WO2014023024A1

WO2014023024A1 - Methods for disparity vector derivation

Info

Publication number: WO2014023024A1
Application number: PCT/CN2012/079962
Authority: WO
Inventors: Jicheng An; Liang Zhao; Kai Zhang; Yi-Wen Chen; Jian-Liang Lin
Original assignee: Mediatek Singapore Pte. Ltd.
Priority date: 2012-08-10
Filing date: 2012-08-10
Publication date: 2014-02-13

Abstract

Methods to manage multiple component carriers (CCs) efficiently in a mobile network with carrier aggregation (CA) enabled are proposed. For CC activation/deactivation, a single LCID value is used to represent both activation and deactivation command. A single command with multiple instructions is provided to activate and/or deactivate multiple CCs. In addition, unnecessary re-activation or re-inactivation of a CC is prevented, and explicit feedback for activation/deactivation is considered. For scheduling mechanism, a novel buffer status reporting (BSR) procedure is provided, where only one BSR is calculated after preparing all the transport blocks (TB) within one transmission time interval (TTI). Novel power headroom reporting (PHR) format and trigger are also provided. For DL-UL linking, various linking types are created based on whether there is carrier indicator field (CIF) in DL grant or UL grant. The various linking types are used in different applications to improve scheduling flexibility and load balancing.

Description

METHODS FOR DISPARITY VECTOR DERIVATION

FIELD OF INVENTION

The invention relates generally to Three-Dimensional (3D) video processing. In particular, the present invention relates to methods for disparity vector derivation in 3D video coding.

BACKGROUND OF THE INVENTION

3D video coding is developed for encoding or decoding video data of multiple views simultaneously captured by several cameras. Since all cameras capture the same scene from different viewpoints, multi-view video data contains a large amount of inter-view redundancy. To exploit the inter-view redundancy, additional tools which employ disparity vectors have been integrated to conventional 3D-HEVC (High Efficiency Video Coding) or 3D-AVC (Advanced Video Coding) codec as follows.

Disparity-compensated prediction

To share the previously encoded texture information of reference views, the well-known concept of disparity-compensated prediction (DCP) has been added as an alternative to motion- compensated prediction (MCP). MCP refers to an inter-picture prediction that uses already coded pictures of the same view in a different access unit, while DCP refers to an inter-picture prediction that uses already coded pictures of other views in the same access unit, as illustrated in Fig. 1. The vector used for DCP is termed disparity vector (DV), which is analog to the motion vector (MV) used in MCP.

Inter- view motion prediction

To share the previously encoded motion information of reference views, the inter-view motion prediction is employed. For deriving candidate motion parameters for a current block in a dependent view, a DV for current block is firstly derived, and then the prediction block in the already coded picture in the reference view is located by adding the DV to the location of current block. If the prediction block is coded using MCP, the associated motion parameters can be used as candidate motion parameters for the current block in the current view. The derived DV can also be directly used as a candidate DV for DCP.

Inter-view residual prediction

To share the previously encoded residual information of reference views, the residual signal for current block can be predicted by the residual signal of the corresponding blocks in reference views. The corresponding block in reference view is located by a DV. As described above, the DV is critical in 3D video coding for inter-view motion prediction, inter-view residual prediction, disparity-compensated prediction (DCP) or any other tools which need to indicate the correspondence between inter- view pictures.

Disparity derivation in current 3DV-HTM

In the current 3DV-HTM, according to JCT2-A0097 [1] Li Zhang, Ying Chen, Marta

Karczewicz, "CE5.h: Disparity vector generation results," JCT2-A0097, July 2012; and JCT2- A0126 [2] Jaewon Sung, Moonmo Koo, Sehoon Yea, "3D-CE5.h: Simplification of disparity vector derivation for HEVC-based 3D video coding," JCT2-A0126, July 2012, the DV is derived from spatial or temporal neighboring blocks. First, several spatial and temporal neighboring blocks are defined. In addition, each block is checked in a given order and once any block is identified as having a DV, the checking process will be terminated. The spatial neighboring blocks, shown in Fig. 2(a), are scanned in the following order: Al, Bl, B0, AO, B2. The temporal neighboring blocks, shown in Fig. 2(b), are scanned in following order: RB, Center.

