Objective method and system for estimating perceived image and video sharpness
View Patent ↗A method and system for estimating the quality of video data without gaining access to the source data, are configured to perform a Fast Fourier Transform (FFT) from at least some of the pixel values of the image in at least the horizontal direction, and at most, in the horizontal and vertical directions for the luminance and chrominance components. An average FFT is used to calculate kurtosis and skewness, then an image-quality metric is determined by calculating an offset from the kurtosis and skewness outputs.
1. A method for estimating the sharpness quality of input-video data, the method comprising the steps of:
converting said input-video data into corresponding digital-image data comprised of pixel values;
generating Fast Fourier Transform (FFT) from at least some of said pixel values and computing an average FFT on said generated FFT;
calculating a statistical reference using said average FFT; and,
producing an image-quality metric from said calculated statistical reference,
wherein the statistical-reference calculating step comprises the step of calculating kurtosis using said average FFT, and wherein said image-quality metric (SH) is produced by calculating an offset of said kurtosis according to one of the following equations:
SH= 2.5546 Ln (kurtosis)−3.8496, and
SH ( var ) 0.485 /100,
where SH represents said image-quality metric and var represents the variance as follows:
var
=
∑
i
=
1
N
(
Y
i
-
Y
_
)
2
(
N
-
1
)
.
wherein {overscore (Y)} represents an average FFT, Y i represents an individual value of said generated FFT, and N represents the size of said generated FFT.
2. The method of claim 1 , wherein said kurtosis is calculated according to the following equation:
Kurtosis
=
∑
i
=
1
N
(
Y
i
-
Y
_
)
4
(
N
-
1
)
s
4
,
and
s
2
=
∑
i
=
1
N
(
Y
i
-
Y
_
)
2
(
N
-
1
)
,
wherein {overscore (Y)} represents an average FFT, s represents variance, Y i represents an individual value of said generated FFT, and N represents the size of said generated FFT.
3. The method of claim 1 , wherein said FFT is generated on every predetermined horizontal line of said pixel values.
4. A method for estimating the sharpness quality of input-video data, the method comprising the steps of:
converting said input-video data into corresponding digital-image data comprised of pixel values;
generating Fast Fourier Transform (FFT) from at least some of said pixel values and computing an average FFT on said generated FFT;
calculating a statistical reference using said average FFT; and,
producing an image-quality metric from said calculated statistical reference,
wherein the statistical reference calculating step comprises the step of calculating kurtosis using said average FFT, and wherein said image-quality metric (SH) is produced by calculating an offset of said kurtosis according to the following equation:
SH
=
(
var
)
0485
/
100
,
and
var
=
∑
i
=
1
N
(
Y
i
-
Y
_
)
2
(
N
-
1
)
.
wherein {overscore (Y)} represents an average FFT, Y i represents an individual value of said generated FFT, and N represents the size of said generated FFT.
5. The method of claim 4 , wherein said FFT is generated on every predetermined horizontal line of said pixel values.
6. A method for estimating the sharpness quality of input-video data, the method comprising the steps of:
converting said input-video data into corresponding digital-image data comprised of pixel values;
generating Fast Fourier Transform (FFT) from at least some of said pixel values and computing an average FFT on said generated FFT;
calculating a statistical reference using said average FFT; and,
producing an image-quality metric from said calculated statistical reference,
wherein the statistical-reference calculating step comprises the step of calculating skewness using said average FFT, and wherein said image-quality metric (SH) is produced by calculating an offset of said skewness according to the following equations:
SH
=
Tan
(
C
1
Sk
var
)
+
C
2
Sk
2
,
and
var
=
∑
i
=
1
N
(
Y
i
-
Y
_
)
2
(
N
-
1
)
,
wherein {overscore (Y)} represents an average FFT, C 1 =5E4, C 2 =0.5, Sk represents said skewness, Y i represents an individual value of said generated FFT, and N represents the size of said generated FFT.
7. The method of claim 6 , wherein said skewness is calculated according to the following equation:
Skewness
=
∑
i
=
1
N
(
Y
i
-
Y
_
)
3
(
N
-
1
)
s
3
.
wherein {overscore (Y)} represents an average FFT, s represents variance, Y i represents an individual value of said generated FFT, and N represents the size of said generated FFT.
8. The method of claim 6 , wherein said FFT is generated on every predetermined horizontal line of said pixel values.
9. A system for estimating the sharpness quality of input-video data, comprising:
a converter for converting said input-video data into corresponding digital-image data comprised of image pixel values;
an FFT generator, adapted to receive an output of said converter, for generating Fast Fourier Transform (FFT) from at least some of said image-pixel values and for computing an average FFT on said generated FFT;
a statistical calculator for calculating a statistical reference based on said average FFT; and,
an image-quality generator for producing an image-quality metric from said calculated statistical reference,
wherein said statistical calculator is further configured to determine a kurtosis of said generated FFT, and wherein said image-quality generator determines said image-quality metric (SH) based on said determined kurtosis using one of the following equations:
SH= 2.5546 Ln (kurtosis)−3.8496, and
SH =( var ) 0.485 /100,
where SH represents said image-quality metric and var represents the variance as follows:
var
=
∑
i
=
1
N
(
Y
i
-
Y
_
)
2
(
N
-
1
)
.
wherein {overscore (Y)} represents an average FFT, k represents said skewness, Y i represents an individual value of said generated FFT, and N represents the size of said generated FFT.
