector derivation,"Document of Joint Collaborative Team on 3D Video CodingExtension Development, JCT3V-B0047,
Shanghai, Oct. 2012) ， 如图 5所示。 DV-MCP块的运动信息是由视间参考帧对应 的块的运动信息得到的， 对应块的位置是由视差矢量定位得到的， 这个视差矢 量即作为需要导出的视差矢量 DV的来源之一。 在该方法中一旦从任意一个邻近 块中获取得到一个视差矢量 DV， 整个搜索过程就停止。 然而， 这种技术的视差
矢量的来源比较少， 邻近块很容易可能都不存在 DCP块和 DV-MCP块， 导致无法 导出视差矢量， 在这种情况下， 3D-HEVC利用零矢量 （0,0) 代替当前块的视差 矢量， 使得导出的视差矢量不够准确。
[0013] 3D-AVS中的视差矢量获取
[0014] 在目前的 3D-AVS测试模型中， 采用了一种特殊的全局的视差矢量获取方案 ( 参见"一种有效的视差矢量获取方法"， 马冀， 张岳欢， 张娜， 范晓鹏， 赵德斌， M3582, AVS第 51次会议， 福州， 2014年 12月） ， 利用当前帧参考帧列表中距当 前帧吋域距离最近的吋域参考帧中包含的所有视差矢量求均值， 作为当前帧的 全局视差矢量， 如图 6所示。 这种方案利用的是吋域的参考帧而并非当前帧， 并 且利用的是全局信息， 这样获取得到的视差矢量是不准确的。
[0015] 现有的相关专利， 例如申请号为 CN104301739A的中国发明专利申请， 涉及一 种多视点视频编码方法， 其将多种编码模式用于副视点中的编码单元， 其中多 种编码模式包含合并模式、 跳过模式、 帧间模式和帧内模式， 且编码单元允许 被分割成更小的编码单元， 上述多视点视频编码方法包含： 接收与当前编码单 元相关的输入数据， 其中当前编码单元对应于当前副视点中当前图像的当前纹 理单元或当前深度单元； 决定与当前编码单元或当前编码单元的相邻单元集合 相关的一或多个编码条件； 以及若上述一或多个编码条件满足， 利用合并模式 或跳过模式编码当前编码单元而无需评估帧间模式与帧内模式中至少一者的编 码性能。 以上所述的多视点视频编码方法能大幅削减编码吋间而不会带来性能 损失的副作用。
[0016] 为此提供一种更加有效的视差矢量的导出方案来提升三维多视点视频编码中视 间预测技术的编码性能十分重要。 。
技术问题
[0017] 本发明的目的是为了解决在 3D-HEVC现有技术中获得视差矢量的来源太少导 致根据邻近单元可能无法导出非零的视差矢量以及 3D-AVS现有技术利用全局信 息导出的视差矢量不够准确的问题， 而提出的一种利用局部信息的三维多视点 视频编码的视差矢量导出方案。
问题的解决方案
技术解决方案
[0018] 一种局部视差矢量的导出方法， 其特征在于： 多视点视频编码中的一种局部视 差矢量的导出方法， 具体是按照以下步骤进行的：
[0019] 步骤一、 选择当前预测单元 PU相邻的左边、 左上、 上边和右上和左下的已经 编码的区域作为 5个候选区域， 分别记作 al、 b2、 bl、 b0和 a0， 区域的大小具体 分别取左边为 RxW， 上边为 RxH， 右边为 RxW; 其中 W和 H为当前预测块 PU的 宽和高； R为区域系数， 初始值取 1 ;
[0020] 步骤二、 将候选区域划分为 lxw大小的若干个单元， 1和 w—般取为 4;
[0021] 步骤三、 在这 5个候选区域中， 按照一定的顺序， 一个一个进行如下判断： 判 断当前候选区域的这些 lxw单元中的是否有可以使用的视差矢量， 如果有， 则将 所有视差矢量进行均值计算得到的视差矢量， 作为当前块的局部视差矢量；
[0022] 步骤四、 根据步骤三得到的当前编码预测块的视差矢量， 可以定位到其在视间 参考帧中的对应块， 根据对应块的运动矢量来预测当前块的运动矢量；
[0023] 步骤五、 当前块的邻近块为运动补偿预测 MCP的块或帧内预测 Mm块吋， 禾 步骤三得到的视差矢量作为视差运动补偿过程中的视差矢量预测；
[0024] 优选地， 步骤三中若当前候选区域所有 lxw单元没有可以使用的视差矢量， 则 其具体的实现过程如下： （1)考察下一个候选区域； （2)若当前 R下所有的候选区 域都没有可以使用的视差矢量， 扩大区域系数 R， 即将区域系数增加 1， 得到新 的候选区域； 按照既定的顺序判断这些候选区域是否有可以使用的视差矢量： 如果有， 则将当前候选区域的所有可以使用的视差矢量进行均值计算导出当前 块的视差矢量； 如果没有， 则继续扩大区域系数 R直到导出视差矢量或者相邻区 域系数 R达到最大值 R M停止。
发明的有益效果
有益效果
[0025] 与现有技术相比， 本发明公幵的多视点视频编码中的一种局部视差矢量的导出 方法， 具有如下有益效果：
