公开(公告)号：US20170310988A1

公开(公告)日：2017-10-26

申请号：US15526083

申请日：2015-12-09

Applicant: MEDIATEK INC.

Inventor： Jian-Liang LIN ,  Yi-Wen CHEN

IPC: H04N19/513 ,  H04N19/56

CPC classification number: H04N19/513 ,  H04N19/52 ,  H04N19/56

Abstract: A method and apparatus for deriving directional-priority based candidates for a block coded in Inter, or Merge or Skip mode are disclosed. One or more motion vectors associated with one or more previously coded blocks for a current block are determined first. One or more directional-priority based candidates for the current block are derived by searching through the previously coded blocks according to a priority order associated with prediction direction of the motion vectors. The motion vectors having a first prediction direction are selected with a higher priority than the motion vectors having a second prediction direction. The derived directional-priority based candidates are inserted into a candidate list. The motion vector predictor (MVP) or Merge/Skip candidate is selected from the candidate list for coding the current block in Inter, or Merge or Skip mode.

公开(公告)号：US20160234510A1

公开(公告)日：2016-08-11

申请号：US15022001

申请日：2015-05-26

Applicant: MEDIATEK INC.

Inventor： Jian-Liang LIN ,  Yi-Wen CHEN ,  Xianguo ZHANG ,  Kai ZHANG ,  Jicheng AN ,  Han HUANG

IPC: H04N19/174 ,  H04N19/50 ,  H04N19/597

CPC classification number: H04N19/174 ,  H04N13/161 ,  H04N19/105 ,  H04N19/119 ,  H04N19/136 ,  H04N19/137 ,  H04N19/182 ,  H04N19/50 ,  H04N19/563 ,  H04N19/597

Abstract: A method of video coding using coding modes including depth-based block partitioning (DBBP) in a multi-view or three-dimensional (3D) video coding system is disclosed. According to the present invention, when DBBP (depth-based block partition) is used to code a current texture coding unit, the DBBP partition mode is signaled so that the decoder does not need to go through complex computations to derive the DBBP partition mode. Various examples of determining the DBBP partition mode are disclosed.

Abstract translation: 公开了一种使用包括在多视点或三维（3D）视频编码系统中的基于深度的块分割（DBBP））的编码模式的视频编码方法。 根据本发明，当使用DBBP（基于深度的块划分）来编码当前的纹理编码单元时，发信号通知DBBP分区模式，使得解码器不需要经过复杂的计算来导出DBBP分区模式。 公开了确定DBBP分区模式的各种示例。

公开(公告)号：US20160198139A1

公开(公告)日：2016-07-07

申请号：US14916462

申请日：2014-10-17

Applicant: MediaTek Inc.

Inventor： Jian-Liang LIN ,  Yi-Wen CHEN

IPC: H04N13/00 ,  G06K9/62 ,  H04N19/176 ,  H04N19/597 ,  H04N19/52

CPC classification number: H04N13/161 ,  G06K9/6201 ,  H04N19/176 ,  H04N19/52 ,  H04N19/597

Abstract: A method for three-dimensional video coding using aligned motion parameter derivation for motion information prediction and inheritance is disclosed. Embodiments according to the present invention utilize motion parameters associated with a corresponding block for motion information prediction or inheritance. The aligned motion parameters may be derived by searching each current reference picture list of the current block to find a matched reference picture having a same POC (Picture Order Count) or a same view index as that of the reference picture pointed by the MV of the corresponding block. The aligned motion parameters may also be derived by searching each current reference picture list to check whether the reference picture index of the reference picture in the reference view to be inherited exceeds a maximum reference picture index of each current reference picture list of the current block.

Abstract translation: 公开了一种使用对准运动参数推导运动信息预测和继承的三维视频编码方法。 根据本发明的实施例利用与相应块相关联的运动参数用于运动信息预测或继承。 对齐的运动参数可以通过搜索当前块的每个当前参考图片列表来找出具有相同POC（图片顺序计数）的相匹配的参考图片或者与由MV指定的参考图片相同的视图索引 相应的块。 对准的运动参数也可以通过搜索每个当前参考图片列表来导出，以检查参考视图中要继承的参考图片的参考图片索引是否超过当前块的每个当前参考图片列表的最大参考图片索引。

公开(公告)号：US20150382019A1

公开(公告)日：2015-12-31

申请号：US14762861

申请日：2014-04-02

Applicant: MEDIATEK INC.