If any DCP coded block is not found in the above mentioned spatial and temporal neighbour blocks, then the disparity information obtained from DV-MCP blocks are used; Fig. 3 shows an example of the DV-MCP block whose motion is predicted from a corresponding block in the inter-view reference picture where the location of the corresponding blocks is specified by a disparity vector. The disparity vector used in the DV-MCP block represents a motion correspondence between the current and inter- view reference picture.

To indicate whether a MCP block is DV-MCP coded or not and to save the disparity vector used for the inter-view motion parameters prediction, two variables are added to store the motion vector information of each block: dvMcpFlag and dvMcpDisparity. When dvMcpFlag is equal to 1, dvMcpDisparity is set to the disparity vector used for the inter- view motion parameter prediction. In the AMVP and merge candidate list construction process, dvMcpFlag of the candidate is set to 1 only for the candidate generated by inter-view motion parameter prediction and 0 for the others. The spatial and temporal neighboring blocks are searched in the following order: AO, Al, B0, Bl, B2, Col. The first block that has dvMcpFlag equal to 1 will be selected and its dvMcpDisparity will used as derived DV for the current block.

The derived DV of the current block will be used for inter-view motion prediction for

AMVP and merge mode and the inter-view residual prediction. In AMVP and merge mode, if the reference picture is inter- view reference picture, the DV is directly used as DV predictor for DCP. If the reference picture is temporal reference picture, the DV is used to locate the prediction block in the reference view and the motion parameter of the prediction block will be used as a candidate motion parameter of the current block. In the inter- view residual prediction, the DV is also used to locate the prediction block in the reference view and residual data of the prediction block will be used for the prediction of residual data of current block.

However, there is one bug in the DV derivation of the current HTM software. When the DV of a neighboring block is used in AMVP and merge mode and inter-view residual prediction for the current block, it is not considered whether the reference view of this DV is equal to the target reference view of the current block. In AMVP and merge mode, when the reference picture of the current block is inter- view reference picture, the target reference view is the view to which the inter- view reference picture belongs; when the reference picture of the current block is temporal reference picture, the target reference view is used to derive the prediction block and its associated motion parameter. In the inter- view residual prediction, the target reference view is also used to derive the prediction block and its residual data. Therefore, when the reference view of DV is not equal to the target reference view, it is obviously unreasonable to use the DV directly.

SUMMARY OF THE INVENTION

In light of the previously described problems, a disparity vector is derived for multi-view video coding and 3D video coding, which can be used for indicating the prediction block in reference view for inter-view motion prediction in AMVP and merge mode, indicating the prediction block in reference view for inter-view residual prediction, predicting the DV of a DCP block in AMVP and merge mode, or indicating the corresponding block in the inter-view picture for any other tools. The DV can be derived using the spatial neighboring blocks and temporal nighboring blocks as proposed in [1] Li Zhang, Ying Chen, Marta Karczewicz, "CE5.h: Disparity vector generation results," JCT2-A0097, July 2012; or [2] Jaewon Sung, Moonmo Koo, Sehoon Yea, "3D-CE5.h: Simplification of disparity vector derivation for HEVC-based 3D video coding," JCT2-A0126, July 2012; no matter the reference view of the DV. When the derived DV is used and if the reference view of DV is not equal to a target reference view, the DV needs to be scaled to the target reference view. The DV can be derived using the spatial neighboring blocks and temporal neighboring blocks with an additional restriction that the reference view of DV must be equal to the target reference view, and the target reference view is given when the derived DV is used. If the DV cannot be found, the DV can be derived using the spatial neighboring blocks and temporal nighboring blocks no matter that reference view of the DV, and then DV should be scaled to the target reference view. The DV can be scaled using the difference of view order index (VOI) or the difference of view positions (VP) between the reference view of DV and the target reference view. The VOI and VP for each view can be signaled in a bitstream.

