Method and apparatus of directional intra prediction

Patent metadata

Field	Value
Publication number	US-10306222-B2
Application number	US-201214123095-A
Country	US
Kind code	B2
Filing date	Jun 20, 2012
Priority date	Jun 20, 2011
Publication date	May 28, 2019
Grant date	May 28, 2019

Abstract

A method and apparatus for intra prediction for a current block using directional intra prediction based on neighboring pixels around the current block are disclosed. Embodiment according to the present invention derives the gradient for the current pixel from the neighboring reconstructed pixels along a prediction direction of directional intra prediction. The gradient derived is used to form the final intra prediction. In another embodiment of the present invention, the intra prediction for slant vertical and horizontal directions is derived based on the neighboring reconstructed pixels in a neighboring row adjacent to the top side of the current block and the neighboring reconstructed pixels in a neighboring column adjacent to the left side of the current block.

First claim

The invention claimed is:

1. A method for block-based intra prediction for a current block using directional intra prediction based on neighboring pixels around the current block, the method comprising:
receiving neighboring reconstructed pixels corresponding to a neighboring region of the current block, wherein one or more reference pixels are selected among the neighboring reconstructed pixels in at least one of: a neighboring-row adjacent to a top side of the current block, and a neighboring-column adjacent to a left side of the current block;
deriving a gradient for each current pixel in the current block from the one or more reference pixels along a prediction direction of directional intra prediction;
providing an intra predictor for each current pixel, wherein the intra predictor comprises the gradient; and
encoding or decoding the current block based on the intra predictors for the current pixels;
wherein the intra predictor for each current pixel is derived by a sum of a first part corresponding to a first intra predictor based on non-gradient block-based intra prediction along the prediction direction and a second part corresponding to the gradient, the first intra predictor not based on a gradient-based determination, wherein the second part corresponding to the gradient is weighted by a weighting factor selected according to the prediction direction and one of a column position or a row position of the current pixel, wherein the weighting factor has a value between 0 and 1.

2. The method for claim 1 , wherein the prediction direction corresponds to vertical direction and wherein the reference pixels are selected from the neighboring reconstructed pixels in the neighboring-column adjacent to the left side of the current block.

3. The method for claim 2 , wherein the reference pixels comprise a first neighboring reconstructed pixel and a second neighboring reconstructed pixel, and wherein the first neighboring reconstructed pixel corresponds to an above-left pixel in the neighboring-column of the current block and the second neighboring reconstructed pixel corresponds to a pixel of the neighboring-column in a same row as the current pixel.

4. The method for claim 3 , wherein the non-gradient intra prediction is based on a reference pixel of the neighboring-row adjacent to the top side of the current block, wherein the reference pixel corresponds to one of the neighboring reconstructed pixels that is in a same column as the current pixel.

5. The method for claim 1 , wherein the prediction direction corresponds to horizontal direction and wherein the reference pixels are selected from the neighboring reconstructed pixels in the neighboring-row adjacent to the top side of the current block.

6. The method for claim 5 , wherein the reference pixels comprise a first neighboring reconstructed pixel and a second neighboring reconstructed pixel, and wherein the first neighboring reconstructed pixel corresponds to an above-left pixel in the neighboring-row of the current block and the second neighboring reconstructed pixel corresponds to a pixel of the neighboring-row in a same column as the current pixel.

7. The method for claim 6 , wherein the non-gradient intra prediction is based on a reference pixel of the neighboring-column adjacent to the left side of the current block, and wherein the reference pixel corresponds to one of the neighboring reconstructed pixels in a same row as the current pixel.

8. The method for claim 1 , wherein the prediction direction corresponds to a slant vertical direction, wherein the reference pixels comprise a first reference pixel in the neighboring-column adjacent to the left side of the current block and a second reference pixel in the neighboring-row adjacent to the top side of the current block, wherein the first reference pixel corresponds to one of the neighboring reconstructed pixels of the neighboring-column that is in a same row as the current pixel, and wherein the second reference pixel is associated with the first reference pixel along the slant vertical direction.

9. The method for claim 8 , wherein the second reference pixel is interpolated from the neighboring reconstructed pixels at integer positions in the neighboring-row according to the slant vertical direction if the second reference pixel is located at a non-integer position.

10. The method for claim 9 , wherein the non-gradient intra prediction is derived from the neighboring reconstructed pixels of the neighboring-row according to the slant vertical direction.

11. The method for claim 1 , wherein the prediction direction corresponds to a slant horizontal direction, wherein the reference pixels comprise a first reference pixel in the neighboring-row adjacent to the top side of the current block and a second reference pixel in the neighboring-column adjacent to the left side of the current block, wherein the first reference pixel corresponds to one of the neighboring reconstructed pixels of the neighboring-row that is in a same column as the current pixel, and wherein the second reference pixel is associated with the first reference pixel along the slant horizontal direction.

12. The method for claim 11 , wherein the second reference pixel is interpolated from the neighboring reconstructed pixels at integer positions of the neighboring column according to the slant horizontal direction if the second reference pixel is located at a non-integer position.

13. The method for claim 12 , wherein the non-gradient intra prediction is derived from the neighboring reconstructed pixels of the neighboring-column according to the slant horizontal direction.

14. An apparatus for block-based intra prediction for a current block using directional intra prediction based on neighboring pixels around the current block, the apparatus comprising at least one circuit configured to:
receive neighboring reconstructed pixels corresponding to a neighboring region of the current block, wherein one or more reference pixels are selected among the neighboring reconstructed pixels in at least one of: a neighboring-row adjacent to a top side of the current block, and a neighboring-column adjacent to a left side of the current block;
derive a gradient for each current pixel in the current block from the reference pixels along a prediction direction of directional intra prediction;
provide an intra predictor for each current pixel, wherein the intra predictor comprises the gradient; and
encoding or decoding the current block based on the intra predictors for the current pixels;
wherein the intra predictor for each current pixel is derived by a sum of a first part corresponding to a first intra predictor based on non-gradient block-based intra prediction along the prediction direction and a second part corresponding to the gradient, the first intra predictor not based on a gradient-based determination, wherein the second part corresponding to the gradient is weighted by a weighting factor selected according to the prediction direction and one of a column position or a row position of the current pixel, wherein the weighting factor has a value between 0 and 1.

15. The apparatus for claim 14 , wherein the prediction direction corresponds to vertical direction and wherein the reference pixels are selected from the neighboring reconstructed pixels in the neighboring-column adjacent to the left side of the current block.

16. The apparatus for claim 14 , wherein the prediction direction corresponds to horizontal direction and wherein the reference pixels are selected from the neighboring reconstructed pixels in the neighboring-row adjacent t

Assignees

• Guo Mei
• Zhao Xin
• Guo Xun
• Lei Shaw Min
• Hfi Innovation Inc

Inventors

• Guo Mei
• Zhao Xin
• Guo Xun
• Lei Shaw-Min

Classifications

• H04N19/105 Primary

  Selection of the reference unit for prediction within a chosen coding or prediction mode, e.g. adaptive choice of position and number of pixels used for prediction · CPC title

• H04N19/593

  involving spatial prediction techniques · CPC title

• H04N19/14

  Coding unit complexity, e.g. amount of activity or edge presence estimation (H04N19/146 takes precedence) · CPC title

• H04N19/176

  the region being a block, e.g. a macroblock · CPC title

• H04N19/157

  Assigned coding mode, i.e. the coding mode being predefined or preselected to be further used for selection of another element or parameter · CPC title