10. The system of claim 9 , wherein said FFT is generated on every predetermined horizontal line of said digital-pixel values.
11. A system for estimating the sharpness quality of input-video data, comprising:
a converter for converting said input-video data into corresponding digital-image data comprised of image pixel values;
an FFT generator, adapted to receive an output of said converter, for generating Fast Fourier Transform (FFT) from at least some of said image-pixel values and for computing an average FFT on said generated FFT;
a statistical calculator for calculating a statistical reference based on said average FFT; and,
an image-quality generator for producing an image-quality metric from said calculated statistical reference,
wherein said statistical calculator is further configured to determine a kurtosis of said generated FFT, and wherein said kurtosis is calculated according to the following equation:
Kurtosis
=
∑
i
=
1
N
(
Y
i
-
Y
_
)
4
(
N
-
1
)
s
4
,
and
s
2
=
∑
i
=
1
N
(
Y
i
-
Y
_
)
2
(
N
-
1
)
,
wherein {overscore (Y)} represents an average FFT, Y i represents an individual value of said generated FFT, N represents the size of said generated FFT, and s represents variance.
12. The system of claim 11 , wherein said FFT is generated on every predetermined horizontal line of said digital-pixel values.
13. A system for estimating the sharpness quality of input-video data, comprising:
a converter for converting said input-video data into corresponding digital-image data comprised of image pixel values;
an FFT generator, adapted to receive an output of said converter, for generating Fast Fourier Transform (FFT) from at least some of said image-pixel values and for computing an average FFT on said generated FFT;
a statistical calculator for calculating a statistical reference based on said average FFT; and,
an image-quality generator for producing an image-quality metric from said calculated statistical reference,
wherein said statistical calculator is further configured to determine a skewness of said generated FFT, and wherein said image-quality generator determines said image-quality metric (SH) based on said skewness using one of the following equations:
SH
=
Tan
(
C
1
Sk
var
)
+
C
2
Sk
2
,
and
var
=
∑
i
=
1
N
(
Y
i
-
Y
_
)
2
(
N
-
1
)
,
wherein {overscore (Y)} represents an average FFT, C 1 =5E4, C2=0.5, Sk represents said skewness, Y i represents an individual value of said generated FFT, and N represents the size of said generated FFT.
14. The system of claim 13 , wherein said FFT is generated on every predetermined horizontal line of said digital-pixel values.
15. A system for estimating the sharpness quality of input-video data, comprising:
a converter for converting said input-video data into corresponding digital-image data comprised of image pixel values;
an FFT generator, adapted to receive an output of said converter, for generating Fast Fourier Transform (FFT) from at least some of said image-pixel values and for computing an average FFT on said generated FFT;
a statistical calculator for calculating a statistical reference based on said average FFT; and,
an image-quality generator for producing an image-quality metric from said calculated statistical reference,
wherein said statistical calculator is further configured to determine a skewness of said generated FFT and wherein said skewness is calculated according to the following equation:
Skewness
=
∑
i
=
1
N
(
Y
i
-
Y
_
)
3
(
N
-
1
)
s
3
.
wherein {overscore (Y)} represents an average FFT, s represents variance, Y i represents an individual value of said generated FFT, and N represents the size of said generated FFT.
16. The system of claim 15 , wherein said FFT is generated on every predetermined horizontal line of said digital-pixel values.
17. A machine-readable medium having stored thereon data representing sequences of instructions, and the sequences of instructions which, when executed by a processor, cause the processor to:
convert an input-video data into corresponding digital-image data comprised of pixel values;
generate Fast Fourier Transform (FFT) from at least some of said image pixel values and compute an average FFT on said generated FFT;
calculate a statistical reference using said average FFT; and,
produce an image-quality metric from said calculated statistical reference,
wherein the statistical-reference calculating step comprises the step of calculating kurtosis using said average FFT, and wherein said image-quality metric (SH) is produced by calculating an offset of said kurtosis according to one of the following equations:
SH= 2.5546 Ln (kurtosis)−3.8496, and
SH =( var ) 0.485 /100,
where SH represents said image-quality metric and var represents the valiance as follows:
var
=
∑
i
=
1
N
(
Y
i
-
Y
_
)
2
(
N
-
1
)
,
wherein {overscore (Y)} represents an average FFT, Y i represents an individual value of said generated FFT, and N represents the size of said generated FFT.