Demosaicking single-sensor camera raw data
An image processing method that demosaicks a mosaic input image of R, G, and B components to generate a full color output image. The image processing method calculates both vertical and horizontal luminance-chrominance difference components for each pixel of the mosaic input image. Next, the image processing method calculates an enhanced version of both the vertical and horizontal luminance-chrominance difference components for each pixel of the mosaic input image. Next, the image processing method evaluates the variations in the enhanced luminance-chrominance difference components in order to create an edge directional map indicating the direction in which demosaicking should be performed. Then, the image processing method interpolates a G component for each of the pixels with the original R and B components using the edge directional map. Next, the image processing method interpolates an R component for each of the pixels with the original G and B components and a B component for each of the pixels with the original R and G components. Finally, the image processing method enhances the interpolated R and B components using the edge directional map.
1. An image processing apparatus that receives a mosaic input image and subjects the mosaic input image subject to a series of image processing to generate a full color output image, the image processing apparatus comprising a memory and a processing unit having:
a luminance-chrominance difference component computation module configured to calculate both vertical and horizontal luminance-chrominance difference components for each pixel of the mosaic input image;
a luminance-chrominance difference component enhancement module configured to calculate enhanced versions of both vertical and horizontal luminance-chrominance difference components for each pixel of the mosaic input image;
an edge direction detection module configured to evaluate the variations in the enhanced luminance-chrominance difference components in order to create an edge directional map indicating the direction in which demosaicking should be performed;
a luminance component interpolation module configured to interpolate a G component for each of the original R and B components using the edge directional map;
a chrominance component interpolation module configured to interpolate an R component for each of the original G and B components and a B component for each of the original R and G components; and
a chrominance component enhancement module configured to enhance the interpolated R and B components using the edge directional map; and wherein
the luminance-chrominance difference component computation module is configured to calculate both vertical and horizontal luminance-chrominance difference components Δ v and Δ h for each pixel (r,s) of the mosaic input image z according to the following equations:
Δ
(
r
,
s
)
v
=
{
z
(
r
,
s
)
-
(
z
(
r
-
1
,
s
)
+
z
(
r
+
1
,
s
)
)
/
2
for
(
odd
r
,
odd
s
)
and
(
even
r
,
even
s
)
(
z
(
r
-
1
,
s
)
+
z
(
r
+
1
,
s
)
)
/
2
-
z
(
r
,
s
)
otherwise
;
and
Δ
(
r
,
s
)
h
=
{
z
(
r
,
s
)
-
(
z
(
r
,
s
-
1
)
+
z
(
r
,
s
+
1
)
)
/
2
for
(
odd
r
,
odd
s
)
and
(
even
r
,
even
s
)
(
z
(
r
,
s
-
1
)
+
z
(
r
,
s
+
1
)
)
/
2
-
z
(
r
,
s
)
otherwise
.
2. The image processing apparatus as recited in claim 1 , wherein:
the luminance-chrominance difference component enhancement module is configured to calculate the enhanced versions {circumflex over (Δ)} v and {circumflex over (Δ)} h of the vertical and horizontal luminance-chrominance difference components Δ v and Δ h , respectively, for each pixel (r, s) of the mosaic input image z according to the following equations:
Δ
^
(
r
,
s
)
v
=
(
Δ
(
r
-
2
,
s
)
v
+
2
Δ
(
r
-
1
,
s
)
v
+
4
Δ
(
r
,
s
)
v
+
2
Δ
(
r
+
1
,
s
)
v
+
Δ
(
r
+
2
,
s
)
v
)
/
10
;
and
Δ
^
(
r
,
s
)
h
=
(
Δ
(
r
,
s
-
2
)
h
+
2
Δ
(
r
,
s
-
1
)
h
+
4
Δ
(
r
,
s
)
h
+
2
Δ
(
r
,
s
+
1
)
h
+
Δ
(
r
,
s
+
2
)
h
)
/
10.
3. The image processing apparatus as recited in claim 2 , wherein the edge direction detection module is configured to evaluate the variations in the enhanced luminance-chrominance difference components {circumflex over (Δ)} v and {circumflex over (Δ)} h in order to create an edge directional map indicating the direction in which demosaicking should be performed according to the following equation:
d
(
r
,
s
)
=
{
1
for
υ
(
r
,
s
)
v
<
υ
(
r
,
s
)
h
3
for
υ
(
r
,
s
)
v
>
υ
(
r
,
s
)
h
2
otherwise
where
:
υ
(
r
,
s
)
v
=
∑
i
=
-
1
1
(
(
Δ
^
(
r
-
1
,
s
+
i
)
v
-
Δ
^
(
r
,
s
+
i
)
v
)
2
+
(
Δ
^
(
r
,
s
+
i
)
v
-
Δ
^
(
r
+
1
,
s
+
i
)
v
)
2
+
(
Δ
^
(
r
-
1
,
s
+
i
)
v
-
Δ
^
(
r
+
1
,
s
+
i
)
v
)
2
)
;
υ
(
r
,
s
)
h
=
∑
i
=
-
1
1
(
(
Δ
^
(
r
+
i
,
s
-
1
)
h
-
Δ
^
(
r
+
i
,
s
)
h
)
2
+
(
Δ
^
(
r
+
i
,
s
)
h
-
Δ
^
(
r
+
i
,
s
+
1
)
h
)
2
+
(
Δ
^
(
r
+
i
,
s
-
1
)
h
-
Δ
^
(
r
+
i
,
s
+
1
)
h
)
2
)
;
and
the d (r,s) value indicates that demosaicking should be performed in the vertical direction (for d (r,s) =1), the horizontal direction (for d (r,s) =3), or using samples located in both the vertical and horizontal directions (for d (r,s) =2).
