Green image correction in phase-detection image sensor remosaicing

1 . An imaging system comprising:

a phase-detection (PD) image sensor comprising a plurality of PD pixel units, each PD pixel unit comprising an upper left PD pixel, an upper right PD pixel, a lower left PD pixel, and a lower right PD pixel, wherein the upper left PD pixel, the upper right PD pixel, the lower left PD pixel, and the lower right PD pixel are covered by a microlens;

wherein the plurality of PD pixel units comprise green PD pixel units, blue PD pixel units, and red PD pixel units; and

wherein a first Bayer pattern comprises two green PD pixel units, a blue PD pixel unit, and a red PD pixel unit of the plurality of PD pixel units;

a processor configured to:

interpolate a green image from an image produced by the PD image sensor comprising the first Bayer pattern to obtain a full resolution interpolated green image including upper left PD pixels, upper right PD pixels, lower left PD pixels, and lower right PD pixels;

convolute the full resolution interpolated green image with an N×N kernel to correct the full resolution interpolated green image, the N×N kernel comprising elements a ij , wherein N is an odd integer larger than unity; and

wherein the elements a ij satisfy an equation:

∑

i

=

1

N

+

1

2

∑

j

=

1

N

+

1

2

a

2

⁢

i

-

1

,

2

⁢

j

-

1

=

∑

i

=

1

N

+

1

2

∑

j

=

1

N

-

1

2

a

2

⁢

i

-

1

,

2

⁢

j

=

∑

i

=

1

N

-

1

2

∑

j

=

1

N

+

1

2

a

2

⁢

i

,

2

⁢

j

-

1

=

∑

i

=

1

N

-

1

2

∑

j

=

1

N

-

1

2

a

2

⁢

i

,

2

⁢

j

.

2 . The imaging system of claim 1 , wherein the N×N kernel has a horizontal line of symmetry, a vertical line of symmetry, a primary diagonal line of symmetry, and a secondary diagonal of symmetry.

3 . The imaging system of claim 2 , wherein N is 3, and the N×N kernel is:

1 2 1

2 4 2

1 2 1.

4 . The imaging system of claim 2 , wherein N is 5, and the N×N kernel is:

1 2 1 2 1

2 4 4 4 2

1 4 8 4 1

2 4 4 4 2

1 2 1 2 1.

5 . The imaging system of claim 2 , wherein the N×N kernel includes negative elements.

6 . The imaging system of claim 5 , wherein N is 5, and the N×N kernel is:

−1 −1 −1 −1 −1

1 1 4 1 −1

−1 4 12 4 −1

—1 1 4 1 −1

−1 −1 −1 −1 −1.

7 . The imaging system of claim 1 , wherein the processor is further configured to:

interpolate a blue image and a red image partially from the image produced by the PD image sensor comprising the first Bayer pattern and partially based on the corrected full resolution interpolated green image; and

form a second Bayer image using the interpolated blue image and red image and part of the corrected full resolution interpolated green image;

wherein the second Bayer image comprises a second Bayer pattern comprising two green PD pixels, a blue PD pixel, and a red PD pixel.

8 . A method for remosaicing a second Bayer image from a first Bayer image, the first Bayer image comprising green pixel units, blue pixel units, and red pixel units, wherein each pixel unit comprises an upper left PD pixel, an upper right PD pixel, a lower left PD pixel, and a lower right PD pixel, wherein the upper left PD pixel, the upper right PD pixel, the lower left PD pixel, and the lower right PD pixel are covered by a microlens, and wherein the first Bayer image comprises a first Bayer pattern comprising two green PD pixel units, a blue PD pixel unit, and a red PD pixel unit, comprising:

interpolating a green image from the first Bayer image to obtain a full resolution interpolated green image including defocused portions having artifacts and in-focus portions having sharp image, wherein the full resolution interpolated green image includes upper left PD pixels, upper right PD pixels, lower left PD pixels, and lower right PD pixels;

correcting the full resolution interpolated green image by convoluting the full resolution interpolated green image with an N×N kernel, wherein N is an odd integer larger than unity;

wherein the N×N kernel comprises elements a ij , and the elements aj satisfy an equation:

∑

i

=

1

N

+

1

2

∑

j

=

1

N

+

1

2

a

2

⁢

i

-

1

,

2

⁢

j

-

1

=

∑

i

=

1

N

+

1

2

∑

j

=

1

N

-

1

2

a

2

⁢

i

-

1

,

2

⁢

j

=

∑

i

=

1

N

-

1

2

∑

j

=

1

N

+

1

2

a

2

⁢

i

,

2

⁢

j

-

1

=

∑

i

=

1

N

-

1

2

∑

j

=

1

N

-

1

2

a

2

⁢

i

,

2

⁢

j

,

wherein the artifacts of the defocused portions are removed from the corrected full resolution interpolated green image, and the sharp image of the in-focus portions is preserved in the corrected full resolution interpolated green image.

9 . The method of claim 8 , wherein the N×N kernel has a horizontal line of symmetry, a vertical line of symmetry, a primary diagonal line of symmetry, and a secondary diagonal of symmetry.

10 . The method of claim 9 , wherein N is 3, and the N×N kernel is:

1 2 1

2 4 2

1 2 1.

11 . The method of claim 9 , wherein N is 5, and the N×N kernel is:

1 2 1 2 1

2 4 4 4 2

1 4 8 4 1

2 4 4 4 2

1 2 1 2 1.

12 . The method of claim 9 , wherein the N×N kernel includes negative elements.

13 . The method of claim 12 , wherein N is 5, and the N×N kernel is:

−1 −1 −1 −1 −1

−1 1 4 1 −1

−1 4 12 4 −1

−1 1 4 1 −1

−1 −1 −1 −1 −1.

14 . The method of claim 8 further comprising:

interpolating a blue image and a red image partially from the first Bayer image and partially based on the corrected full resolution interpolated green image; and

forming a second Bayer image using the interpolated blue image and red image and part of the corrected full resolution interpolated green image;

wherein the second Bayer image comprises a second Bayer pattern comprising two green PD pixels, a blue PD pixel, and a red PD pixel.