Apparatus and method for restoring image based on distance-specific point spread function

1. An apparatus for restoring an image, comprising:

an edge-strength-detection module which detects a red (R)-channel edge strength, a green (G)-channel edge strength and a blue (B)-channel edge strength from an input image and generates an R-channel edge map, a G-channel edge map and a B-channel edge map based on the R-channel edge strength, the G-channel edge strength and the B-channel edge strength;

an edge-strength-compensation module which compensates for the R-channel edge strength and the B-channel edge strength based on the G-channel edge strength;

a valid-value-generation module which generates a valid value indicating a blur level based on the compensated R-channel edge strength and the compensated B-channel edge strength;

a point spread function (PSF) estimation module which estimates a PSF corresponding to the valid value; and

an image-restoration module which restores the input image using the PSF.

2. The apparatus of claim 1 , wherein the edge-strength-detection module detects the R-channel edge strength, the G-channel edge strength and the B-channel edge strength using a high-pass filter.

3. The apparatus of claim 2 , wherein the high-pass filter comprises a Laplacian filter.

4. The apparatus of claim 1 , wherein the edge-strength-compensation module compensates for the R-channel edge strength and the B-channel edge strength using a weight which is generated based on the G-channel edge strength.

5. The apparatus of claim 4 , wherein the edge-strength-compensation module obtains the compensated R-channel edge strength or the compensated B-channel edge strength by multiplying the R-channel edge strength or the B-channel edge strength by the weight.

6. The apparatus of claim 1 , wherein the valid value is a ratio of the sum of compensated R-channel edge strengths and the sum of compensated B-channel edge strengths.

7. The apparatus of claim 1 , wherein the PSF-estimation module compares the valid value with a plurality of default valid values, determines a distance between a subject of the input image and an image sensor based on a default valid value that matches the valid value, determines which of R, G and B channels is sharpest based on the distance between the subject of the input image and the image sensor, and estimates a PSF corresponding to whichever of the R, G and B channels is determined to be sharpest, the default valid values being respectively obtained from a plurality of images of subjects at different distances from the image sensor.

8. The apparatus of claim 1 , wherein the image-restoration module restores the input image by applying a deconvolution filter to the PSF and thus removing blurs in the input image.

9. The apparatus of claim 8 , wherein the deconvolution filter comprises a Wiener filter or a Lucy-Richardson filter.

10. An image-capturing system comprising:

a photographing device which photographs an image of a subject;

a driving circuit which adjusts a lens and a diaphragm in order for the photographing device to photograph an image of the subject; and

an apparatus for restoring an image which removes blurs in the image photographed by the photographing device,

wherein the apparatus comprises:

an edge-strength-detection module which detects an R-channel edge strength, a G-channel edge strength and a B-channel edge strength from an input image, and generates an R-channel edge map, a G-channel edge map and a B-channel edge map based on the R-channel edge strength, the G-channel edge strength and the B-channel edge strength;

an edge-strength-compensation module which compensates for the R-channel edge strength and the B-channel edge strength based on the G-channel edge strength;

a valid-value-generation module which generates a valid value indicating a blur level based on the compensated R-channel edge strength and the compensated B-channel edge strength;

a PSF-estimation module which estimates a PSF corresponding to the valid value; and

an image-restoration module which restores the input image using the PSF.

11. A method of restoring an image, comprising:

detecting an R-channel edge strength, a G-channel edge strength and a B-channel edge strength from an input image and generating an R-channel edge map, a G-channel edge map and a B-channel edge map based on the R-channel edge strength, the G-channel edge strength and the B-channel edge strength;

compensating for the R-channel edge strength and the B-channel edge strength based on the G-channel edge strength;

generating a valid value indicating a blur level based on the compensated R-channel edge strength and the compensated B-channel edge strength;

estimating a PSF corresponding to the valid value; and

restoring, by way of a processor, the input image using the PSF.

12. The method of claim 11 , wherein the detecting the R-channel edge strength, the G-channel edge strength and the B-channel edge strength, comprises detecting the R-channel edge strength, the G-channel edge strength and the B-channel edge strength using a high-pass filter.

13. The method of claim 12 , wherein the high-pass filter comprises a Laplacian filter.

14. The method of claim 11 , wherein the compensating for the R-channel edge strength and the B-channel edge strength, compensating for the R-channel edge strength or the B-channel edge strength using a weight which is generated based on the G-channel edge strength.

15. The method of claim 14 , wherein the compensating for the R-channel edge strength and the B-channel edge strength further comprises obtaining the compensated R-channel edge strength or the compensated B-channel edge strength by multiplying the R-channel edge strength or the B-channel edge strength by the weight.

16. The method of claim 11 , wherein the valid value is a ratio of the sum of compensated R-channel edge strengths and the sum of compensated B-channel edge strengths.

17. The method of claim 11 , wherein the estimating of the PSF comprises comparing the valid value with a plurality of default valid values, determining a distance between a subject of the input image and an image sensor based on a default valid value that matches the valid value, determining which of R, G and B channels is sharpest based on the distance between the subject of the input image and the image sensor, and estimating a PSF corresponding to whichever of the R, G and B channels is determined to be sharpest, the default valid values being respectively obtained from a plurality of images of subjects at different distances from the image sensor.

18. The method of claim 11 , wherein the restoring the input image comprises restoring the input image by applying a deconvolution filter to the PSF, and thus removing blurs in the input image.

19. The method of claim 18 , wherein the deconvolution filter comprises a Wiener filter or a Lucy-Richardson filter.