4. The image processing apparatus as recited in claim 3 , wherein the luminance component interpolation module is configured to interpolate a luminance component for each of the original R and B components using the edge directional map d (r,s) according to the following equation:
x
(
r
,
s
)
2
=
{
z
(
r
,
s
)
+
(
Δ
^
(
r
-
1
,
s
)
v
+
Δ
^
(
r
+
1
,
s
)
v
)
/
2
for
d
(
r
,
s
)
=
1
z
(
r
,
s
)
+
(
Δ
^
(
r
,
s
-
1
)
h
+
Δ
^
(
r
,
s
+
1
)
h
)
/
2
for
d
(
r
,
s
)
=
3
z
(
r
,
s
)
+
(
Δ
^
(
r
-
1
,
s
)
v
+
Δ
^
(
r
+
1
,
s
)
v
+
Δ
^
(
r
,
s
-
1
)
h
+
Δ
^
(
r
,
s
+
1
)
h
)
/
4
otherwise
.
5. The image processing apparatus as recited in claim 4 , wherein the chrominance component interpolation module is configured to interpolate an R component for each of the original G and B components and a B component for each of the original R and G components according to the following equations:
x
(
r
,
s
)
1
=
x
(
r
,
s
)
2
+
(
(
x
(
r
-
1
,
s
-
1
)
1
-
x
(
r
-
1
,
s
-
1
)
2
)
+
(
x
(
r
-
1
,
s
+
1
)
1
-
x
(
r
-
1
,
s
+
1
)
2
)
+
(
x
(
r
+
1
,
s
-
1
)
1
-
x
(
r
+
1
,
s
-
1
)
2
)
+
(
x
(
r
+
1
,
s
+
1
)
1
-
x
(
r
+
1
,
s
+
1
)
2
)
)
/
4
,
for
(
even
r
,
odd
s
)
;
x
(
r
,
s
)
3
=
x
(
r
,
s
)
2
+
(
(
x
(
r
-
1
,
s
-
1
)
3
-
x
(
r
-
1
,
s
-
1
)
2
)
+
(
x
(
r
-
1
,
s
+
1
)
3
-
x
(
r
-
1
,
s
+
1
)
2
)
+
(
x
(
r
+
1
,
s
-
1
)
3
-
x
(
r
+
1
,
s
-
1
)
2
)
+
(
x
(
r
+
1
,
s
+
1
)
3
-
x
(
r
+
1
,
s
+
1
)
2
)
)
/
4
,
for
(
odd
r
,
even
s
)
;
x
(
r
,
s
)
1
=
x
(
r
,
s
)
2
+
(
(
x
(
r
-
1
,
s
)
1
-
x
(
r
-
1
,
s
)
2
)
+
(
x
(
r
,
s
-
1
)
1
-
x
(
r
,
s
-
1
)
2
)
+
(
x
(
r
,
s
+
1
)
1
-
x
(
r
,
s
+
1
)
2
)
+
(
x
(
r
+
1
,
s
)
1
-
x
(
r
+
1
,
s
)
2
)
)
/
4
,
for
(
odd
r
,
odd
s
)
and
(
even
r
,
even
s
)
;
and
x
(
r
,
s
)
3
=
x
(
r
,
s
)
2
+
(
(
x
(
r
-
1
,
s
)
3
-
x
(
r
-
1
,
s
)
2
)
+
(
x
(
r
,
s
-
1
)
3
-
x
(
r
,
s
-
1
)
2
)
+
(
x
(
r
,
s
+
1
)
3
-
x
(
r
,
s
+
1
)
2
)
+
(
x
(
r
+
1
,
s
)
3
-
x
(
r
+
1
,
s
)
2
)
)
/
4
,
for
(
odd
r
,
odd
s
)
and
(
even
r
,
even
s
)
.
