Source-consistent techniques for predicting absolute perceptual video quality

Patent metadata

Field	Value
Publication number	US-10798387-B2
Application number	US-201715782586-A
Country	US
Kind code	B2
Filing date	Oct 12, 2017
Priority date	Dec 12, 2016
Publication date	Oct 6, 2020
Grant date	Oct 6, 2020

Abstract

In various embodiments, a perceptual quality application computes an absolute quality score for encoded video content. In operation, the perceptual quality application selects a model based on the spatial resolution of the video content from which the encoded video content is derived. The model associates a set of objective values for a set of objective quality metrics with an absolute quality score. The perceptual quality application determines a set of target objective values for the objective quality metrics based on the encoded video content. Subsequently, the perceptual quality application computes the absolute quality score for the encoded video content based on the selected model and the set of target objective values. Because the absolute quality score is independent of the quality of the video content, the absolute quality score accurately reflects the perceived quality of a wide range of encoded video content when decoded and viewed.

First claim

What is claimed is:

1. A computer-implemented method, comprising:
selecting a first model based on a first spatial resolution of first video content, wherein the first model is included in a plurality of models and associates a set of objective values for a set of objective quality metrics with an absolute quality score, and each model included in the plurality of models is trained using source video content having a different spatial resolution;
computing a first set of values for the set of objective quality metrics based on first encoded video content derived from the first video content;
computing a first absolute quality score for the first encoded video content based on the first model and the first set of values; and
comparing the first encoded video content to another encoded video content based on the first absolute quality score or transmitting at least a portion of the first encoded video content based on the first absolute quality score.

2. The computer-implemented method of claim 1 , wherein computing the first set of values comprises:
decoding the first encoded video content to generate a first decoded video content;
up-sampling the first decoded video content to generate first re-constructed video content having the first spatial resolution; and
computing a value for a first objective quality metric included in the set of objective quality metrics based on the first re-constructed video content and the first video content.

3. The computer-implemented method of claim 1 , wherein the first set of values includes values for at least one of temporal information, a detail loss measure, or visual information fidelity.

4. The computer-implemented method of claim 1 , wherein selecting the first model comprises:
determining a highest spatial resolution included in the plurality of resolutions that is not greater than the first spatial resolution; and
identifying a model included in the plurality of models that is associated with the highest resolution.

5. The computer-implemented method of claim 1 , further comprising, prior to selecting the first model:
encoding training video content having the first spatial resolution to generate encoded training video content;
generating training data based on the encoded training video content; and
generating the first model via a machine learning algorithm trained using the training data.

6. The computer-implemented method of claim 5 , wherein generating the training data comprises:
computing a second set of values for the set of objective quality metrics based on the encoded training video content; and
associating the second set of values with a second absolute quality score for the encoded training video content.

7. The computer-implemented method of claim 6 , further comprising generating the second absolute quality score based on one or more human-observed visual quality scores for re-constructed video content derived from the encoded training video content.

8. The computer-implemented method of claim 1 , wherein the first spatial resolution comprises 480p, 720p, 1080p, or 4K.

9. The computer-implemented method of claim 1 , further comprising:
selecting a second model based on a second spatial resolution of second video content, wherein the second model is included in the plurality of models;
computing a second set of values for the set of objective quality metrics based on second encoded video content derived from the second video content; and
computing a second absolute quality score for the second encoded video content based on the second model and the second set of values,
wherein the second absolute quality score is aggregated with the first absolute quality score in evaluating the visual quality of streamed media content.

10. A non-transitory computer-readable storage medium including instructions that, when executed by a processor, cause the processor to perform the steps of:
selecting a first model based on a first spatial resolution of first video content, wherein the first model is included in a plurality of models, and each model included in the plurality of models is trained using source video content having a different spatial resolution;
computing a first set of values for a set of objective quality metrics based on the first video content and first encoded video content derived from the first video content, wherein the set of objective quality metrics represent a plurality of deterministic video characteristics;
computing a first absolute quality score for the first encoded video content based on the first model and the first set of values; and
comparing the first encoded video content to another encoded video content based on the first absolute quality score or transmitting at least a portion of the first encoded video content based on the first absolute quality score.

11. The non-transitory computer-readable storage medium of claim 10 , wherein computing the first set of values comprises:
decoding the first encoded video content to generate a first decoded video content;
up-sampling the first decoded video content to generate first re-constructed video content having the first spatial resolution; and
computing a value for a first objective quality metric included in the set of objective quality metrics based on the first re-constructed video content and the first video content.

12. The non-transitory computer-readable storage medium of claim 11 , wherein computing the value for the first objective quality metric comprises performing one or more comparison operations between the first re-constructed video content and the first video content.

13. The non-transitory computer-readable storage medium of claim 10 , wherein selecting the first model comprises:
determining a highest spatial resolution included in the plurality of resolutions that is not greater than the first spatial resolution; and
identifying a model included in the plurality of models that is associated with the highest spatial resolution.

14. The non-transitory computer-readable storage medium of claim 10 , the steps further comprising, prior to selecting the first model:
encoding training video content having the first spatial resolution to generate encoded training video content;
generating training data based on the encoded training video content; and
generating the first model via a machine learning algorithm trained using the training data.

15. The non-transitory computer-readable storage medium of claim 14 , the steps further comprising, prior to encoding the training video content, deriving the training video content from base video content having a second spatial resolution that exceeds the first spatial resolution.

16. The non-transitory computer-readable storage medium of claim 14 , wherein the machine learning algorithm comprises a support vector machine algorithm, an artificial neural network algorithm, or a random forest algorithm.

17. The non-transitory computer-readable storage medium of claim 14 , wherein generating the training data comprises:
computing a second set of values for the set of objective quality metrics based on the encoded training video content; and
associating the second set of values with a second absolute quality score for the encoded training video content.

18. The non-transitory computer-readable storage medium of claim 10 , the steps further comprising:
selecting a second model based on a second spatial resolution of second video content, wherein the second model is included in the plurality of models;
computing a second set of values for the

Assignees

• Netflix Inc

Inventors

• Li Zhi
• Aaron Anne
• Moorthy Anush
• Bampis Christos

Classifications

• H04N17/004

  for digital television systems · CPC title

• H04N19/00

  Methods or arrangements for coding, decoding, compressing or decompressing digital video signals · CPC title

• H04N21/23439

  for generating different versions · CPC title

• H04N21/23418

  involving operations for analysing video streams, e.g. detecting features or characteristics (television picture signal circuitry for scene change detection H04N5/147; filtering for image enhancement G06T5/00; methods or arrangements for recognising scenes G06V20/00; arrangements characterised by components specially adapted for monitoring, identification or recognition of video in broadcast systems H04H60/59) · CPC title

• H04N19/146

  Data rate or code amount at the encoder output · CPC title