Inventor： Yi-Wen CHEN ,  Jicheng AN ,  Jian-Liang LIN

IPC: H04N19/597 ,  H04N13/00 ,  H04N19/52 ,  H04N19/463 ,  H04N19/196

CPC classification number: H04N19/597 ,  H04N13/161 ,  H04N19/196 ,  H04N19/197 ,  H04N19/463 ,  H04N19/513 ,  H04N19/52

Abstract: A method and apparatus for a three-dimensional encoding or decoding system incorporating view synthesis prediction (VSP) with reduced computational complexity and/or memory access bandwidth are disclosed. The system applies the VSP process to the texture data only and applies non-VSP process to the depth data. Therefore, when a current texture block in a dependent view is coded according to VSP by backward warping the current texture block to the reference picture using an associated depth block and the motion parameter inheritance (MPI) mode is selected for the corresponding depth block in the dependent view, the corresponding depth block in the dependent view is encoded or decoded using non-VSP inter-view prediction based on motion information inherited from the current texture block.

Abstract translation: 公开了一种具有缩减计算复杂度和/或存储器访问带宽的视图合成预测（VSP）的三维编码或解码系统的方法和装置。 系统仅将VSP处理应用于纹理数据，并将非VSP处理应用于深度数据。 因此，当依赖视图中的当前纹理块通过使用相关联的深度块将当前纹理块向后弯曲到参考图片时根据VSP进行编码，并且为相应的深度块选择运动参数继承（MPI）模式 依赖视图中，依赖视图中的相应深度块基于从当前纹理块继承的运动信息使用非VSP视角间预测进行编码或解码。

公开(公告)号：US20150358598A1

公开(公告)日：2015-12-10

申请号：US14758413

申请日：2014-01-07

Applicant: MEDIATEK INC.

Inventor： Jian-Liang LIN ,  Yi-Wen CHEN

IPC: H04N13/00 ,  H04N19/105 ,  H04N19/597

CPC classification number: H04N13/161 ,  H04N19/105 ,  H04N19/52 ,  H04N19/597

Abstract: A method and apparatus using a single converted DV (disparity vector) from the depth data for a conversion region are disclosed. Embodiments according to the present invention receive input data and depth data associated with a conversion region of a current picture in a current dependent view. The conversion region is checked to determine whether it is partitioned into multiple motion prediction sub-blocks. If the conversion region is partitioned into multiple motion prediction sub-blocks, then a single converted DV from the depth data associated with the conversion region is determined and each of the multiple motion prediction sub-blocks of the conversion region is processed according to a first coding tool using the single converted DV. If the conversion region is not partitioned into multiple motion prediction sub-blocks, the conversion region is processed according to the first coding tool or a second coding tool using the single converted DV.

Abstract translation: 公开了从转换区域的深度数据中使用单个转换的DV（视差矢量）的方法和装置。 根据本发明的实施例在当前依赖视图中接收与当前画面的转换区域相关联的输入数据和深度数据。 检查转换区域以确定其是否被划分为多个运动预测子块。 如果将转换区域划分为多个运动预测子块，则确定与转换区域相关联的深度数据中的单个转换后的DV，并根据第一个处理转换区域的多个运动预测子块中的每一个 编码工具使用单转换DV。 如果转换区域未划分为多个运动预测子块，则根据第一编码工具或使用单个转换的DV的第二编码工具来处理转换区域。

公开(公告)号：US20150264399A1

公开(公告)日：2015-09-17

申请号：US14656993

申请日：2015-03-13

Applicant: MediaTek Inc.

Inventor： Jian-Liang LIN ,  Yi-Wen CHEN ,  Xianguo ZHANG ,  Kai ZHANG ,  Jicheng AN ,  Han HUANG

IPC: H04N19/597 ,  H04N13/00 ,  H04N19/17 ,  H04N19/186 ,  H04N19/159 ,  H04N19/105

CPC classification number: H04N19/597 ,  H04N13/161 ,  H04N19/119 ,  H04N19/159 ,  H04N19/176 ,  H04N19/70

Abstract: A method of signaling depth-based block partitioning (DBBP) for multi-view or three-dimensional (3D) video coding is disclosed. In one embodiment, the DBBP flag is signaled for all candidate prediction modes of the current texture coding unit including a non-2N×N partition mode. The group of candidate prediction modes may consist of 2N×N partition mode and N×2N partition mode. If the DBBP flag indicates the DBBP being used for the current texture coding unit, DBBP encoding is applied to the current texture coding unit or DBBP decoding is applied to one or more PUs associated with the current texture coding unit to recover the current texture coding unit. If the DBBP flag indicates the DBBP being not used for the current texture coding unit, a prediction partition mode is signaled at an encoder side or parsed at a decoder side from a CU (coding unit) level of the bitstream.

