Constraints on neighboring block based disparity vector (NBDV) techniques for 3D video

摘要

Techniques are described for determining whether a block in a candidate reference picture is available. A video coder may determine a location of a co-located largest coding unit (CLCU) in the candidate reference picture, where the CLCU is co-located with a LCU in a current picture, and the LCU includes a current block that is to be inter-predicted. The video coder may determine whether a block in the candidate reference picture is available based on a location of the block in the candidate reference picture relative to the location of the CLCU. If the block in the candidate reference picture is unavailable, the video coder may derive a disparity vector for the current block from a block other than the block determined to be unavailable.

主权项

1. A method for decoding video data, the method comprising:
determining a location of a co-located largest coding unit (CLCU) in a candidate reference picture, wherein the CLCU is co-located with a largest coding unit (LCU) that covers a current block in a current picture of the video data; determining a location of a block in the candidate reference picture relative to a location of the CLCU within the candidate reference picture; determining whether the block in the candidate reference picture is available for purposes of determining a disparity vector for the current block based on whether the block in the candidate reference picture is determined to be located inside or outside a bottom boundary of the CLCU within the candidate reference picture, wherein the block in the candidate reference picture is determined to be unavailable for purposes of determining the disparity vector for the current block based on the block in the candidate reference picture being located outside the bottom boundary of the CLCU; determining whether the block in the candidate reference picture is inter-predicted with a disparity motion vector for purposes of determining the disparity vector for the current block only when the block in the candidate reference picture is inside the bottom boundary of the CLCU, wherein the disparity motion vector refers to a reference picture used to inter-predict the block in the candidate reference picture and is in a view other than a view that includes the candidate reference picture; based on the block in the candidate reference picture being inter-predicted with the disparity motion vector, deriving the disparity vector for the current block from the disparity motion vector; and inter-prediction decoding the current block based in part on the disparity vector.