[0026] 本发明利用当前帧中的当前块的邻近块都不是视差补偿预测的块吋， 利用当前 预测单元导出的局部视差矢量作为视差补偿预测过程中的视差矢量的预测； 并
且用导出的视差矢量定位到当前块在视间参考帧中的对应块， 根据对应块的运 动矢量来预测当前块的运动矢量。
[0027] 因此将导出的当前预测单元的局部视差矢量作为视间参考帧对应的视差信息， 可以用于视间运动预测、 视差补偿预测或任何其他需要为当前块在参考视点中 定位到对应块的工具； 使得本发明在没有增加多视点视频编码吋间的基础上， 在低延迟配置条件下提升了多视点视频编码视点 1.96<¾-2.01<¾和合成视点 1.23%-1 .32%的 BD-rate值。
对附图的简要说明
附图说明
[0028] 通过阅读下文优选实施方式的详细描述， 各种其他的优点和益处对于本领域普 通技术人员将变得清楚明了。 附图仅用于示出优选实施方式的目的， 而并不认 为是对本发明的限制。 而且在整个附图中， 用相同的参考符号表示相同的部件
[0029] 图 1为本发明实施例提供的视间补偿预测示意图；
[0030] 图 2为本发明实施例提供的视间运动预测示意图；
[0031] 图 3为本发明实施例提供的视间残差预测示意图；
[0032] 图 4为本发明实施例提供的 3D-HEVC视差矢量获取的吋域、 空域相邻块示意图
[0033] 图 5为本发明实施例提供的 3D-HEVC视差矢量获取的视差矢量运动补偿预测块 示意图；
[0034] 图 6为本发明实施例提供的 3D-AVS视差矢量获取的全局视差矢量示意图；
[0035] 图 7为本发明实施例提供的一种当前预测单元 PU的五个相邻候选区域的示意图
本发明的实施方式
[0036] 下面将参照附图更详细地描述本公幵的示例性实施方式。 虽然附图中显示了本 公幵的示例性实施方式， 然而应当理解， 可以以各种形式实现本公幵而不应被 这里阐述的实施方式所限制。 相反， 提供这些实施方式是为了能够更透彻地理
解本公幵， 并且能够将本公幵的范围完整的传达给本领域的技术人员。
[0037] 本发明公幵的多视点视频编码中的一种局部视差矢量获取方法， 主要解决当前 3D-HEVC中获取视差矢量的来源较少， 易由默认导出的零矢量代替视差矢量导 致不准确， 3D-AVS中利用全局信息导出视差矢量不够准确而提出的一种多视点 视频编码中的局部视差矢量获取方法。 该方法是主要将当前预测单元 PU相邻已 经编码的左边、 左上、 上边、 右上、 左下五个候选区域划分为若干个 lxw大小的 块， 按照一定的顺序考察当前候选区域所有划分的块中是否含有视差矢量来决 定是否考察下一个候选区域， 如果当前候选区域均没有可以使用的视差矢量则 扩大相邻区域得到新的候选区域， 直到当前候选区域所有划分的块中存在视差 矢量或候选区域到达设置的最大相邻区域为止， 最后利用当前候选区域 lxw块中 的所有视差矢量求均值获得当前预测单元的局部视差矢量达到视差矢量获取的 目的。
[0038] 在一个优选实施方案中， 本发明公幵多视点视频编码中的一种局部视差矢量的 导出方法， 包括：
[0039] 步骤一、 选择当前预测单元 PU相邻的左边、 左上、 上边和右上和左下的已经 编码的区域作为 5个候选区域， 分别记作 al、 b2、 bl、 b0和 a0， 如图 7所示， 区域 的大小具体分别取左边为 RxW， 上边为 RxH， 右边为 RxW; 其中 W和 H为当前 预测块 PU的宽和高； R为区域系数， 初始值取 1 ; 相邻区域的大小根据当前预测 块 PU的宽和高的不同而不同。
[0040] 步骤二、 将候选区域划分为 lxw大小的若干个单元， 1和 w—般取为 4;
[0041] 步骤三、 在这 5个候选区域中， 按照一定的顺序， 一个一个进行如下判断： 判 断当前候选区域的这些 lxw单元中的是否有可以使用的视差矢量， 如果有， 则将 所有视差矢量进行均值计算得到的视差矢量， 作为当前块的局部视差矢量， 并 停止寻找；
[0042] 步骤四、 若当前候选区域所有 lxw单元没有可以使用的视差矢量， 则考察下一 个候选区域；
[0043] 步骤五、 若当前 R下所有的候选区域都没有可以使用的视差矢量， 扩大区域系 数 R， 即将区域系数增加 1， 得到新的候选区域；
[0044] 步骤六、 循环步骤三、 四、 五， 直到导出当前块的视差矢量或者相邻区域系数 达到最大值 R M停止；
[0045] 步骤七、 根据步骤三、 四、 五、 六得到的当前编码预测块的视差矢量， 可以定 位到其在视间参考帧中的对应块， 根据对应块的运动矢量来预测当前块的运动 矢量；
[0046] 步骤八、 当前块的邻近块为运动补偿预测 MCP的块或帧内预测 Mm块吋， 利用 步骤三、 四、 五、 六得到的视差矢量作为视差运动补偿过程中的视差矢量预测
[0047] 即完成了多视点视频编码中的一种局部视差矢量的导出方案。
[0048] 为了进一步描述本公幵， 下面进一步描述了示例性的实施方式。
