Method of reparing scratches in digital images

• US 20050058366A1
• Filed: 09/11/2003
• Published: 03/17/2005
• Est. Priority Date: 09/11/2003
• Status: Active Grant

• Current Assignee: Primax Electronics Limited
• Original Assignee: Primax Electronics Limited
• Inventors: Yeh, Po-Chiang, Jiang, Yan-Ru, Dong, Rui-Jing
• Granted Patent: US 7,324,704 B2
• Time in Patent Office: Days
• US Class Current: 382/275
• CPC Class Codes: G06T 5/77 Retouching; Inpainting; Scratch removal

Abstract

A method of repairing scratches in a digital image includes counting the number of abnormal pixels among neighboring pixels of a selected pixel. The selected pixels that are surrounded by abnormal pixels in a number greater than a predetermined value are designed as particular pixels. An area surrounding each particular pixel is divided into a plurality of blocks. Furthermore, brightness difference between two of the blocks are calculated. Scratch pixels are found from the particular pixels based on brightness difference between blocks. Then, an area surrounded by scratch pixels are subdivided and filled up. Thereby, the scratch pixels can be precisely found and repaired without the need of hardware.

Claims

1. A method of repairing scratches in digital images, the method comprising:

calculating a brightness difference between a selected pixel and one of neighboring pixels surrounding the selected pixel;

counting the number of abnormal pixels among the neighboring pixels;

designating as particular pixels the selected pixels surrounded by the abnormal pixels, wherein the number of abnormal pixels is more than a predetermined value;

dividing an area surrounding each particular pixel into a plurality of blocks, and calculating a brightness difference between the blocks;

finding out scratch pixels from the particular pixels based on brightness difference between blocks; and

subdividing and filling up an area surrounded by scratch pixels.

2. The method of claim 1, wherein a neighboring pixel is an abnormal pixel when an absolute value of the brightness difference between the neighboring pixel and its corresponding selected pixel is greater than an abnormal standard.

3. The method of claim 1, wherein the particular pixels include scratch pixels and edge pixels.

4. The method of claim 1, wherein the area surrounding each particular pixel is divided into 4 blocks.

5. The method of claim 1, wherein the area surrounding each particular pixel is divided into 3 blocks.

6. The method of claim 1, wherein the area surrounding each particular pixel is divided into 8 blocks.

7. The method of claim 1, wherein the step of calculating brightness difference between the blocks includes estimating a mean square error of each block which has a plurality of pixels, and calculating the difference between the mean square errors.

8. The method of claim 7, wherein if the number of blocks is 4, then the difference between the mean square errors is obtained according to the following expression:

Dif=MaxVar−

Mid1Var−

Mid2Var+MinVar,

wherein Dif is a difference of the mean square errors, MaxVar is a maximum mean square error, MinVar is a minimum mean square error, and Mid1Var and Mid2Var are middles among the four mean square error values.

9. The method of claim 7, wherein if the number of blocks is 3, then the difference between the mean square errors is obtained according to the following expression:

Dif=MaxVar−

2MidVar+MinVar,

wherein Dif is a difference of the mean sqaure errors, MaxVar is a maximum mean square error, MinVar is a minimum mean square error, and MidVar is a middle among the three mean square error values.

10. The method of claim 1, wherein at the step of finding out scratch pixels from the particular pixels based on brightness difference between blocks, if the particular pixel has a Dif smaller than a difference standard, then the particular pixel is a scratch pixel, otherwise the particular pixel is a edge pixel.