Method and apparatus for template-based intra prediction in image and video coding
A method and apparatus for video coding using template-based Intra prediction are disclosed. According to one method, the template-based Intra prediction searches N template matching candidates to reduce the complexity. In another method, the template-based Intra prediction assigns different weights for different pixel locations of template or uses different pixel precision for interpolation filter during cost evaluation among template matching candidates. In yet another method, truncated template size is used to reduce the complexity. In yet another method, the left and top templates may have different sizes for non-square block. In yet another method, inverse templated-based Intra prediction is used.
1. A method of video encoding and decoding used by a video encoding system and video decoding system respectively, the method comprising:
receiving input data associated with a current block in a current image, wherein template-based Intra prediction is enabled for the current image;
receiving a plurality of neighboring reconstructed samples of the current block;
determining one or more size-reduced templates from said plurality of neighboring reconstructed samples, wherein determining the one or more size-reduced templates comprises at least one of:
identifying, from said plurality of neighboring reconstructed samples, a first set of neighboring reconstructed samples as a top template located on a top side of the current block with a template width less than a width of the current block; and
identifying, from said plurality of neighboring reconstructed samples, a second set of neighboring reconstructed samples as a left template located on a left side of the current block with a template height less than a height of the current block;
determining a target Intra mode or an Intra-mode candidate set using the template-based Intra prediction according to said one or more size-reduced templates, wherein determining the target Intra mode or the Intra-mode candidate set comprises:
generating prediction samples for said one or more size-reduced templates according to a corresponding available Intra mode; and
determining whether the corresponding available Intra mode is selected as the target Intra mode or one of the Intra-mode candidate set based on a comparison between the prediction samples and neighboring reconstructed samples in said one or more size-reduced templates; and
encoding or decoding the current block using Intra prediction with a current Intra mode selected from an Intra mode group comprising the target Intra mode or the Intra-mode candidate set.
2. The method of claim 1 , wherein the template height of the left template is equal to a half of the height of the current block and the template width of the top template is equal to a half of the width of the current block.
3. The method of claim 1 , wherein the template height of the left template is equal to ¾ of the height of the current block and the template width of the top template is equal to ¾ of the width of the current block.
4. The method of claim 1 , wherein the template height of the left template is smaller than the height of the current block and the template width of the top template is smaller than the width of the current block.
5. An apparatus for video encoding and decoding used by a video encoding system and video decoding system respectively, the apparatus comprising one or more electronic circuits or processors arranged to:
receive input data associated with a current block in a current image, wherein template-based Intra prediction is enabled for the current image; and
receive a plurality of neighboring reconstructed samples of the current block;
if the template-based Intra prediction is selected for the current block:
determine one or more size-reduced templates from said plurality of neighboring reconstructed samples, wherein determining the one or more size-reduced templates comprises at least one of:
identifying, from said plurality of neighboring reconstructed samples, a first set of neighboring reconstructed samples as a top template located on a top side of the current block with a template width less than a width of the current block; and
identifying, from said plurality of neighboring reconstructed samples, a second set of neighboring reconstructed samples as a left template located on a left side of the current block with a template height less than a height of the current block;
determine a target Intra mode or an Intra-mode candidate set using the template-based Intra prediction according to said one or more size-reduced templates, wherein determining the target Intra mode or the Intra-mode candidate set comprises:
generating prediction samples for said one or more size-reduced templates according to a corresponding available Intra mode; and
determining whether the corresponding available Intra mode is selected as the target Intra mode or one of the Intra-mode candidate set based on a comparison between the prediction samples and neighboring reconstructed samples in said one or more size-reduced templates; and
encode or decode the current block using Intra prediction with a current Intra mode selected from an Intra mode group comprising the target Intra mode or the Intra-mode candidate set.