[0049] 具体实施方式一：
[0050] 本实施方式的一种多视点视频编码中的视差矢量的导出方式， 具体是按照以下 步骤获得的：
[0051] 步骤一、 选择当前预测单元 PU相邻的左边、 左上、 上边和右上和左下的已经 编码的区域作为 5个候选区域， 分别记作 al、 b2、 bl、 b0和 a0， 如图 7所示， 区域 的大小具体分别取左边为 RxW， 上边为 RxH， 右边为 RxW; 其中 W和 H为当前预 测块 PU的宽和高； R为区域系数， 初始值取 1， R M本实施方案取为 4; 相邻区域 的大小根据当前预测块 PU的宽和高的不同而不同。
[0052] 步骤二、 将候选区域划分为 4x4大小的若干个单元；
[0053] 步骤三、 在这 5个候选区域中， 按照左边 al、 上边 bl、 左上 b2、 右上 b0、 左下 a 0的顺序， 一个一个进行如下判断： 判断当前候选区域的这些 4x4单元中的是否 有可以使用的视差矢量， 如果有， 则将所有视差矢量按照公式 （1.1) 进行均值 计算得到的视差矢量， 作为当前块的局部视差矢量， 并停止寻找；
[0054] 式中 Ω κ表示在给定 R下的候选区域中所有 DV的集合， N R表示 Ω κ这个集合的
大小。 优选地， 在具体实施的过程中， 为了方便硬件的设计， 固定最大操作数
， 当前区域N R的最大数目取 256， 如果超过 256， 则停止寻找按照 256计算。
[0055] 步骤四、 若当前候选区域所有 4x4单元没有可以使用的视差矢量， 则考察下一 个候选区域；
[0056] 步骤五、 若当前 R下所有的候选区域都没有可以使用的视差矢量， 扩大区域系 数 R， 即将区域系数增加 1， 得到新的候选区域；
[0057] 步骤六、 循环步骤三、 四、 五， 直到导出当前块的视差矢量或者相邻区域系数 达到最大值 R M=4停止。
[0058] 本实施方式效果：
[0059] 本实施方式利用导出的视差矢量定位到当前块在视间参考帧中的对应块， 根据 对应块的运动矢量来预测当前块的运动矢量。 并且当当前块的邻近块都不是视 差补偿预测的块吋； 利用导出的视差矢量作为视差补偿预测过程中的视差矢量 的预测。 因此将导出的当前预测块的局部视差矢量作为视间参考帧对应的视差 信息， 可以用于视间运动预测、 视差补偿预测或任何其他需要为当前块在参考 视点中定位到对应块的工具； 使得本实施方式在没有增加多视点视频编码吋间 的基础上， 在低延迟配置条件下提升了多视点视频编码视点 1.99%的 BD-mte值， 提升了合成视点的 1.32%的 BD-mte值。
[0060] 具体实施方式二：
[0061] 本实施方式的一种多视点视频编码中的视差矢量的导出方式， 具体是按照以下 步骤获得的：
[0062] 步骤一、 选择当前预测单元 PU相邻的左边、 左上、 上边和右上和左下的已经 编码的区域作为 5个候选区域， 分别记作 al、 b2、 bl、 b0和 a0， 如图 7所示， 区域 的大小具体分别取左边为 RxW， 上边为 RxH， 右边为 RxW; 其中 W和 H为当前预 测块 PU的宽和高； R为区域系数， 初始值取 1， R M本实施方案取为 4; 相邻区域 的大小根据当前预测块 PU的宽和高的不同而不同。
[0063] 步骤二、 将候选区域划分为 4x4大小的若干个单元；
[0064] 步骤三、 在这 5个候选区域中， 按照左边 al、 上边 bl、 右上 b0、 左下 a0、 左上 b
2的顺序， 一个一个进行如下判断： 判断当前候选区域的这些 4x4单元中的是否
有可以使用的视差矢量， 如果有， 则将所有视差矢量按照具体实施方式一中的 公式 （1.1) 进行均值计算得到的视差矢量， 作为当前块的局部视差矢量， 并停 止寻找；
[0065] 步骤四、 若当前候选区域所有 4x4单元没有可以使用的视差矢量， 则考察下一 个候选区域；
[0066] 步骤五、 若当前 R下所有的候选区域都没有可以使用的视差矢量， 扩大区域系 数 R， 即将区域系数增加 1， 得到新的候选区域；
[0067] 步骤六、 循环步骤三、 四、 五， 直到导出当前块的视差矢量或者相邻区域系数 达到最大值 R M=4停止。
[0068] 本实施方式效果：
[0069] 本实施方式与具体实施方式一相同， 将导出的当前预测块的局部视差矢量作为 视间参考帧对应的视差信息， 可以用于视间运动预测、 视差补偿预测或任何其 他需要为当前块在参考视点中定位到对应块的工具； 使得本实施方式在没有增 加多视点视频编码吋间的基础上， 在低延迟配置条件下提升了多视点视频编码 视点 1.96%的 BD-rate值， 提升了合成视点的 1.23%的 BD-rate值。
[0070] 具体实施方式三：
[0071] 本实施方式的一种多视点视频编码中的视差矢量的导出方式， 具体是按照以下 步骤获得的：