6. The image processing apparatus as recited in claim 5 , wherein the directional chrominance component interpolation module is configured to use the edge directional map d (r,s) to direct the interpolated chrominance component enhancement process along the edges within the image z according to the following equations:
x
(
r
,
s
)
1
=
{
x
(
r
,
s
)
2
+
(
x
(
r
-
1
,
s
)
1
-
x
(
r
-
1
,
s
)
2
+
x
(
r
+
1
,
s
)
1
-
x
(
r
+
1
,
s
)
2
)
/
2
if
D
<
2
x
(
r
,
s
)
2
+
(
x
(
r
,
s
-
1
)
1
-
x
(
r
,
s
-
1
)
2
+
x
(
r
,
s
+
1
)
1
-
x
(
r
,
s
+
1
)
2
)
/
2
if
D
>
2
x
(
r
,
s
)
1
otherwise
x
(
r
,
s
)
3
=
{
x
(
r
,
s
)
2
+
(
x
(
r
-
1
,
s
)
3
-
x
(
r
-
1
,
s
)
2
+
x
(
r
+
1
,
s
)
3
-
x
(
r
+
1
,
s
)
2
)
/
2
if
D
<
2
x
(
r
,
s
)
2
+
(
x
(
r
,
s
-
1
)
3
-
x
(
r
,
s
-
1
)
2
+
x
(
r
,
s
+
1
)
3
-
x
(
r
,
s
+
1
)
2
)
/
2
if
D
>
2
x
(
r
,
s
)
3
otherwise
where
D
=
(
d
(
r
-
1
,
s
)
+
d
(
r
,
s
-
1
)
+
d
(
r
,
s
+
1
)
+
d
(
r
+
1
,
s
)
)
/
4
or
D
=
d
(
r
,
s
)
.
7. The image processing apparatus as recited in claim 2 , wherein the edge direction detection module is configured to evaluate the variations in the enhanced luminance-chrominance difference components {circumflex over (Δ)} v and {circumflex over (Δ)} h in order to create an edge directional map indicating the direction in which demosaicking should be performed according to the following equation:
d
(
r
,
s
)
=
{
1
for
υ
(
r
,
s
)
v
<
υ
(
r
,
s
)
h
and
υ
(
r
,
s
)
v
>
Φ
3
for
υ
(
r
,
s
)
v
>
υ
(
r
,
s
)
h
and
υ
(
r
,
s
)
h
>
Φ
2
otherwise
where:
Φ is a predetermined variation constraint;
υ
(
r
,
s
)
v
=
∑
i
=
-
1
1
(
(
Δ
^
(
r
-
1
,
s
+
i
)
v
-
Δ
^
(
r
,
s
+
i
)
v
)
2
+
(
Δ
^
(
r
,
s
+
i
)
v
-
Δ
^
(
r
+
1
,
s
+
i
)
v
)
2
+
(
Δ
^
(
r
-
1
,
s
+
i
)
v
-
Δ
^
(
r
+
1
,
s
+
i
)
v
)
2
)
;
υ
(
r
,
s
)
h
=
∑
i
=
-
1
1
(
(
Δ
^
(
r
+
i
,
s
-
1
)
h
-
Δ
^
(
r
+
i
,
s
)
h
)
2
+
(
Δ
^
(
r
+
i
,
s
)
h
-
Δ
^
(
r
+
i
,
s
+
1
)
h
)
2
+
(
Δ
^
(
r
+
i
,
s
-
1
)
h
-
Δ
^
(
r
+
i
,
s
+
1
)
h
)
2
)
;
and
the d (r,s) value indicates that demosaicking should be performed in the vertical direction (for d (r,s) =1), the horizontal direction (for d (r,s) =3), or using samples located in both the vertical and horizontal directions (for d (r,s) =2).
8. The image processing apparatus as recited in claim 7 , wherein the luminance component interpolation module is configured to interpolate a luminance component for each of the original R and B components using the edge directional map d (r,s) according to the following equation:
x
(
r
,
s
)
2
=
{
z
(
r
,
s
)
+
(
Δ
^
(
r
-
1
,
s
)
v
+
Δ
^
(
r
+
1
,
s
)
v
)
/
2
for
d
(
r
,
s
)
=
1
z
(
r
,
s
)
+
(
Δ
^
(
r
,
s
-
1
)
h
+
Δ
^
(
r
,
s
+
1
)
h
)
/
2
for
d
(
r
,
s
)
=
3
z
(
r
,
s
)
+
υ
(
r
,
s
)
h
(
Δ
^
(
r
-
1
,
s
)
v
+
Δ
^
(
r
+
1
,
s
)
v
)
/
2
+
υ
(
r
,
s
)
v
(
Δ
^
(
r
,
s
-
1
)
h
+
Δ
^
(
r
,
s
+
1
)
h
)
/
2
υ
(
r
,
s
)
h
+
υ
(
r
,
s
)
v
otherwise
.
9. The image processing apparatus as recited in claim 1 , further comprising a clamping module configured to detect a signal overshoot or undershoot in each interpolated G component and to clamp each interpolated G component with a detected signal overshoot or undershoot to the closest neighboring original G component in the direction in which demosaicking is being performed.
10. A method for demosaicking a mosaic input image to generate a full color output image, the method comprising the steps of using a processing unit to:
a) calculate both vertical and horizontal luminance-chrominance difference components for each pixel of the mosaic input image;
b) calculate an enhanced version of both the vertical and horizontal luminance-chrominance difference components for each pixel of the mosaic input image;
c) evaluate the variations in the enhanced luminance-chrominance difference components in order to create an edge directional map indicating the direction in which demosaicking should be performed;
d) interpolate a G component for each of the original R and B components using the edge directional map;
e) interpolate an R component for each of the original G and B components and a B component for each of the original R and G components; and
f) enhance the interpolated R and B components using the edge directional map; and
wherein the step a) comprises calculating both vertical and horizontal luminance-chrominance difference components Δ v and Δ h for each pixel (r,s) of the mosaic input image z according to the following equations:
Δ
(
r
,
s
)
v
=
{
z
(
r
,
s
)
-
(
z
(
r
-
1
,
s
)
+
z
(
r
+
1
,
s
)
)
/
2
for
(
odd
r
,
odd
s
)
and
(
even
r
,
even
s
)
(
z
(
r
-
1
,
s
)
+
z
(
r
+
1
,
s
)
)
/
2
-
z
(
r
,
s
)
otherwise
;
and
Δ
(
r
,
s
)
h
=
{
z
(
r
,
s
)
-
(
z
(
r
,
s
-
1
)
+
z
(
r
,
s
+
1
)
)
/
2
for
(
odd
r
,
odd
s
)
and
(
even
r
,
even
s
)
(
z
(
r
,
s
-
1
)
+
z
(
r
,
s
+
1
)
)
/
2
-
z
(
r
,
s
)
otherwise
.