Abstract translation: 公开了用于多视图或三维（3D）视频编码的基于深度的块分割（DBBP）的方法。 在一个实施例中，为包括非2N×N分区模式的当前纹理编码单元的所有候选预测模式发信号通知DBBP标志。 候选预测模式组可以由2N×N分区模式和N×2N分区模式组成。 如果DBBP标志指示用于当前纹理编码单元的DBBP，则将DBBP编码应用于当前纹理编码单元，或者将DBBP解码应用于与当前纹理编码单元相关联的一个或多个PU，以恢复当前纹理编码单元 。 如果DBBP标志指示DBBP未被用于当前纹理编码单元，则在编码器侧发信号通知预测分区模式，或者在解码器侧从比特流的CU（编码单元）级别解析。

公开(公告)号：US20200322631A1

公开(公告)日：2020-10-08

申请号：US16910374

申请日：2020-06-24

Applicant: MEDIATEK INC.

Inventor： Cheng-Hsuan SHIH ,  Jian-Liang LIN

IPC: H04N19/597 ,  H04N19/176 ,  H04N19/119 ,  H04N19/96

Abstract: Methods and apparatus of processing 360-degree virtual reality (VR360) pictures are disclosed. According to one method, if a leaf coding unit contains one or more face edges, the leaf processing unit is split into sub-processing units along the face edges without the need to signal the partition. In another method, if the quadtree (QT) of binary tree (BT) partition depth for a processing unit has not reached the maximum QT or BT depth, the processing unit is split. If the processing unit contains a horizontal face edge, QT or horizontal BT partition is applied. If the processing unit contains a vertical face edge, QT or vertical BT partition is applied.

公开(公告)号：US20200077092A1

公开(公告)日：2020-03-05

申请号：US16557686

申请日：2019-08-30

Applicant: MEDIATEK INC.

Inventor： Sheng Yen LIN ,  Lin LIU ,  Jian-Liang LIN

IPC: H04N19/117 ,  H04N19/61 ,  H04N19/70 ,  H04N19/172

Abstract: Method and apparatus of coding pictures containing one or more virtual boundaries, such as 360-degree virtual reality (VR360) video are disclosed. According to this method, a reconstructed filtered unit associated with a loop filter for a current reconstructed pixel is received. The loop filtering process associated with the loop filter is applied to the current reconstructed pixel to generate a filtered reconstructed pixel, where if the loop filtering process for the current reconstructed pixel is across a virtual boundary of the picture, the loop filtering process is disabled when fixed-size loop filtering is used or a smaller-size loop filter is selected when adaptive-size loop filtering is used for the current reconstructed pixel, where the filtered reconstructed pixel is the same as the current reconstructed pixel when the loop filtering process is disabled. The filtered reconstructed pixel is the same as the current reconstructed pixel.

公开(公告)号：US20190289328A1

公开(公告)日：2019-09-19

申请号：US16299320

申请日：2019-03-12

Applicant: MEDIATEK INC.

Inventor： Cheng-Hsuan SHIH ,  Jian-Liang LIN

IPC: H04N19/597 ,  H04N19/96 ,  H04N19/119 ,  H04N19/176

Abstract: Methods and apparatus of processing 360-degree virtual reality (VR360) pictures are disclosed. According to one method, if a leaf coding unit contains one or more face edges, the leaf processing unit is split into sub-processing units along the face edges without the need to signal the partition. In another method, if the quadtree (QT) of binary tree (BT) partition depth for a processing unit has not reached the maximum QT or BT depth, the processing unit is split. If the processing unit contains a horizontal face edge, QT or horizontal BT partition is applied. If the processing unit contains a vertical face edge, QT or vertical BT partition is applied.

公开(公告)号：US20190289327A1

公开(公告)日：2019-09-19

申请号：US16286874

申请日：2019-02-27

Applicant: MEDIATEK INC.

Inventor： Sheng-Yen LIN ,  Jian-Liang LIN

IPC: H04N19/597 ,  G06N3/08 ,  H04N19/96 ,  H04N19/82 ,  H04N19/119

Abstract: Methods and apparatus of processing 360-degree virtual reality (VR360) pictures are disclosed. A target reconstructed VR picture in a reconstructed VR picture sequence is divided into multiple processing units and whether a target processing unit contains any discontinuous edge corresponding to a face boundary in the target reconstructed VR picture is determined. If the target processing unit contains any discontinuous edge: the target processing unit is split into two or more sub-processing units along the discontinuous edges; and NN processing is applied to each of the sub-processing units to generate a filtered processing unit. If the target processing unit contains no discontinuous edge, the NN processing is applied to the target processing unit to generate the filtered processing unit. A method and apparatus for CNN training process are also disclosed. The input reconstructed VR pictures and original pictures are divided into sub-frames along discontinuous boundaries for the training process.