[0072] 步骤一、 选择当前预测单元 PU相邻的左边、 左上、 上边和右上和左下的已经 编码的区域作为 5个候选区域， 分别记作 al、 b2、 bl、 b0和 a0， 如图 7所示， 区域 的大小具体分别取左边为 RxW， 上边为 RxH， 右边为 RxW; 其中 W和 H为当前预 测块 PU的宽和高； R为区域系数， 初始值取 1， R M本实施方案取为 4; 相邻区域 的大小根据当前预测块 PU的宽和高的不同而不同。
[0073] 步骤二、 将候选区域划分为 4x4大小的若干个单元；
[0074] 步骤三、 在这 5个候选区域中， 按照左边 al、 上边 bl、 左上 b2、 左下 a0、 右上 b 0的顺序， 一个一个进行如下判断： 判断当前候选区域的这些 4x4单元中的是否 有可以使用的视差矢量， 如果有， 则将所有视差矢量按照具体实施方式一中的 公式 （1.1) 进行均值计算得到的视差矢量， 作为当前块的局部视差矢量， 并停
止寻找；
[0075] 步骤四、 若当前候选区域所有 4x4单元没有可以使用的视差矢量， 则考察下一 个候选区域；
[0076] 步骤五、 若当前 R下所有的候选区域都没有可以使用的视差矢量， 扩大区域系 数 R， 即将区域系数增加 1， 得到新的候选区域；
[0077] 步骤六、 循环步骤三、 四、 五， 直到导出当前块的视差矢量或者相邻区域系数 达到最大值 R M=4停止。
[0078] 本实施方式效果：
[0079] 本实施方式与具体实施方式一相同， 将导出的当前预测块的局部视差矢量作为 视间参考帧对应的视差信息， 可以用于视间运动预测、 视差补偿预测或任何其 他需要为当前块在参考视点中定位到对应块的工具； 使得本实施方式在没有增 加多视点视频编码吋间的基础上， 在低延迟配置条件下提升了多视点视频编码 视点 2.01 <¾的 BD-rate值， 提升了合成视点的 1.28%的 BD-rate值。
[0080] 以上所述， 仅为本发明较佳的具体实施方式， 但本发明的保护范围并不局限于 此， 任何熟悉本技术领域的技术人员在本发明揭露的技术范围内， 可轻易想到 的变化或替换， 都应涵盖在本发明的保护范围之内。 因此， 本发明的保护范围 应所述以权利要求的保护范围为准。
Claims
[权利要求 1] 一种局部视差矢量的导出方法， 其特征在于多视点视频编码标准中的 一种局部视差矢量的导出方法， 具体是按照以下步骤进行的：
步骤一、 选择当前预测单元 PU相邻的左边、 左上、 上边和右上和 左下的已经编码的区域作为 5个候选区域， 分别记作 al、 b2、 bl、 b0 和 a0， 区域的大小具体分别取左边为 RxW， 上边为 RxH， 右边为 Rx W; 其中 W和 H为当前预测块 PU的宽和高； R为区域系数， 初始值取 1 步骤二、 将候选区域划分为 lxw大小的若干个单元， 1和 w—般取 为 4;
步骤三、 在这 5个候选区域中， 按照一定的顺序， 一个一个进行如 下判断： 判断当前候选区域的这些 lxw单元中的是否有可以使用的视 差矢量， 如果有， 则将所有视差矢量进行均值计算得到的视差矢量， 作为当前块的局部视差矢量；
步骤四、 根据步骤三得到的当前编码预测块的视差矢量， 可以定 位到其在视间参考帧中的对应块， 根据对应块的运动矢量来预测当前 块的运动矢量；
步骤五、 当前块的邻近块为运动补偿预测 MCP的块或帧内预测 Intr a块吋， 利用步骤三得到的视差矢量作为视差运动补偿过程中的视差 矢量预测；
[权利要求 2] 根据权利要求 1所述的方法， 其特征在于:步骤三中若当前候选区域所 有 lxw单元没有可以使用的视差矢量， 则其具体的实现过程如下： (1)考察下一个候选区域；
(2)若当前 R下所有的候选区域都没有可以使用的视差矢量， 扩大 区域系数 R， 即将区域系数增加 1， 得到新的候选区域； 按照既定的 顺序判断这些候选区域是否有可以使用的视差矢量： 如果有， 则将当 前候选区域的所有可以使用的视差矢量进行均值计算导出当前块的视 差矢量； 如果没有， 则继续扩大区域系数 R直到导出视差矢量或者相
邻区域系数 R达到最大值 R M停止。
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710086852.0 | 2017-02-17 | ||
| CN201710086852.0A CN106803963B (zh) | 2017-02-17 | 2017-02-17 | 一种局部视差矢量的导出方法 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2018149092A1 true WO2018149092A1 (zh) | 2018-08-23 |
Family
ID=58988552
