Objective 3D video quality assessment model

Patent metadata

Field	Value
Publication number	US-9686526-B2
Application number	US-201214349054-A
Country	US
Kind code	B2
Filing date	Mar 29, 2012
Priority date	Oct 4, 2011
Publication date	Jun 20, 2017
Grant date	Jun 20, 2017

Abstract

There is provided a method of determining the quality of a three dimensional (3D) video stream, wherein the 3D video comprises at least one 3D view composition, each 3D view composition defining a plurality of two dimensional (2D) views. The method comprises determining a characteristic of each of a plurality of 2D views. The method further comprises calculating a variation in the characteristic, the variation calculated between the 2D views corresponding to the same 3D view composition. The method further still comprises calculating a quality score for the 3D video sequence, the quality score based upon the calculated variation.

First claim

The invention claimed is:

1. A method of determining the quality of a three dimensional (3D) video stream, wherein the 3D video stream comprises at least one 3D view composition, the at least one 3D view composition defining a plurality of two dimensional (2D) views, the method comprising:
determining a characteristic of each of the plurality of 2D views wherein the characteristic of the 2D views comprises a quantization parameter of the 2D views;
calculating a variation in the characteristic, the variation calculated between the 2D views corresponding to the at least one 3D view composition, wherein the characteristic of the 2D views comprises horizontal shift;
generating a disparity map from the horizontal shift between at least two of the 2D views corresponding to the at least one 3D view composition;
detecting at least one view synthesis artifact of a plurality of view synthesis artifacts by performing operations comprising:
performing edge detection on at least one view of the plurality of 2D views to create a first edge map;
performing edge detection on a disparity map corresponding to the at least one view to create a second edge map, wherein the disparity map corresponding to the at least one view to create a second edge map comprises a depth map based on two views of the plurality of 2D views; and
calculating a metric difference between the first and the second edge maps; and

calculating a quality score for the 3D video stream based upon the variation in the characteristic, and based upon the detecting the at least one view synthesis artifact.

2. The method of claim 1 , wherein the quality score is calculated based upon detection of at least one item present in one of the 2D views but missing from another one of the 2D views; wherein an item is detected as missing based on the horizontal shift measured in the vicinity of the item exceeding the distance of the item from a view boundary.

3. The method of claim 1 , wherein the quality score is calculated based upon detection of inverted parallax; wherein inverted parallax is detected based on a majority of the calculated points of the disparity map being negative.

4. The method of claim 1 , wherein the edge detection is a Sobel operator, or Canny edge detection.

5. The method of claim 1 , further comprising:
aligning the first and second edge maps prior to calculating the symmetric difference between the first and second edge map.

6. The method of claim 1 , further comprising determining a characteristic of each of a plurality of the 2D views for a subset of a plurality of the 3D view compositions of the 3D video stream.

7. The method of claim 1 , wherein an exponential function is applied to calculate the impact the variation has on the quality score.

8. The method of claim 1 , further comprising taking an average of the variation between the determined characteristics of 2D views in a 3D view composition over a period of time or a number of video frames.

9. A non-transitory computer-readable medium, carrying instructions, which, when executed by computer logic, causes said computer logic to carry out the methods defined by method of claim 1 .

10. An apparatus for determining the quality of a three dimensional (3D) video stream, wherein the 3D video stream comprises at least one 3D view composition, the at least one 3D view composition defining a plurality of two dimensional (2D) views, the apparatus comprising:
a circuit configured to perform operations comprising:
determining a characteristic of each of a plurality of 2D views wherein the characteristic of the 2D views comprises a quantization parameter of the 2D views;
calculating a variation in the characteristic, the variation calculated between the 2D views corresponding to the at least one 3D view composition, wherein the characteristic of the 2D views comprises horizontal shift;
generating a disparity map from the horizontal shift between at least two of the 2D views corresponding to the at least one 3D view composition;
detecting at least one view synthesis artifact of a plurality of view synthesis artifacts by performing operations comprising:
performing edge detection on at least one view of the plurality of 2D views to create a first edge map;
performing edge detection on a disparity map corresponding to the at least one view to create a second edge map, wherein the disparity map corresponding to the at least one view to create a second edge map comprises a depth map based on two views of the plurality of 2D views; and
calculating a metric difference between the first and the second edge maps; and

calculating a quality score for the 3D video stream based upon the variation in the characteristic, and based upon the detecting the at least one view synthesis artifact.

11. The apparatus of method of claim 10 , wherein the circuit is further configured to calculate the quality score based upon the detection of at least one item present in one of the 2D views but missing from another one of the 2D views; wherein an item is detected as missing based on the horizontal shift calculated by the disparity map generating circuit in the vicinity of the item exceeding the distance of the item from a view boundary.

12. The apparatus of claim 10 , further comprising an inverted parallax detection circuit configured to detect inverted parallax based on a majority of the calculated points of the disparity map being negative, and wherein the output of the inverted parallax detection circuit is used by the circuit to calculate a quality score for the 3D video stream.

13. The apparatus of claim 10 , wherein the edge detection is a Sobel operator, or Canny edge detection.

Assignees

• Pettersson Martin
• Hakeem Usman
• Ericsson Telefon Ab L M

Inventors

• Pettersson Martin
• Hakeem Usman

Classifications

• H04N13/0007 Primary

  Electricity · mapped topic

• H04N2213/002

  Eyestrain reduction by processing stereoscopic signals or controlling stereoscopic devices · CPC title

• H04N13/0059

  Electricity · mapped topic

• H04N13/0033

  Electricity · mapped topic

• H04N13/0037

  Electricity · mapped topic