Method for measuring video quality using a reference, and apparatus for measuring video quality using a reference
View Patent ↗The purpose of an objective video quality evaluation is to automatically assess the quality of video sequences in agreement with human quality perception. The invention addresses the effects of the introduction of a temporal dimension by focusing on the temporal evolutions of spatial distortions, since it has been found that a spatial quality variation cannot be evaluated by simple subtraction of the spatial quality of neighboring frames. An improved method for estimating perceived video quality comprises steps of calculating a first similarity map between adjacent frames of a current sequence, calculating a second similarity map between the corresponding reference frames, and calculating (smg 3 ) a third similarity map, which provides a numerical quality value.
1. A method for measuring video quality of a sequence of distorted video pictures, wherein the respective video pictures are also available as undistorted video pictures, comprising steps of
dividing the pictures into blocks of equal size;
generating a first similarity map, wherein the blocks of a distorted picture f and the corresponding collocated blocks of a temporally neighboring distorted picture f′ are input to a first similarity function f sim,1 (f,f′)={right arrow over (CM)}(f,f′) that outputs a first similarity map;
generating a second similarity map, wherein the blocks of an undistorted picture f 0 and the corresponding collocated blocks of a temporally neighboring undistorted picture f′ 0 are input to the first similarity function f sim,1 (f 0 ,f′ 0 )={right arrow over (CM)}(f 0 ,f′ 0 ) that outputs a second similarity map;
calculating the similarity between the first similarity map and the second similarity map, wherein a second similarity function f sim,2 (X,Y) is used according to VQM variation =f sim 2 ({right arrow over (CM)}(f,f′),{right arrow over (CM)}(f 0 ,f′ 0 )) and wherein a single numerical value is obtained, and wherein the second similarity function performs a calculation according to
f
sim
2
(
X
,
Y
)
=
∑
i
=
1
n
x
i
-
y
i
n
;
and
providing said single numerical value as a measure for the video quality of said sequence of distorted video pictures.
2. Method according to claim 1 , wherein the first similarity function performs a Pearson Correlation according to
f
sim
1
(
X
,
Y
)
=
PearsonCorrelation
(
X
,
Y
)
=
∑
i
=
1
n
(
x
i
-
X
_
)
(
y
i
-
Y
_
)
(
n
-
1
)
σ
X
σ
Y
,
with x and y being pixel signals, X and Y being the average of the pixel signals, and σ x and σ y being the standard deviation of the pixel signals.
3. Method according to claim 1 , wherein the first and the second similarity functions are different from each other.
4. Method according to claim 1 , wherein all blocks have equal size.
5. Method according to claim 1 , wherein the measurement is performed only on a portion of a picture.
6. An apparatus for measuring video quality of a sequence of distorted video pictures, wherein the respective video pictures are also available as undistorted video pictures, comprising
picture dividing unit for dividing the pictures into blocks of equal size;
first similarity map generating unit for generating a first similarity map, wherein the blocks of a distorted picture f and the corresponding collocated blocks of a temporally neighboring distorted picture f′ are input to a first similarity function f sim,1 (f,f′)={right arrow over (CM)}(f,f′) that outputs a first similarity map;
second similarity map generating unit for generating a second similarity map, wherein the blocks of an undistorted picture f 0 and the corresponding collocated blocks of a temporally neighboring undistorted picture f′ 0 are input to the first similarity function f sim,1 (f,f′ 0 )={right arrow over (CM)}(f 0 ,f′ 0 ) that outputs a second similarity map;
third similarity map generating unit (smg 3 ) for calculating a similarity between the first similarity map and the second similarity map, wherein a second similarity function f sim,2 (X,Y) is used according to VQM variation =f sim 2 ({right arrow over (CM)}(f,f′),{right arrow over (CM)}(f 0 ,f′ 0 )), and wherein a single numerical value (VQM) is obtained, and wherein the second similarity function performs a calculation according to
f
sim
2
(
X
,
Y
)
=
∑
i
=
1
n
x
i
-
y
i
n
;
and
output unit for providing said single numerical value as a measure for the video quality of said sequence of distorted video pictures.
7. Apparatus according to claim 6 , wherein the first similarity function performs a Pearson Correlation according to
f
sim
1
(
X
,
Y
)
=
PearsonCorrelation
(
X
,
Y
)
=
∑
i
=
1
n
(
x
i
-
X
_
)
(
y
i
-
Y
_
)
(
n
-
1
)
σ
X
σ
Y
,
with x and y being pixel signals, X and Y being the average of the pixel signals, and σ x and σ y being the standard deviation of the pixel signals.
8. Apparatus according to claim 6 , wherein the first and the second similarity map generating units perform the same function, and both perform a different function than the third similarity map generating unit.
9. Apparatus according to claim 6 , wherein all blocks of the picture have equal size.
10. Apparatus according to claim 6 , wherein the measurement is performed only on a portion of a picture.
11. A non-transitory computer readable storage medium having stored thereon computer executable instructions that when executed by a computer causes the computer to perform a method for measuring video quality of a sequence of distorted video pictures, wherein the respective video pictures are also available as undistorted video pictures, the method comprising the computer executed steps of
dividing the pictures into blocks of equal size;
generating a first similarity map, wherein the blocks of a distorted picture f and the corresponding collocated blocks of a temporally neighboring distorted picture f′ are input to a first similarity function f sim,1 (f,f′)={right arrow over (CM)}(f,f′) that outputs a first similarity map;
generating a second similarity map, wherein the blocks of an undistorted picture f 0 and the corresponding collocated blocks of a temporally neighboring undistorted picture f′ 0 are input to the first similarity function f sim,1 (f 0 ,f′ 0 )={right arrow over (CM)}(f 0 ,f′ 0 ) that outputs a second similarity map;
calculating the similarity between the first similarity map and the second similarity map, wherein a second similarity function f sim,2 (X,Y) is used according to VQM variation =f sim 2 ({right arrow over (CM)}(f,f′), {right arrow over (CM)}(f 0 ,f′ 0 )) and wherein a single numerical value is obtained, and wherein the second similarity function performs a calculation according to
f
sim
2
(
X
,
Y
)
=
∑
i
=
1
n
x
i
-
y
i
n
;
and
providing said single numerical value as a measure for the video quality of said sequence of distorted video pictures.