Technique for video quality estimation

Patent metadata

Field	Value
Publication number	US-10728538-B2
Application number	US-201013521172-A
Country	US
Kind code	B2
Filing date	Feb 1, 2010
Priority date	Jan 11, 2010
Publication date	Jul 28, 2020
Grant date	Jul 28, 2020

Abstract

An objective video quality estimation technique is disclosed. The technique may be based on a video bitstream model, using parameters taken from the video coding layer of the bitstream for estimating the quality of the video. The technique can be implemented as a method, a computer program, a computer program product, a device, or any one of a server node, a client terminal and a network node comprising the device. As a method embodiment, the technique comprises receiving a video bitstream comprising a series of picture frames; determining an error occurrence in a picture frame of the video bitstream; determining at least one of a temporal propagation and a spatial propagation of the error; and estimating the quality of the video bitstream based on result of the determination.

First claim

The invention claimed is:

1. A video quality estimation method, comprising:
receiving an encoded video bitstream comprising a series of picture frames;
determining, from the encoded video bitstream, that an error has occurred in a macro block in one of the picture frames of the encoded video bitstream;
keeping track of a temporal propagation of the error that has occurred in the macro block by using intra prediction based on the encoded video bitstream that relates to the macro block affected by the error in response to determining that the error has occurred; and
determining, from the encoded video bitstream, a visual degradation caused by the error based on keeping track of the temporal propagation of the error that has occurred in the macro block by determining a penalty value for the macro block affected by the error, wherein the penalty value is indicative of an amount of the visual degradation caused by the error that has occurred in the macro block,
wherein the penalty value is calculated based on the encoded video bitstream according to one or more of the following rules:
assigning a higher penalty value to the error if the error occurs in a scene change;
assigning a higher penalty value to the error if the encoded video bitstream comprises motion data directly after an error-affected picture frame and if a relative magnitude of the motion is large;
assigning no penalty value to the error for the macro block if the macro block is a skip or is an estimated skip, unless there is a scene change or an estimated scene change;
decreasing the penalty value assigned to the error for at least one subsequent picture frame by a predetermined amount or factor; and
resetting the penalty value upon reception of an intra-coded macro block or an intra-coded picture frame.

2. The method according to claim 1 , wherein the error occurs at or affects a plurality of picture frame partitions in the one of the picture frames, and wherein determining the visual degradation comprises calculating a separate penalty value for each of the plurality of partitions as a value indicative of an amount of the visual degradation caused by the error.

3. The method according to claim 1 ,
wherein the series of picture frames comprises a reference picture frame and a prediction picture frame, and
wherein the prediction picture frame comprises a motion vector pointing at a position in the reference picture, and
wherein the temporal propagation of the error is determined based on the motion vector information including the motion vector.

4. The method according to claim 3 , wherein keeping track of the temporal propagation of the error comprises keeping track of the temporal propagation of the error by using the motion vector information including the motion vector and the reference picture which the motion vector points at, and wherein the motion vector relates to the macro block affected by the error.

5. The method according to claim 1 , further comprising:
collecting the visual degradation in a loss penalty map, wherein the loss penalty map has the same resolution as the associated picture or a lower resolution than the picture.

6. The method according to claim 5 , wherein the picture frame comprises a plurality of partitions arranged in the form of a partition map, and the loss penalty map has the same resolution as the partition map.

7. The method according to claim 1 , wherein keeping track of the temporal propagation of the error comprises weighting a visual degradation for a previous error.

8. The method according to claim 1 , further comprising:
aggregating the visual degradation caused by the error over at least one of time and space; and
based on a result of the aggregation, calculating a score indicative of quality of the encoded video bitstream, wherein calculating the score comprises calculating the score periodically or using a sliding window approach where a new score for a predefined time duration is calculated at a predefined rate.

9. The method according to claim 1 , wherein the encoded video bitstream comprises at least a portion of a video coding layer, wherein the video coding layer comprises one or more parameters, and wherein determining that the error has occurred is based on at least one of the parameters.

10. A computer program product comprising a non-transitory computer-readable recording medium and program code portions, which when run on a computer cause the computer to perform the method of claim 1 , stored on the non-transitory computer-readable recording medium.

11. The method of claim 1 , wherein keeping track of the temporal propagation of the error comprises determining an error propagation of the error based on keeping track of the temporal propagation of the error, wherein determining the error propagation comprises determining the error propagation of the error from a partition of the one of the picture frames into other partitions of the one of the picture frames or into other ones of the picture frames.

12. The method according to claim 1 , further comprising:
determining whether the encoded video bitstream comprises motion directly after the error occurs; and
assigning a higher penalty value to the error in response to determining that the encoded video bitstream comprises motion directly after the error occurs, wherein the higher penalty value factors into an overall quality score for the encoded video bitstream.

13. The method according to claim 1 , further comprising visually presenting a position of the error in the one of the picture frames in a loss penalty map.

14. The method according to claim 1 , wherein keeping track of a temporal propagation of the error comprises determining an error propagation of the error by keeping track of the temporal propagation of the error using the intra prediction that relates to the macro block affected by the error in response to determining that the error has occurred, the method further comprising:
estimating a quality of the encoded video bitstream based on the error propagation.

15. The method according to claim 1 , wherein keeping track of a temporal propagation of the error comprises determining an error propagation of the error by keeping track of the temporal propagation of the error using the intra prediction that relates to the macro block affected by the error in response to determining that the error has occurred, the method further comprising:
estimating a quality of the encoded video bitstream based on the error propagation and based on the visual degradation.

16. The method according to claim 1 , wherein the penalty value is calculated by assigning no penalty value to the error for the macro block responsive to the macro block being a skip or an estimated skip.

17. A video quality estimation method, comprising:
receiving a video bitstream comprising a series of picture frames;
determining that an error has occurred in a macro block in one of the picture frames of the video bitstream;
keeping track of at least one of a temporal propagation and a spatial propagation of the error that has occurred in the macro block by using at least one of motion vector information and intra prediction that relates to the macro block affected by the error in response to determining that the error has occurred; and
determining a visual degradation caused by the error based on keeping track of the at least one of the temporal propagation and the spatial propagation of the error that has occurred in the macro block by determining a penalty value for the macro block affected by the error, wherein the penalty

Assignees

• Pettersson Martin
• Rossholm Andreas
• Ericsson Telefon Ab L M

Inventors

• Pettersson Martin
• Rossholm Andreas

Classifications

• H04N17/004 Primary

  for digital television systems · CPC title