Block size decision for video coding

1. A method for encoding video data, the method comprising:

determining, by a video encoder, a respective variance value for pixels of each respective sub-block of at least four sub-blocks of a current block of video data of a particular block size of a plurality of block sizes, wherein the current block of video data is included in a current picture of video data that includes a plurality of other blocks of video data;

determining, by the video encoder, an average of the determined variance values for the at least four sub-blocks;

determining, by the video encoder, a maximum difference value for the current block as a largest absolute difference between a variance value of each sub-block of the at least four sub-blocks and the determined average;

determining, by the video encoder, that the pixels of the current block are homogeneous based on the maximum difference value for the current block being less than a threshold difference value;

in response to determining that the pixels of the current block are homogeneous, selectively including, by the video encoder, block sizes of the plurality of block sizes that are greater than or equal to a threshold block size of the plurality of block sizes in a sub-set of block sizes to evaluate for use when intra encoding the pixels of the current block; and

evaluating block sizes in the sub-set of block sizes for use when intra encoding the pixels of the current block.

2. The method of claim 1 , wherein the threshold block size is a first threshold block size, the method further comprising:

determining, by the video encoder, that pixels of another block are not homogeneous if a maximum difference value for the other block is not less than the threshold difference value; and

in response to determining that the pixels of the other block are not homogeneous, selectively including block sizes of the plurality of block sizes that are less than or equal to a second threshold block size of the plurality of block sizes in a sub-set of block sizes to evaluate for use when intra encoding the pixels of the other block.

3. The method of claim 2 , wherein the first threshold block size equals the second threshold block size.

4. The method of claim 2 , wherein the first threshold block size is different than the second threshold block size.

5. The method of claim 1 , wherein the threshold block size is a first threshold block size, the method further comprising, in response to determining that the pixels of the current block in the particular block size are not homogeneous:

dividing a current sub-block of the at least four sub-blocks of video data into at least four second sub-blocks of video data;

determining a respective variance value for each respective second sub-block of the at least four second sub-blocks;

determining an average of the determined variance values for the at least four second sub-blocks;

determining a maximum difference value for the current sub-block based on the determined average and the determined variance values for the at least four second sub-blocks;

determining, based on the maximum difference value for the current sub-block, whether pixels of the current sub-block are homogeneous; and

in response to determining that the pixels of the current sub-block are homogeneous, selectively including block sizes of the plurality of block sizes that are greater than or equal to a second threshold block size of the plurality of block sizes but less than the first threshold block size in a sub-set of block sizes to evaluate for use when intra encoding the pixels of the current sub-block.

6. The method of claim 5 , further comprising:

in response to determining that the pixels of the current sub-block are not homogeneous, selectively including block sizes of the plurality of block sizes that are less than or equal to a third threshold block size of the plurality of block sizes in the sub-set of block sizes to evaluate for use when intra encoding the pixels of the current sub-block.

7. The method of claim 1 , further comprising:

determining, for the current block, a horizontal sum of absolute differences (SAD) value using a first pixel in each respective row of the current block as a predictor for the respective row;

determining, for the current block, a vertical SAD value using a first pixel in each respective column of the current block as a predictor for the respective column,

wherein the determination of whether the pixels of the current block are homogeneous is further based on whether a lesser of the horizontal SAD value and the vertical SAD value satisfies a threshold SAD value.

8. The method of claim 7 , further comprising:

determining a variance value for the current block,

wherein the determination of whether the pixels of the current block are homogeneous is further based on whether the variance value for the current block satisfies a threshold variance value.

9. The method of claim 8 , wherein determining whether the pixels of the current block are homogeneous comprises determining that the pixels of the current block are homogeneous in response to determining that:

the maximum difference value for the current block is less than the threshold difference value,

the lesser of the horizontal SAD value and the vertical SAD value is less than the threshold SAD value, or

the variance value for the current block is less than the threshold variance value.

10. The method of claim 8 , wherein the threshold difference value, the threshold SAD value, and the threshold variance value are fixed.

11. The method of claim 8 , further comprising:

adaptively determining one or more of the threshold difference value, the threshold SAD value, and the threshold variance value.

12. The method of claim 1 , wherein evaluating the block sizes in the sub-set of block sizes comprises:

determining a respective rate-distortion (RD) cost for each respective block size in the sub-set of block sizes, the method further comprising:

encoding, in a coded video bitstream, an indication that at least a portion of the pixels of the current block are intra encoded using the block size in the sub-set of block sizes with the lowest RD cost.