If the cameras are arranged in parallel in a horizontal line, the DV can be scaled only for horizontal component; if the cameras are arranged in parallel in a vertical line, the DV can be scaled only for vertical component; if the cameras are not arranged in a line, the DV can be scaled for both horizontal and vertical components.

Other aspects and features of the invention will become apparent to those with ordinary skill in the art upon review of the following descriptions of specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

The invention can be more fully understood by reading the subsequent detailed description and examples with references made to the accompanying drawings, wherein:

Fig. 1 is a diagram illustrating disparity-compensated prediction as an alternative to motion-compensated prediction according to an embodiment of the invention;

Fig. 2(a) and Fig. 2(b) are diagrams illustrating (a) Location of spatial neighboring blocks; and (b) Location of temporal neighboring blocks according to an embodiment of the invention;

Fig. 3 illustrates an exemplary DV-MCP block.

DETAILED DESCRIPTION

The following description is of the best-contemplated mode of carrying out the invention.

This description is made for the purpose of illustrating the general principles of the invention and should not be taken in a limiting sense. The scope of the invention is best determined by reference to the appended claims.

In this invention, we propose several methods to improve the DV derivation in the current HTM.

The first proposed method is applied according to the following steps:

First step, derive the DV of the current block according to the spatial or temporal neighbouring blocks as in reference [1] Li Zhang, Ying Chen, Marta Karczewicz, "CE5.h:

Disparity vector generation results," JCT2-A0097, July 2012; and [2] Jaewon Sung, Moonmo Koo, Sehoon Yea, "3D-CE5.h: Simplification of disparity vector derivation for HEVC-based

3D video coding," JCT2-A0126, July 2012.

Second step, in AMVP or merge mode or inter-view residual prediction, if the reference view of the derived DV is not equal to the target reference view, the DV will be scaled to the target reference view; else the DV will be used directly. The second proposed method is applied according to the following steps:

First step, given a target reference view in AMVP or merge mode or inter-view residual prediction, derive the DV of the current block according to the spatial or temporal neighboring blocks as in [1] Li Zhang, Ying Chen, Marta Karczewicz, "CE5.h: Disparity vector generation results," JCT2-A0097, July 2012; and [2] Jaewon Sung, Moonmo Koo, Sehoon Yea, "3D- CE5.h: Simplification of disparity vector derivation for HEVC-based 3D video coding," JCT2- A0126, July 2012, with an additional restriction that the reference view of the DV must be equal to the target reference view.

Second step, if the DV is not found in the first step, derive the DV as the original method without that restriction, and scale the DV to the target reference view.

In both the first and second proposed methods, the DV can be scaled using the difference of view order index (VOI) between the reference view of DV and target reference view, or scaled using the difference of view positions (VPs) between the reference view of DV and target reference view.

Cameras are arranged in a horizontal line

Assume that the cameras are arranged in parallel horizontally in a line. For ordering the multiple views, each view is associated with an identifier called view order index (VOI). The VOI is a signed integer values, which specifies the ordering of the coded views from left to right. If a view A has a smaller value of VOI than a view B, the camera for view A is located to the left of the camera of view B.

The view position (VP) represents the coordinate of camera/view in the camera line horizontally. For example, if view A, view B, and view C are located from left to right and the distance between view B and C are twice of distance between view A and B, then the difference of VP between view B and C should be twice of the difference of VP between view A and B.

