Repairing scratch impairments to an image

1. Apparatus for repairing scratch impairments by modifying impaired brightness values of an input image to provide a repaired output image, the apparatus comprising: at least one scratch detection circuit configured to operate on the input image to produce a detected scratch signal; a gate control unit configured to receive a measure of unimpaired brightness and provide a gate control signal in dependence upon a predicted brightness error;

a gate configured to receive said gate control signal and said detected scratch signal and operating on said detected scratch signal to produce a gated scratch signal;

and a subtractor configured to receive the input image and said gated scratch signal and subtracting said gated scratch signal from the input image to provide a repaired output image.

2. Apparatus according to claim 1 , wherein said gate control unit receives said measure of unimpaired brightness from said or one of said scratch detection circuits.

3. Apparatus according to claim 1 , comprising a correlation unit serving to compare a relationship between impaired brightness values and corresponding modified brightness with an expected relationship measure.

4. Apparatus according to claim 3 , wherein said gate is configured to reduce the gated scratch signal as said relationship departs from said expected relationship.

5. Apparatus according to claim 3 in which:

the relationship is used to predict the expected brightness error due to a scratch for a point in the image; and

a brightness change that repairs scratch impairment at the said point is limited in dependence upon the predicted brightness error.

6. Apparatus according to claim 3 in which the relationship between the impaired brightness values and corresponding modified brightness values is represented by a mathematical model.

7. Apparatus according to claim 6 in which the parameters of the model are derived by statistical analysis of the said set of brightness values.

8. Apparatus according to claim 7 in which the model is a linear model and its parameters are derived by linear regression.

9. Apparatus according to claim 1 comprising:

a first path providing the input mage to said subtractor; and

a second path containing a first said scratch detection circuit and said gate.

10. Apparatus according to claim 9 , further comprising a noise reducer in said second path upstream of said scratch detection circuit.

11. Apparatus according to claim 9 , further comprising a third path, connected to receive the input image in parallel with said first and second paths, said third path containing a second said scratch detection circuit.

12. Apparatus according to claim 11 , wherein said third path contains a correlation unit serving to compare a relationship between impaired brightness values and corresponding modified brightness with an expected relationship measure.

13. Apparatus according to claim 11 , wherein said third path contains a vertical filter upstream of said second scratch detection circuit.

14. A method for repairing scratch impairments to an impaired input image to provide a repaired output image comprising the steps of: identifying brightness values for a set of points within a narrow region of the impaired input image that is aligned with an expected scratch direction, said points having respective impaired brightness values;

modifying brightness values for said points in a first scratch repair process, said first scratch repair process serving at each point in said set of points to replace an impaired brightness value by a modified brightness value;

evaluating a relationship measure between the impaired brightness values and corresponding modified brightness values for the said set of points;

and controlling a second scratch repair process that provides said repaired output image in dependence upon a predicted brightness error.

15. A method according to claim 14 in which:

said relationship measure is compared with an expected relationship measure for said set of scratch impaired points; and

a scratch repair that is applied to the said image region is inhibited when the result of the said comparison is unlike the expected result.

16. A method according to claim 14 in which:

said relationship measure is used to predict the expected brightness error due to a scratch for a point in the image; and

a brightness change that repairs scratch impairment at the said point is limited in dependence upon the predicted brightness error.

17. A method according to claim 14 in which the relationship between the impaired brightness values and corresponding modified brightness values is represented by a mathematical model.

18. A method according to claim 17 in which said mathematical model is a linear model and its parameters are derived by linear regression.

19. A method according to claim 14 in which the scratch repair process that provides a repaired output image creates a correction signal that is combined with the input image to create a repaired output image by the steps of:

noise reducing the input image;

applying a scratch repair process to the noise-reduced input image; and

taking the difference between the input and the output of that scratch repair process as the said scratch correction signal.

20. A non-transitory computer program product comprising: code causing programmable apparatus to implement a method for repairing scratch impairments to an impaired input image to provide a repaired output image, the method comprising the steps of:

identifying brightness values for a set of points within a narrow region of the impaired input image that is aligned with an expected scratch direction, said points having respective impaired brightness values;

modifying brightness values for said points in a first scratch repair process, said first scratch repair process serving at each point in said set of points to replace an impaired brightness value by a modified brightness value;

evaluating a measure of the relationship between the impaired brightness values and corresponding modified brightness values for the said set of points; and controlling a second scratch repair process that provides said repaired output image in dependence upon a predicted brightness error.