Refining filter for inter layer prediction of scalable video coding

1. A computer-implemented method for performing scalable video coding, comprising:

applying lower layer processing to reconstructed reference layer video data, wherein the lower layer processing comprises at least one of up-sampling for spatial scalability or tone mapping for bit-depth scalability;

applying each of a plurality of fixed refining filters to the processed reconstructed reference layer video picture to generate a plurality of refined pictures, each of the plurality of refined pictures corresponding to one of the plurality of fixed refining filters being separately applied to the reconstructed reference layer video picture, wherein the plurality of fixed refining filters comprise a low frequency enhancement filter, a high frequency enhancement filter, a particular frequency band enhancement filter, a high-pass frequency filter, a band-pass frequency filter, a low-pass frequency filter, a symmetric fixed refining filter, an asymmetric fixed refining filter, a one-dimensional fixed refining filter, and a two-dimensional fixed refining filter;

selecting a chosen refined picture from the plurality of refined pictures based on a comparison of each of the plurality of refined pictures to the processed reconstructed reference layer video picture;

determining an individual fixed refining filter from the plurality of fixed refining filters, the individual fixed refining filter corresponding to the fixed refining filter applied to the reconstructed reference layer video picture to generate the chosen refined picture;

performing inter layer prediction for an enhancement layer to generate enhancement layer video data associated with the reconstructed reference layer video picture based at least in part on the chosen refined picture of the plurality of refined pictures;

generating a fixed refining filter indicator corresponding to the selected individual fixed refining filter; and

encoding a scalable bitstream with compressed coefficients corresponding to the reference layer video picture, compressed coefficients corresponding to the enhancement layer video data, and the fixed refining filter indicator corresponding to the selected individual fixed refining filter.

2. The method of claim 1 , wherein applying a first fixed refining filter of the plurality of fixed refining filters comprises taking the dot product of the reconstructed reference layer video data and filter coefficients and adding offset values to generate a first refined picture of the plurality of refined pictures.

3. The method of claim 1 , further comprising:

generating enhancement layer output video data associated with the enhancement layer based at least in part on the inter layer prediction.

4. A system for video coding on a computer, comprising:

one or more memory stores;

one or more processors coupled to the one or more memory stores, the one or more processors to:

apply lower layer processing to reconstructed reference layer video data, wherein the lower layer processing comprises at least one of up-sampling for spatial scalability or tone mapping for bit-depth scalability;

apply each of a plurality of fixed refining filters to the processed reconstructed reference layer video picture to generate a plurality of refined pictures, each of the plurality of refined pictures corresponding to one of the plurality of fixed refining filters being separately applied to the reconstructed reference layer video picture, wherein the plurality of fixed refining filters comprise a low frequency enhancement filter, a high frequency enhancement filter, a particular frequency band enhancement filter, a high-pass frequency filter, a band-pass frequency filter, a low-pass frequency filter, a symmetric fixed refining filter, an asymmetric fixed refining filter, a one-dimensional fixed refining filter, and a two-dimensional fixed refining filter;

select a chosen refined picture from the plurality of refined pictures based on a comparison of each of the plurality of refined pictures to the processed reconstructed reference layer video picture;

determine an individual fixed refining filter from the plurality of fixed refining filters, the individual fixed refining filter corresponding to the fixed refining filter applied to the reconstructed reference layer video picture to generate the chosen refined picture;

perform inter layer prediction for an enhancement layer to generate enhancement layer video data associated with the reconstructed reference layer video picture based at least in part on the chosen refined picture of the plurality of refined pictures;

generate a fixed refining filter indicator corresponding to the selected individual fixed refining filter; and

encode a scalable bitstream with compressed coefficients corresponding to the reference layer video picture, compressed coefficients corresponding to the enhancement layer video data, and the fixed refining filter indicator corresponding to the selected individual fixed refining filter.

5. The system of claim 4 , wherein the one or more processors are to apply a first fixed refining filter of the plurality of fixed refining filters by determination of a dot product of the reconstructed reference layer video data and filter coefficients and addition of offset values to generate a first refined picture of the plurality of refined pictures.

6. At least one non-transitory machine readable medium comprising a plurality of instructions that in response to being executed on a computing device, cause the computing device to perform scalable video coding by:

applying lower layer processing to reconstructed reference layer video data, wherein the lower layer processing comprises at least one of up-sampling for spatial scalability or tone mapping for bit-depth scalability;

applying each of a plurality of fixed refining filters to the processed reconstructed reference layer video picture to generate a plurality of refined pictures, each of the plurality of refined pictures corresponding to one of the plurality of fixed refining filters being separately applied to the reconstructed reference layer video picture, wherein the plurality of fixed refining filters comprise a low frequency enhancement filter, a high frequency enhancement filter, a particular frequency band enhancement filter, a high-pass frequency filter, a band-pass frequency filter, a low-pass frequency filter, a symmetric fixed refining filter, an asymmetric fixed refining filter, a one-dimensional fixed refining filter, and a two-dimensional fixed refining filter;

selecting a chosen refined picture from the plurality of refined pictures based on a comparison of each of the plurality of refined pictures to the processed reconstructed reference layer video picture;

determining an individual fixed refining filter from the plurality of fixed refining filters, the individual fixed refining filter corresponding to the fixed refining filter applied to the reconstructed reference layer video picture to generate the chosen refined picture;

performing inter layer prediction for an enhancement layer to generate enhancement layer video data associated with the reconstructed reference layer video picture based at least in part on the chosen refined picture of the plurality of refined pictures;

generating a fixed refining filter indicator corresponding to the selected individual fixed refining filter; and

encoding a scalable bitstream with compressed coefficients corresponding to the reference layer video picture, compressed coefficients corresponding to the enhancement layer video data, and the fixed refining filter indicator corresponding to the selected individual fixed refining filter.

7. The machine readable medium of claim 6 , wherein applying a first fixed refining filter of the plurality of fixed refining filters comprises taking the dot product of the reconstructed reference layer video data and filter coefficients and adding offset values to generate a first refined picture of the plurality of refined pictures.