11. The method as recited in claim 10 , wherein the step b) comprises calculating an enhanced version {circumflex over (Δ)} v and {circumflex over (Δ)} h of both the vertical luminance-chrominance difference components Δ v and Δ h for each pixel (r, s) of the mosaic input image z according to the following equations:
Δ
^
(
r
,
s
)
v
=
(
Δ
(
r
-
2
,
s
)
v
+
2
Δ
(
r
-
1
,
s
)
v
+
4
Δ
(
r
,
s
)
v
+
2
Δ
(
r
+
1
,
s
)
v
+
Δ
(
r
+
2
,
s
)
v
)
/
10
;
and
Δ
^
(
r
,
s
)
h
=
(
Δ
(
r
,
s
-
2
)
h
+
2
Δ
(
r
,
s
-
1
)
h
+
4
Δ
(
r
,
s
)
h
+
2
Δ
(
r
,
s
+
1
)
h
+
Δ
(
r
,
s
+
2
)
h
)
/
10.
12. The method as recited in claim 11 , wherein the step c) comprises evaluating the variations in the enhanced luminance-chrominance difference components {circumflex over (Δ)} v and {circumflex over (Δ)} h in order to create an edge directional map indicating the direction in which demosaicking should be performed according to the following equations:
d
(
r
,
s
)
=
{
1
for
υ
(
r
,
s
)
v
<
υ
(
r
,
s
)
h
3
for
υ
(
r
,
s
)
v
>
υ
(
r
,
s
)
h
2
otherwise
where
:
υ
(
r
,
s
)
v
=
∑
i
=
-
1
1
(
(
Δ
^
(
r
-
1
,
s
+
i
)
v
-
Δ
^
(
r
,
s
+
i
)
v
)
2
+
(
Δ
^
(
r
,
s
+
i
)
v
-
Δ
^
(
r
+
1
,
s
+
i
)
v
)
2
+
(
Δ
^
(
r
-
1
,
s
+
i
)
v
-
Δ
^
(
r
+
1
,
s
+
i
)
v
)
2
)
;
υ
(
r
,
s
)
h
=
∑
i
=
-
1
1
(
(
Δ
^
(
r
+
i
,
s
-
1
)
h
-
Δ
^
(
r
+
i
,
s
)
h
)
2
+
(
Δ
^
(
r
+
i
,
s
)
h
-
Δ
^
(
r
+
i
,
s
+
1
)
h
)
2
+
(
Δ
^
(
r
+
i
,
s
-
1
)
h
-
Δ
^
(
r
+
i
,
s
+
1
)
h
)
2
)
;
and
the d (r,s) value indicates that demosaicking should be performed in the vertical direction (for d (r,s) =1), the horizontal direction (for d (r,s)= 3), or using samples located in both the vertical and horizontal directions (for d (r,s) =2).
13. The method as recited in claim 12 , wherein the step d) comprises interpolating a G component for each of the original R and B components using the edge directional map according to the following equation:
x
(
r
,
s
)
2
=
{
z
(
r
,
s
)
+
(
Δ
^
(
r
-
1
,
s
)
v
+
Δ
^
(
r
+
1
,
s
)
v
)
/
2
for
d
(
r
,
s
)
=
1
z
(
r
,
s
)
+
(
Δ
^
(
r
,
s
-
1
)
h
+
Δ
^
(
r
,
s
+
1
)
h
)
/
2
for
d
(
r
,
s
)
=
3
z
(
r
,
s
)
+
(
Δ
^
(
r
-
1
,
s
)
v
+
Δ
^
(
r
+
1
,
s
)
v
+
Δ
^
(
r
,
s
-
1
)
h
+
Δ
^
(
r
,
s
+
1
)
h
)
/
4
otherwise
.