13. A video encoding device comprising:

a memory configured to store a current block of video data; and

one or more processors configured to:

determine a respective variance value for pixels of each respective sub-block of at least four sub-blocks of the current block of video data of a particular block size of a plurality of block sizes, wherein the current block of video data is included in a current picture of video data that includes a plurality of other blocks of video data;

determine an average of the determined variance values for the at least four sub-blocks;

determine a maximum difference value for the current block as a largest absolute difference between a variance value of each sub-block of the at least four sub-blocks and the determined average;

determine that the pixels of the current block are homogeneous if the maximum difference value for the current block is less than a threshold difference value;

selectively include, in response to determining that the pixels of the current block are homogeneous, block sizes of the plurality of block sizes that are greater than or equal to a threshold block size of the plurality of block sizes in a sub-set of block sizes to evaluate for use when intra encoding the pixels of the current block; and

evaluate block sizes in the sub-set of block sizes for use when intra encoding the pixels of the current block.

14. The video encoding device of claim 13 , wherein the threshold block size is a first threshold block size, and wherein the one or more processors are further configured to:

determine that the pixels of the current block are not homogeneous if the maximum difference value for the current block is not less than the threshold difference value; and

selectively include, in response to determining that the pixels of the current block are not homogeneous, block sizes of the plurality of block sizes that are less than or equal to a second threshold block size of the plurality of block sizes in the sub-set of block sizes to evaluate for use when intra encoding the pixels of the current block.

15. The video encoding device of claim 13 , wherein the threshold block size is a first threshold block size, and wherein, in response to determining that the pixels of the current block in the particular block size are not homogeneous, the one or more processors are configured to:

divide a current sub-block of the at least four sub-blocks of video data into at least four second sub-blocks of video data;

determine a respective variance value for each respective second sub-block of the at least four second sub-blocks;

determine an average of the determined variance values for the at least four second sub-blocks;

determine a maximum difference value for the current sub-block based on the determined average and the determined variance values for the at least four second sub-blocks;

determine, based on the maximum difference value for the current sub-block, whether pixels of the current sub-block are homogeneous; and

selectively include, in response to determining that the pixels of the current sub-block are homogeneous, block sizes of the plurality of block sizes that are greater than or equal to a second threshold block size of the plurality of block sizes but less than the first threshold block size in a sub-set of block sizes to evaluate for use when intra encoding the pixels of the current sub-block.

16. The video encoding device of claim 15 , wherein the one or more processors are further configured to:

selectively include, in response to determining that the pixels of the current sub-block are not homogeneous, block sizes of the plurality of block sizes that are less than or equal to a third threshold block size of the plurality of block sizes in the sub-set of block sizes to evaluate for use when intra encoding the pixels of the current sub-block.

17. The video encoding device of claim 13 , wherein the one or more processors are further configured to:

determine, for the current block, a horizontal sum of absolute differences (SAD) value using a first pixel in each respective row of the current block as a predictor for the respective row;

determine, for the current block, a vertical SAD value using a first pixel in each respective column of the current block as a predictor for the respective column,

wherein the determination of whether the pixels of the current block are homogeneous is further based on whether a lesser of the horizontal SAD value and the vertical SAD value satisfies a threshold SAD value.

18. The video encoding device of claim 17 , wherein the one or more processors are further configured to:

determine a variance value for the current block,

wherein the determination of whether the pixels of the current block are homogeneous is further based on whether the variance value for the current block satisfies a threshold variance value.

19. The video encoding device of claim 18 , wherein the one or more processors are configured to determine that the pixels of the current block are homogeneous in response to determining that:

the maximum difference value for the current block is less than the threshold difference value,

the lesser of the horizontal SAD value and the vertical SAD value is less than the threshold SAD value, or

the variance value for the current block is less than the threshold variance value.

20. A non-transitory computer-readable storage medium storing instructions that, when executed by one or more processors of a video encoder, cause the one or more processors to:

determine a respective variance value for pixels of each respective sub-block of at least four sub-blocks of a current block of video data of a particular block size of a plurality of block sizes, wherein the current block of video data is included in a current picture of video data that includes a plurality of other blocks of video data;

determine an average of the determined variance values for the at least four sub-blocks;

determine a maximum difference value for the current block as a largest absolute difference between a variance value of each sub-block of the at least four sub-blocks and the determined average;

determine that the pixels of the current block are homogeneous if the maximum difference value for the current block is less than a non-zero threshold difference value;

selectively include, in response to determining that the pixels of the current block are homogeneous, block sizes of the plurality of block sizes that are greater than or equal to a threshold block size of the plurality of block sizes in a sub-set of block sizes to evaluate for use when intra encoding the pixels of the current block; and

evaluate block sizes in the sub-set of block sizes for use when intra encoding the pixels of the current block.