Motion vector processing

A non-transitory computer-readable medium comprising, stored thereupon, a computer program for predicting motion vectors for multi-view video coding of video content from multiple camera views, said computer program comprising code that, when run on a computer, causes said computer to:

• determine a motion vector type for a motion vector, from multiple predefined motion vector types, wherein said motion vector is estimated for a current pixel block in a current frame of a current view of said multiple camera views and said motion vector identifies a reference pixel area in a reference frame, wherein said motion vector type is determined based on at least one of i) said current camera view and a camera view of said reference frame among said multiple camera views, and ii) a point in time of said current frame and a point in time of said reference frame;
• identify at least one candidate motion vector predictor of said determined motion vector type;
• and
• determine a motion vector predictor for said motion vector from only among said at least one candidate motion vector predictor of said determined motion vector type.

A device for motion vector prediction for multi-view video coding of video content from multiple camera views, said device comprising:

• a type determiner configured to determine a motion vector type for a motion vector, from multiple predefined motion vector types, wherein said motion vector is estimated for a current pixel block in a current frame of a current camera view of said multiple camera views and said motion vector identifies a reference pixel area in a reference frame, wherein said type determiner is configured to determine said motion vector type based on at least one of i) said current camera view and a camera view of said reference frame among said multiple camera views, and ii) a point in time of said current frame and a point in time of said reference frame;
• a predictor identifier configured to identify at least one candidate motion vector predictor of said motion vector type determined by said type determiner;
• and
• a predictor determiner configured to determine a motion vector predictor for said motion vector from only among said at least one candidate motion vector predictor of said determined motion vector type identified by said predictor identifier.

A method of motion vector decoding for encoded multi-view video from multiple camera views, said method comprises:

• providing, for a current pixel block in a current frame of a current camera view of said multiple camera views, a reference frame index identifying a reference frame;
• determining a motion vector type for said current pixel block, from multiple predefined motion vector types and based on at least one of i) said current camera view and a camera view of said reference frame among said multiple camera views, and ii) a point in time of said current frame and a point in time of said reference frame;
• generating a candidate list comprising at least one candidate motion vector predictor of said determined motion vector type, wherein said candidate list comprises only candidate motion vector predictors of said determined motion vector type;
• and
• determining a motion vector for said current pixel block based on a candidate motion vector predictor of said candidate list.

A non-transitory computer-readable medium comprising, stored thereupon, a computer program for decoding motion vectors for encoded multi-view video from multiple camera views, said computer program comprising code that, when run on a computer, causes said computer to:

• provide, for a current pixel block in a current frame of a current camera view of said multiple camera views, a reference frame index identifying a reference frame;
• determine a motion vector type for said current pixel block, from multiple predefined motion vector types and based on at least one of i) said current camera view and a camera view of said reference frame among said multiple camera views, and ii) a point in time of said current frame and a point in time of said reference frame;
• generate a candidate list comprising at least one candidate motion vector predictor of said determined motion vector type, wherein said candidate list comprises only candidate motion vector predictors of said determined motion vector type;
• and
• determine a motion vector for said current pixel block based on a candidate motion vector predictor of said candidate list.

A device for motion vector decoding for encoded multi-view video from multiple camera views, said device comprising:

• an index provider configured to provide, for a current pixel block in a current frame of a current camera view of said multiple camera views, a reference frame index identifying a reference frame;
• a type determiner configured to determine a motion vector type for said current pixel block, from multiple predefined motion vector types and based on at least one of i) said current camera view and a camera view of said reference frame among said multiple camera views, and ii) a point in time of said current frame and a point in time of said reference frame;
• a list generator configured to generate a candidate list comprising at least one candidate motion vector predictor of said motion vector type determined by said type determiner, wherein said candidate list comprises only candidate motion vector predictors of said determined motion vector type;
• and
• a vector determiner configured to determine a motion vector for said current pixel block based on a candidate motion vector predictor of said candidate list.