14. The method as recited in claim 13 , wherein the step e) comprises interpolating an R component for each of the original G and B components and a B component for each of the original R and G components according to the following equations:
x
(
r
,
s
)
1
=
x
(
r
,
s
)
2
+
(
(
x
(
r
-
1
,
s
-
1
)
1
-
x
(
r
-
1
,
s
-
1
)
2
)
+
(
x
(
r
-
1
,
s
+
1
)
1
-
x
(
r
-
1
,
s
+
1
)
2
)
+
(
x
(
r
+
1
,
s
-
1
)
1
-
x
(
r
+
1
,
s
-
1
)
2
)
+
(
x
(
r
+
1
,
s
+
1
)
1
-
x
(
r
+
1
,
s
+
1
)
2
)
)
/
4
,
for
(
even
r
,
odd
s
)
;
x
(
r
,
s
)
3
=
x
(
r
,
s
)
2
+
(
(
x
(
r
-
1
,
s
-
1
)
3
-
x
(
r
-
1
,
s
-
1
)
2
)
+
(
x
(
r
-
1
,
s
+
1
)
3
-
x
(
r
-
1
,
s
+
1
)
2
)
+
(
x
(
r
+
1
,
s
-
1
)
3
-
x
(
r
+
1
,
s
-
1
)
2
)
+
(
x
(
r
+
1
,
s
+
1
)
3
-
x
(
r
+
1
,
s
+
1
)
2
)
)
/
4
,
for
(
odd
r
,
even
s
)
;
x
(
r
,
s
)
1
=
x
(
r
,
s
)
2
+
(
(
x
(
r
-
1
,
s
)
1
-
x
(
r
-
1
,
s
)
2
)
+
(
x
(
r
,
s
-
1
)
1
-
x
(
r
,
s
-
1
)
2
)
+
(
x
(
r
,
s
+
1
)
1
-
x
(
r
,
s
+
1
)
2
)
+
(
x
(
r
+
1
,
s
)
1
-
x
(
r
+
1
,
s
)
2
)
)
/
4
,
for
(
odd
r
,
odd
s
)
and
(
even
r
,
even
s
)
;
and
x
(
r
,
s
)
3
=
x
(
r
,
s
)
2
+
(
(
x
(
r
-
1
,
s
)
3
-
x
(
r
-
1
,
s
)
2
)
+
(
x
(
r
,
s
-
1
)
3
-
x
(
r
,
s
-
1
)
2
)
+
(
x
(
r
,
s
+
1
)
3
-
x
(
r
,
s
+
1
)
2
)
+
(
x
(
r
+
1
,
s
)
3
-
x
(
r
+
1
,
s
)
2
)
)
/
4
,
for
(
odd
r
,
odd
s
)
and
(
even
r
,
even
s
)
.
15. The method as recited in claim 14 , wherein the step f) comprises enhancing the interpolated R and B components using the edge directional map according to the following equations:
x
(
r
,
s
)
1
=
{
x
(
r
,
s
)
2
+
(
x
(
r
-
1
,
s
)
1
-
x
(
r
-
1
,
s
)
2
+
x
(
r
+
1
,
s
)
1
-
x
(
r
+
1
,
s
)
2
)
/
2
if
D
<
2
x
(
r
,
s
)
2
+
(
x
(
r
,
s
-
1
)
1
-
x
(
r
,
s
-
1
)
2
+
x
(
r
,
s
+
1
)
1
-
x
(
r
,
s
+
1
)
2
)
/
2
if
D
>
2
x
(
r
,
s
)
1
otherwise
x
(
r
,
s
)
3
=
{
x
(
r
,
s
)
2
+
(
x
(
r
-
1
,
s
)
3
-
x
(
r
-
1
,
s
)
2
+
x
(
r
+
1
,
s
)
3
-
x
(
r
+
1
,
s
)
2
)
/
2
if
D
<
2
x
(
r
,
s
)
2
+
(
x
(
r
,
s
-
1
)
3
-
x
(
r
,
s
-
1
)
2
+
x
(
r
,
s
+
1
)
3
-
x
(
r
,
s
+
1
)
2
)
/
2
if
D
>
2
x
(
r
,
s
)
3
otherwise
where
D
=
(
d
(
r
-
1
,
s
)
+
d
(
r
,
s
-
1
)
+
d
(
r
,
s
+
1
)
+
d
(
r
+
1
,
s
)
)
/
4
or
D
=
d
(
r
,
s
)
.
16. The method as recited in claim 11 , wherein the step c) comprises evaluating the variations in the enhanced luminance-chrominance difference components {circumflex over (Δ)} v and {circumflex over (Δ)} h in order to create an edge directional map indicating the direction in which demosaicking should be performed according to the following equations:
d
(
r
,
s
)
=
{
1
for
υ
(
r
,
s
)
v
<
υ
(
r
,
s
)
h
and
υ
(
r
,
s
)
v
>
Φ
3
for
υ
(
r
,
s
)
v
>
υ
(
r
,
s
)
h
and
υ
(
r
,
s
)
h
>
Φ
2
otherwise
where:
Φ is a predetermined variation constraint;
υ
(
r
,
s
)
v
=
∑
i
=
-
1
1
(
(
Δ
^
(
r
-
1
,
s
+
i
)
v
-
Δ
^
(
r
,
s
+
i
)
v
)
2
+
(
Δ
^
(
r
,
s
+
i
)
v
-
Δ
^
(
r
+
1
,
s
+
i
)
v
)
2
+
(
Δ
^
(
r
-
1
,
s
+
i
)
v
-
Δ
^
(
r
+
1
,
s
+
i
)
v
)
2
)
;
υ
(
r
,
s
)
h
=
∑
i
=
-
1
1
(
(
Δ
^
(
r
+
i
,
s
-
1
)
h
-
Δ
^
(
r
+
i
,
s
)
h
)
2
+
(
Δ
^
(
r
+
i
,
s
)
h
-
Δ
^
(
r
+
i
,
s
+
1
)
h
)
2
+
(
Δ
^
(
r
+
i
,
s
-
1
)
h
-
Δ
^
(
r
+
i
,
s
+
1
)
h
)
2
)
;
and
the d (r,s) value indicates that demosaicking should be performed in the vertical direction (for d (r,s) =1), the horizontal direction (for d (r,s) =3), or using samples located in both the vertical and horizontal directions (for d (r,s) =2).