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/CN2017/093881 Ceased WO2018149092A1 (zh) | 2017-02-17 | 2017-07-21 | 一种局部视差矢量的导出方法 |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN106803963B (zh) |
| WO (1) | WO2018149092A1 (zh) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106803963B (zh) * | 2017-02-17 | 2019-07-02 | 北京大学 | 一种局部视差矢量的导出方法 |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140241430A1 (en) * | 2013-02-26 | 2014-08-28 | Qualcomm Incorporated | Neighboring block disparity vector derivation in 3d video coding |
| CN104394417A (zh) * | 2014-12-15 | 2015-03-04 | 哈尔滨工业大学 | 一种多视点视频编码中的视差矢量获取方法 |
| CN106803963A (zh) * | 2017-02-17 | 2017-06-06 | 北京大学 | 一种局部视差矢量的导出方法 |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120189060A1 (en) * | 2011-01-20 | 2012-07-26 | Industry-Academic Cooperation Foundation, Yonsei University | Apparatus and method for encoding and decoding motion information and disparity information |
| US9350970B2 (en) * | 2012-12-14 | 2016-05-24 | Qualcomm Incorporated | Disparity vector derivation |
| US9516332B2 (en) * | 2012-12-21 | 2016-12-06 | Qualcomm Incorporated | Disparity vector derivation in three-dimensional video coding |
-
2017
- 2017-02-17 CN CN201710086852.0A patent/CN106803963B/zh active Active
- 2017-07-21 WO PCT/CN2017/093881 patent/WO2018149092A1/zh not_active Ceased
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140241430A1 (en) * | 2013-02-26 | 2014-08-28 | Qualcomm Incorporated | Neighboring block disparity vector derivation in 3d video coding |
| CN104394417A (zh) * | 2014-12-15 | 2015-03-04 | 哈尔滨工业大学 | 一种多视点视频编码中的视差矢量获取方法 |
| CN106803963A (zh) * | 2017-02-17 | 2017-06-06 | 北京大学 | 一种局部视差矢量的导出方法 |
Non-Patent Citations (3)
| Title |
|---|
| JAEWON SUNG: "3D-CE5.h: Simplification of disparity vector derivation for HEVC-based 3D video coding", JOINT COLLABORATIVE TEAM ON 3D VIDEO CODING EXTENSION DEVELOPMENT OF ITU-T SG 16 WP 3 AND ISO/IEC JTC 1/SC 29/WG 11, 20 July 2012 (2012-07-20) * |
| JEWON KANG: "3D-CE5.h related: Improvements for disparity vector derivation", JOINT COLLABORATIVE TEAM ON 3D VIDEO CODING EXTENSION DEVELOPMENT OF ITU-T SG 16 WP 3 AND ISO/IEC JTC 1/SC 29/WG 11, 19 October 2012 (2012-10-19) * |
