System and method for controlling video coding at frame level

1. A method for controlling video coding, comprising:

obtaining a target rate for an image data unit to be encoded using one or more coding parameters, wherein the one or more coding parameters are determined based on the target rate and a rate control model associated with one or more model parameters, the one or more coding parameters including a quantization parameter (QP);

determining the one or more model parameters associated with the rate control model based on an encoding of one or more reference image data units using one or more reference coding parameters;

determining the one or more coding parameters for encoding the image data unit based on the target rate and the rate control model associated with the one or more determined model parameters;

using the one or more determined coding parameters to encode the image data unit;

determining whether the image data unit contains a significant scene change, the image data unit being determined to contain the significant scene change in response to an intra-predicting block ratio in encoding the image data unit being above a threshold, the intra-predicting block ratio being defined as a ratio based on a number of intra-predicting blocks in encoding an image frame; and

updating the one or more model parameters based on a gradient method with a learning rate, the learning rate being dynamically determined as reversely related to the QP.

2. The method of claim 1 , further comprising determining the target rate based on a bit allocation or communication constraint.

3. The method of claim 1 , wherein the target rate is a target bit rate or target bits per pixel (bpp).

4. The method of claim 1 , wherein the image data unit is a group of pictures, an image frame, or a portion of an image frame.

5. The method of claim 1 , further comprising configuring a sliding window with a sequence of reference image data units.

6. The method of claim 5 , wherein at least two of the sequence of reference image data units are encoded using at least one identical coding parameter.

7. The method of claim 5 , further comprising updating the one or more model parameters based on an optimization of the rate control model over multiple sample points within the sliding window, wherein each sample point corresponds to a reference image data unit of the sequence of reference image data units.

8. A video encoder, comprising:

a memory storing one or more computer-executable instructions; and

one or more processors configured to access the memory and execute the one or more computer-executable instructions to:

obtain a target rate for an image data unit to be encoded using one or more coding parameters, wherein the one or more coding parameters are determined based on the target rate and a rate control model associated with one or more model parameters, the one or more coding parameters including a quantization parameter (QP);

determine the one or more model parameters associated with the rate control model based on an encoding of one or more reference image data units using one or more reference coding parameters;

determine the one or more coding parameters for encoding the image data unit based on the target rate and the rate control model associated with the one or more determined model parameters;

use the one or more determined coding parameters to encode the image data unit;

determine whether the image data unit contains a significant scene change, the image data unit being determined to contain the significant scene change in response to an intra-predicting block ratio in encoding the image data unit being above a threshold, the intra-predicting block ratio being defined as a ratio based on a number of intra-predicting blocks in encoding an image frame; and

update the one or more model parameters based on a gradient method with a learning rate, the learning rate being dynamically determined as reversely related to the QP.

9. The video encoder of claim 8 , wherein the target rate is determined based on a bit allocation or communication constraint.

10. The video encoder of claim 8 , wherein the target rate is a target bit rate or target bits per pixel (bpp).

11. The video encoder of claim 8 , wherein the image data unit is a group of pictures, an image frame, or a portion of an image frame.

12. The video encoder of claim 8 , wherein a sliding window is configured with a sequence of reference image data units.

13. The video encoder of claim 12 , wherein at least two of the sequence of reference image data units are encoded using at least one identical coding parameter.

14. The video encoder of claim 12 , wherein the one or more model parameters are updated further based on an optimization of the rate control model over multiple sample points within the sliding window, wherein each sample point corresponds to a reference image data unit of the sequence of reference image data units.

15. One or more non-transitory computer-readable storage media storing computer-executable instructions that, when executed by a computing system, configure the computing system to perform operations comprising:

obtaining a target rate for an image data unit to be encoded using one or more coding parameters, wherein the one or more coding parameters are determined based on the target rate and a rate control model associated with one or more model parameters, the one or more coding parameters including a quantization parameter (QP);

determining the one or more model parameters associated with the rate control model based on an encoding of one or more reference image data units using one or more reference coding parameters;

determining the one or more coding para ers for encoding the image data unit based on the target rate and the rate control model associated with the one or more determined model parameters;

using the one or more determined coding parameters to encode the image data unit;

determining whether the image data unit contains a significant scene change, the image data unit being determined to contain the significant scene change in response to an intra-predicting block ratio in encoding the image data unit being above a threshold, the intra-predicting block ratio being defined as a ratio based on a number of intra-predicting blocks in encoding an image frame

updating the one or more model parameters based on a gradient method with a learning rate, the learning rate being dynamically determined as reversely related to the QP.

16. The method of claim 1 , wherein the learning rate is dynamically determined as 1/(QP 2 +1).

17. The video encoder of claim 8 , wherein the learning rate is dynamically determined as 1/(QP 2 +1).

18. The method of claim 1 , wherein the target rate is target bits per pixel (bpp).