17. The method as recited in claim 16 , wherein the step d) comprises interpolating a G component for each of the original R and B components using the edge directional map according to the following equation:
x
(
r
,
s
)
2
=
{
z
(
r
,
s
)
+
(
Δ
^
(
r
-
1
,
s
)
v
+
Δ
^
(
r
+
1
,
s
)
v
)
/
2
for
d
(
r
,
s
)
=
1
z
(
r
,
s
)
+
(
Δ
^
(
r
,
s
-
1
)
h
+
Δ
^
(
r
,
s
+
1
)
h
)
/
2
for
d
(
r
,
s
)
=
3
z
(
r
,
s
)
+
υ
(
r
,
s
)
h
(
Δ
^
(
r
-
1
,
s
)
v
+
Δ
^
(
r
+
1
,
s
)
v
)
/
2
υ
(
r
,
s
)
v
(
Δ
^
(
r
,
s
-
1
)
h
+
Δ
^
(
r
,
s
+
1
)
h
)
/
2
υ
(
r
,
s
)
h
+
υ
(
r
,
s
)
v
otherwise
.
18. The method as recited in claim 10 , further comprising the following steps after step d) and before step e):
d1) detecting a signal overshoot or undershoot in each interpolated G component; and
d2) clamping each interpolated G component with a detected signal overshoot or undershoot to the closest neighboring original G component in the direction in which demosaicking is being performed.
19. One or more non-transitory computer-readable media having computer-readable instructions thereon which, when executed, implement a method for demosaicking a mosaic input image subject to generate a full color output image, the method comprising the steps of:
a) calculating both vertical and horizontal luminance-chrominance difference components for each pixel of the mosaic input image;
b) calculating an enhanced version of both the vertical and horizontal luminance-chrominance difference components for each pixel of the mosaic input image;
c) evaluating the variations in the enhanced luminance-chrominance difference components in order to create an edge directional map indicating the direction in which demosaicking should be performed;
d) interpolating a G component for each of the original R and B components using the edge directional map;
e) interpolating an R component for each of the original G and B components and a B component for each of the original R and G components; and
f) enhancing the interpolated R and B components by directing a chrominance demosaicking process along the edges within the image using the edge directional map; and
wherein the step a) comprises calculating both vertical and horizontal luminance-chrominance difference components Δ v and Δ h for each pixel (r,s) of the mosaic input image z according to the following equations:
Δ
(
r
,
s
)
v
=
{
z
(
r
,
s
)
-
(
z
(
r
-
1
,
s
)
+
z
(
r
+
1
,
s
)
)
/
2
for
(
odd
r
,
odd
s
)
and
(
even
r
,
even
s
)
(
z
(
r
-
1
,
s
)
+
z
(
r
+
1
,
s
)
)
/
2
-
z
(
r
,
s
)
otherwise
;
and
Δ
(
r
,
s
)
h
=
{
z
(
r
,
s
)
-
(
z
(
r
,
s
-
1
)
+
z
(
r
,
s
+
1
)
)
/
2
for
(
odd
r
,
odd
s
)
and
(
even
r
,
even
s
)
(
z
(
r
,
s
-
1
)
+
z
(
r
,
s
+
1
)
)
/
2
-
z
(
r
,
s
)
otherwise
.
20. The one or more non-transitory computer-readable media as recited in claim 19 , wherein:
the step b) comprises calculating an enhanced version {circumflex over (Δ)} v and {circumflex over (Δ)} h of both the vertical luminance-chrominance difference components Δ v and Δ h for each pixel (r,s) of the mosaic input image z according to the following equations:
Δ
^
(
r
,
s
)
v
=
(
Δ
(
r
-
2
,
s
)
v
+
2
Δ
(
r
-
1
,
s
)
v
+
4
Δ
(
r
,
s
)
v
+
2
Δ
(
r
+
1
,
s
)
v
+
Δ
(
r
+
2
,
s
)
v
)
/
10
;
and
Δ
^
(
r
,
s
)
h
=
(
Δ
(
r
,
s
-
2
)
h
+
2
Δ
(
r
,
s
-
1
)
h
+
4
Δ
(
r
,
s
)
h
+
2
Δ
(
r
,
s
+
1
)
h
+
Δ
(
r
,
s
+
2
)
h
)
/
10.