| YANG, QIUXIANG ET AL.: "Improved NBDV Derivation Method in 3D-HEVC", COMPUTER ENGINEERING AND DESIGN, vol. 37, no. 9, 30 September 2016 (2016-09-30) * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106803963A (zh) | 2017-06-06 |
| CN106803963B (zh) | 2019-07-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| KR101753171B1 (ko) | 3ｄ 비디오 코딩에서의 간략화된 뷰 합성 예측 방법 | |
| CN104718760B (zh) | 用于三维和多视图视频编码的方法和装置 | |
| JP5970609B2 (ja) | ３ｄビデオ符号化における統一された視差ベクトル導出の方法と装置 | |
| KR101638752B1 (ko) | 3ｄ 비디오 코딩에서 변이 벡터 도출을 제한하는 방법 | |
| JP6000463B2 (ja) | ３ｄ映像符号化の仮想深度値の方法および装置 | |
| CN105453561B (zh) | 三维以及多视图视频编码中导出默认视差向量的方法 | |
| AU2013284038B2 (en) | Method and apparatus of disparity vector derivation in 3D video coding | |
| JP6042556B2 (ja) | ３ｄビデオ符号化における制約される視差ベクトル導出の方法と装置 | |
| JP2016513925A (ja) | ３ｄビデオ符号化におけるビュー合成予測の方法と装置 | |
| Zhang et al. | A fast depth-map wedgelet partitioning scheme for intra prediction in 3D video coding | |
| US20150365649A1 (en) | Method and Apparatus of Disparity Vector Derivation in 3D Video Coding | |
| CN101895749B (zh) | 一种快速视差估计和运动估计方法 | |
| WO2015006984A1 (en) | Reference view selection for 3d video coding | |
| CN100571401C (zh) | 确定参考帧的方法及其装置 | |
| WO2018149092A1 (zh) | 一种局部视差矢量的导出方法 | |
| Zhang et al. | Improved disparity vector derivation in 3D-HEVC | |
| CN104394417B (zh) | 一种多视点视频编码中的视差矢量获取方法 | |
| CN105247862A (zh) | 三维视频编码中的视点合成预测的方法及装置 | |
| CN105144714B (zh) | 三维或多视图视频编码或解码的方法及装置 | |
| CN105432084A (zh) | 用于三维视频编码的参考视图选择的方法 | |
| Zhao et al. | Synthesized disparity vectors for 3D 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: 17896998 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: 17896998 Country of ref document: EP Kind code of ref document: A1 |
|
| 32PN | Ep: public notification in the ep bulletin as address of the adressee cannot be established |
Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 03.02.2020) |
|
| 122 | Ep: pct application non-entry in european phase |
Ref document number: 17896998 Country of ref document: EP Kind code of ref document: A1 |