The scaled DV (with two components SDV_X, SDV_Y) is derived according to original

DV (with two components DV_X, DV_Y) as follows:

SDV_X=Clip3(-32768,32767,(ScaleFactor *DV_X+127+(ScaleFactor*DV_X < 0)) » 8);

(1)

SDV_Y = DV_Y; (2)

The scaling factor is calculated using the following equations:

ScaleFactor = Clip3( -4096, 4095, ( tb * tx + 32 ) » 6 ) (3)

tx = ( 16384 + Abs( td / 2 ) ) / td (4) where td and tb can be derived as:

td = Clip3( -128, 127, CurrVOI - DVRefVOI ) (5)

tb = Clip3( -128, 127, CurrVOI - TargetRefVOI ) (6)

The variables in the above equations are specified as follows:

CurrVOI : The VOI of current view;

DVRefVOI: The VOI of reference view of DV;

TargetRefV 01: The VOI of target reference view.

The td and tb can also be derived as:

td = Clip3( -128, 127, CurrVP - DVRefYP ) (7)

tb = Clip3( -128, 127, CurrVP - TargetRefYP) (8)

The variables in the above equations are specified as follows:

CurrVP: The VP of current view;

DVRefYP: The VP of reference view of DV;

TargetRefYP: The VP of target reference view.

Cameras are arranged in a vertical line

If the cameras are arranged in parallel vertically in a line, the VOI specifies the ordering of coded views from top to bottom, and the VP also represents the coordinate of camera in a line vertically, and then the scaled DV can be derived as:

SDV_X = DV_X; (9)

SDV_Y= Clip3(-32768,32767, (ScaleFactor * DV_Y + 127 + (ScaleFactor *DV_Y < 0)) » 8 ); (10)

The scaling factor can also be derived as the above equations (3)-(8).

Cameras are not arranged in a line

If the cameras are not arranged in a line horizontally or vertically, the VOI and VP will have two components for horizontal and vertical directions respectively, and then the scaled DV can be derived as:

SDV_X = Clip3( -32768, 32767, (ScaleFactorX * DV_X + 127 + (ScaleFactorX *DV_X<0)) » 8); (11)

SDV_Y = Clip3( -32768, 32767, (ScaleFactorY * DV_Y + 127 + (ScaleFactorY*DV_Y<

0))» 8); (12)

The scaling factor in horizontal direction ScaleFactorX is derived according to the above equations (3)-(8) by replacing the VOI and VP as the horizontal component of VOI and VP respectively. The scaling factor in vertical direction ScaleFactorY is derived according to the above equations (3)-(8) by replacing the VOI and VP as the vertical component of VOI and VP respectively.

The disparity vector derived for multi-view video coding and 3D video coding can be used for indicating the prediction block in reference view for inter-view motion prediction in AMVP and merge mode, indicating the prediction block in reference view for inter-view residual prediction, predicting the DV of a DCP block in AMVP and merge mode, and indicating the corresponding block in the inter-view picture for any other tools. The DV can be derived using the spatial neighboring blocks and temporal nighboring blocks no matter the reference view of the DV. When the derived DV is used and if the reference view of DV is not equal to the target reference view, the DV needs to be scaled to the target reference view. The DV can be derived using the spatial neighboring blocks and temporal neighboring blocks with an additional restriction that the reference view of DV must be equal to the target reference view, and the target reference view is given when the derived DV is used. If the DV cannot be found, the DV can be derived using the spatial neighboring blocks and temporal nighboring blocks no matter that reference view of the DV, and then DV should be scaled to the target reference view. The DV can be scaled using the difference of view order index (VOI) or the difference of view positions (VP) between the reference view of DV and the target reference view. The VOI and VP for each view can be signaled in a bitstream.