21. The one or more non-transitory computer-readable media as recited in claim 20 , wherein the step c) comprises evaluating the variations in the enhanced luminance-chrominance difference components {circumflex over (Δ)} v and {circumflex over (Δ)} h in order to create an edge directional map indicating the direction in which demosaicking should be performed according to the following equations:
d
(
r
,
s
)
=
{
1
for
υ
(
r
,
s
)
v
<
υ
(
r
,
s
)
h
3
for
υ
(
r
,
s
)
v
>
υ
(
r
,
s
)
h
2
otherwise
where
:
υ
(
r
,
s
)
v
=
∑
i
=
-
1
v
(
(
Δ
^
(
r
-
1
,
s
+
i
)
v
-
Δ
^
(
r
,
s
+
i
)
v
)
2
+
(
Δ
^
(
r
,
s
+
i
)
v
-
Δ
^
(
r
+
1
,
s
+
i
)
v
)
2
+
(
Δ
^
(
r
-
1
,
s
+
i
)
v
-
Δ
^
(
r
+
1
,
s
+
i
)
v
)
2
)
;
υ
(
r
,
s
)
h
=
∑
i
=
-
1
1
(
(
Δ
^
(
r
+
i
,
s
-
1
)
h
-
Δ
^
(
r
+
i
,
s
)
h
)
2
+
(
Δ
^
(
r
+
i
,
s
)
h
-
Δ
^
(
r
+
i
,
s
+
1
)
h
)
2
+
(
Δ
^
(
r
+
i
,
s
-
1
)
h
-
Δ
^
(
r
+
i
,
s
+
1
)
h
)
2
)
;
and
the d (r,s) value indicates that demosaicking should be performed in the vertical direction (for d (r,s) =1), the horizontal direction (for d (r,s) =3), or using samples located in both the vertical and horizontal directions (for d (r,s) =2).
22. The one or more non-transitory computer-readable media as recited in claim 21 , wherein:
the step d) comprises interpolating a G component for each of the original R and B components using the edge directional map according to the following equation:
x
(
r
,
s
)
2
=
{
z
(
r
,
s
)
+
(
Δ
^
(
r
-
1
,
s
)
v
+
Δ
^
(
r
+
1
,
s
)
v
)
/
2
for
d
(
r
,
s
)
=
1
z
(
r
,
s
)
+
(
Δ
^
(
r
,
s
-
1
)
h
+
Δ
^
(
r
,
s
+
1
)
h
)
/
2
for
d
(
r
,
s
)
=
3
z
(
r
,
s
)
+
(
Δ
^
(
r
-
1
,
s
)
v
+
Δ
^
(
r
+
1
,
s
)
v
+
Δ
^
(
r
,
s
-
1
)
h
+
Δ
^
(
r
,
s
+
1
)
h
)
/
4
otherwise
.
;
and
the step e) comprises interpolating an R component for each of the original G and B components and a B component for each of the original R and G components according to the following equations:
x
(
r
,
s
)
1
=
x
(
r
,
s
)
2
+
(
(
x
(
r
-
1
,
s
-
1
)
1
-
x
(
r
-
1
,
s
-
1
)
2
)
+
(
x
(
r
-
1
,
s
+
1
)
1
-
x
(
r
-
1
,
s
+
1
)
2
)
+
(
x
(
r
+
1
,
s
-
1
)
1
-
x
(
r
+
1
,
s
-
1
)
2
)
+
(
x
(
r
+
1
,
s
+
1
)
1
-
x
(
r
+
1
,
s
+
1
)
2
)
)
/
4
,
for
(
even
r
,
odd
s
)
;
x
(
r
,
s
)
3
=
x
(
r
,
s
)
2
+
(
(
x
(
r
-
1
,
s
-
1
)
3
-
x
(
r
-
1
,
s
-
1
)
2
)
+
(
x
(
r
-
1
,
s
+
1
)
3
-
x
(
r
-
1
,
s
+
1
)
2
)
+
(
x
(
r
+
1
,
s
-
1
)
3
-
x
(
r
+
1
,
s
-
1
)
2
)
+
(
x
(
r
+
1
,
s
+
1
)
3
-
x
(
r
+
1
,
s
+
1
)
2
)
)
/
4
,
for
(
odd
r
,
even
s
)
;
x
(
r
,
s
)
1
=
x
(
r
,
s
)
2
+
(
(
x
(
r
-
1
,
s
)
1
-
x
(
r
-
1
,
s
)
2
)
+
(
x
(
r
,
s
-
1
)
1
-
x
(
r
,
s
-
1
)
2
)
+
(
x
(
r
,
s
+
1
)
1
-
x
(
r
,
s
+
1
)
2
)
+
(
x
(
r
+
1
,
s
)
1
-
x
(
r
+
1
,
s
)
2
)
)
/
4
,
for
(
odd
r
,
odd
s
)
and
(
even
r
,
even
s
)
;
and
x
(
r
,
s
)
3
=
x
(
r
,
s
)
2
+
(
(
x
(
r
-
1
,
s
)
3
-
x
(
r
-
1
,
s
)
2
)
+
(
x
(
r
,
s
-
1
)
3
-
x
(
r
,
s
-
1
)
2
)
+
(
x
(
r
,
s
+
1
)
3
-
x
(
r
,
s
+
1
)
2
)
+
(
x
(
r
+
1
,
s
)
3
-
x
(
r
+
1
,
s
)
2
)
)
/
4
,
for
(
odd
r
,
odd
s
)
and
(
even
r
,
even
s
)
.
23. The one or more non-transitory computer-readable media as recited in claim 22 , wherein the step f) comprises enhancing the interpolated R and B components using the edge directional map according to the following equations:
x
(
r
,
s
)
1
=
{
x
(
r
,
s
)
2
+
(
x
(
r
-
1
,
s
)
1
-
x
(
r
-
1
,
s
)
2
+
x
(
r
+
1
,
s
)
2
-
x
(
r
+
1
,
s
)
2
)
/
2
if
D
<
2
x
(
r
,
s
)
2
+
(
x
(
r
,
s
-
1
)
1
-
x
(
r
,
s
-
1
)
2
+
x
(
r
,
s
+
1
)
1
-
x
(
r
,
s
+
1
)
2
)
/
2
if
D
>
2
x
(
r
,
s
)
1
otherwise
x
(
r
,
s
)
3
=
{
x
(
r
,
s
)
2
+
(
x
(
r
-
1
,
s
)
3
-
x
(
r
-
1
,
s
)
2
+
x
(
r
+
1
,
s
)
3
-
x
(
r
+
1
,
s
)
2
)
/
2
if
D
<
2
x
(
r
,
s
)
2
+
(
x
(
r
,
s
-
1
)
3
-
x
(
r
,
s
-
1
)
2
+
x
(
r
,
s
+
1
)
3
-
x
(
r
,
s
+
1
)
2
)
/
2
if
D
>
2
x
(
r
,
s
)
3
otherwise
where
D
=
(
d
(
r
-
1
,
s
)
+
d
(
r
,
s
-
1
)
+
d
(
r
,
s
+
1
)
+
d
(
r
+
1
,
s
)
)
/
4
or
D
=
d
(
r
,
s
)
.
24. The one or more non-transitory computer-readable media as recited in claim 20 , wherein the step c) comprises evaluating the variations in the enhanced luminance-chrominance difference components {circumflex over (Δ)} v and {circumflex over (Δ)} h in order to create an edge directional map indicating the direction in which demosaicking should be performed according to the following equations:
d
(
r
,
s
)
=
{
1
for
υ
(
r
,
s
)
v
<
υ
(
r
,
s
)
h
and
υ
(
r
,
s
)
v
>
Φ
3
for
υ
(
r
,
s
)
v
>
υ
(
r
,
s
)
h
and
υ
(
r
,
s
)
h
>
Φ
2
otherwise
where:
Φ is a predetermined variation constraint;
υ
(
r
,
s
)
v
=
∑
i
=
-
1
1
(
(
Δ
^
(
r
-
1
,
s
+
i
)
v
-
Δ
^
(
r
,
s
+
i
)
v
)
2
+
(
Δ
^
(
r
,
s
+
i
)
v
-
Δ
^
(
r
+
1
,
s
+
i
)
v
)
2
+
(
Δ
^
(
r
-
1
,
s
+
i
)
v
-
Δ
^
(
r
+
1
,
s
+
i
)
v
)
2
)
;
υ
(
r
,
s
)
h
=
∑
i
=
-
1
1
(
(
Δ
^
(
r
+
i
,
s
-
1
)
h
-
Δ
^
(
r
+
i
,
s
)
h
)
2
+
(
Δ
^
(
r
+
i
,
s
)
h
-
Δ
^
(
r
+
i
,
s
+
1
)
h
)
2
+
(
Δ
^
(
r
+
i
,
s
-
1
)
h
-
Δ
^
(
r
+
i
,
s
+
1
)
h
)
2
)
;
and
the d (r,s) value indicates that demosaicking should be performed in the vertical direction (for d (r,s) =1), the horizontal direction (for d (r,s) =3), or using samples located in both the vertical and horizontal directions (for d (r,s) =2).
25. The one or more non-transitory computer-readable media as recited in claim 24 , wherein:
the step d) comprises interpolating a G component for each of the original R and B components using the edge directional map according to the following equation:
x
(
r
,
s
)
2
=
{
z
(
r
,
s
)
+
(
Δ
^
(
r
-
1
,
s
)
v
+
Δ
^
(
r
+
1
,
s
)
v
)
/
2
for
d
(
r
,
s
)
=
1
z
(
r
,
s
)
+
(
Δ
^
(
r
,
s
-
1
)
h
+
Δ
^
(
r
,
s
+
1
)
h
)
/
2
for
d
(
r
,
s
)
=
1
z
(
r
,
s
)
+
υ
(
r
,
s
)
h
(
Δ
^
(
r
-
1
,
s
)
v
+
Δ
^
(
r
+
1
,
s
)
v
)
/
2
υ
(
r
,
s
)
v
(
Δ
^
(
r
,
s
-
1
)
h
+
Δ
^
(
r
,
s
+
1
)
h
)
/
2
υ
(
r
,
s
)
h
+
υ
(
r
,
s
)
v
otherwise
.
26. The one or more non-transitory computer-readable media as recited in claim 19 , further comprising the following steps after step d) and before step e):
d1) detecting a signal overshoot or undershoot in each interpolated G component; and
d2) clamping each interpolated G component with a detected signal overshoot or undershoot to the closest neighboring original G component in the direction in which demosaicking is being performed.