Specifically, if the cameras are arranged in parallel in a horizontal line, the DV can only be scaled for horizontal component as shown in equation (1) and (2); if the cameras are arranged in parallel in a vertical line, the DV can only be scaled for vertical component as shown in equation (9) and (10); if the cameras are not arranged in a line, the DV can be scaled for both horizontal and vertical components as shown in equation (11) and (12). If the cameras are arranged in parallel in a horizontal or vertical line, the DV can also be scaled for both horizontal and vertical components. Specifically, if the cameras are arranged in parallel in a horizontal or vertical line, the DV can also be scaled for both horizontal and vertical components are shown in equation (11) and (12). The scaling factor in those equations (1),(2), (9), and (10) can be derived from the equation (3)-(8). The scaling factor in equation (11) is derived according to the equations (3)-(8) by replacing the VOI and VP as the horizontal component of VOI and VP respectively. The scaling factor in equation (12) is derived according to the equations (3)-(8) by replacing the VOI and VP as the vertical component of VOI and VP respectively.

The disparity vector derivation methods described above can be used in a video encoder as well as in a video decoder. Embodiments of disparity vector derivation methods according to the present invention as described above may be implemented in various hardware, software codes, or a combination of both. For example, an embodiment of the present invention can be a circuit integrated into a video compression chip or program codes integrated into video compression software to perform the processing described herein. An embodiment of the present invention may also be program codes to be executed on a Digital Signal Processor (DSP) to perform the processing described herein. The invention may also involve a number of functions to be performed by a computer processor, a digital signal processor, a microprocessor, or field programmable gate array (FPGA). These processors can be configured to perform particular tasks according to the invention, by executing machine-readable software code or firmware code that defines the particular methods embodied by the invention. The software code or firmware codes may be developed in different programming languages and different format or style. The software code may also be compiled for different target platform. However, different code formats, styles and languages of software codes and other means of configuring code to perform the tasks in accordance with the invention will not depart from the spirit and scope of the invention.

The invention may be embodied in other specific forms without departing from its spirit or essential characteristics. The described examples are to be considered in all respects only as illustrative and not restrictive. To the contrary, it is intended to cover various modifications and similar arrangements (as would be apparent to those skilled in the art). Therefore, the scope of the appended claims should be accorded the broadest interpretation so as to encompass all such modifications and similar arrangements.

Claims

1. A method of disparity vector derivation for multi-view video coding or 3D video coding, comprising:

deriving a disparity vector (DV) used for (a) indicating a prediction block in a reference view for inter- view motion prediction in AMVP (advance motion vector prediction) and merge mode; (b) indicating the prediction block in the reference view for inter-view residual prediction; (c) predicting the disparity vector (DV) of a DCP (disparity-compensated prediction) block in the AMVP and merge mode; or (d) indicating a corresponding block in an inter- view picture for another tool.

2. The method as claimed in claim 1, wherein the DV can be derived using spatial neighboring blocks and temporal neighboring blocks, no matter the reference view of the DV.

3. The method as claimed in claim 2, wherein when the derived DV is used, if the reference view of the DV is not equal to a target reference view, the DV needs to be scaled to the target reference view.

4. The method as claimed in claim 1, wherein the DV can be derived using spatial neighboring blocks and temporal neighboring blocks with an additional restriction that the reference view of the DV must be equal to a target reference view, and the target reference view is given when the derived DV is used.

5. The method as claimed in claim 4, wherein if the DV cannot be found, then the DV is derived using spatial neighboring blocks and temporal neighboring blocks no matter the reference view of the DV, and then DV is scaled to the target reference view.

6. The method as claimed in claim 5, wherein the DV is scaled using a difference of view order index (VOI) or a difference of view positions (VP) between the reference view of the DV and the target reference view.

7. The method as claimed in claim 6, wherein if cameras are arranged in parallel in a horizontal line, the DV is scaled only for horizontal component; if the cameras are arranged in parallel in a vertical line, the DV is scaled only for vertical component; if the cameras are not arranged in a line, the DV is scaled for both horizontal and vertical components.

8. The method as claimed in claim 7, wherein if the cameras are arranged in parallel in horizontal or vertical line, the DV is scaled for both horizontal and vertical components.

9. The method as claimed in claim 6, the VOI and VP for each view is signaled in bitstream.