# US 2010/0166332 A1 · Methods of deblurring image and recording mediums having the same recorded thereon

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2010/0166332 A1
- Document ID: US20100166332A1
- Status: Published application
- Kind: A1
- Application number: 12/580,975
- Filed: 2009-10-16
- Published: 2010-07-01
- Earliest claimed: 2008-12-31
- Assignees: Postech Academy-Industry Foundation
- Inventors: Seung-Yong Lee; Sung-Hyun Cho
- Web page: https://patentgrep.com/patent/US20100166332A1

## CPC

- G06T 5/73: Deblurring; Sharpening
- G06V 10/30: Noise filtering
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING

## Abstract

A method of deblurring an image by which blur can be easily and rapidly eliminated from one image and the quality of the image can be improved is provided. The method includes receiving a blurred image, an image estimation step of estimating a non-blurred image from the blurred image, a blur information estimation step of estimating blur information from the blurred image and the estimated non-blurred image, and a deblurring step of deblurring the blurred image based on the blurred image and the estimated blur information, wherein the image estimation step and the blur information estimation step are iteratively performed. Thus, blur can be rapidly and effectively eliminated from one image, thereby improving the quality of an image.

## Claims

### Claim 1

A method of deblurring an image, comprising:
receiving a blurred image;
an image estimation step of estimating a non-blurred image from the blurred image;
a blur information estimation step of estimating blur information from the blurred image and the estimated non-blurred image; and
a deblurring step of deblurring the blurred image based on the blurred image and the estimated blur information,
wherein the image estimation step and the blur information estimation step are iteratively performed.

### Claim 2 (depends on 1)

The method of claim 1, wherein the image estimation step comprises:
a gradient information prediction step of predicting gradient information for the blurred image; and
a first deconvolution step of performing first deconvolution based on the estimated blur information and the blurred image.

### Claim 3 (depends on 2)

The method of claim 2, wherein the gradient information prediction step comprises:
applying bidirectional filtering to the blurred image;
applying shock filtering to the bidirectional filtered image;
calculating a gradient map for the shock filtered image; and
applying a threshold value to the calculated gradient map.

### Claim 4 (depends on 3)

The method of claim 3, wherein applying the threshold value to the calculated gradient map comprises:
creating a histogram based on a direction and a size of the calculated gradient;
setting a size of a gradient capable of including as many pixels as a predetermined multiple or more of a maximum value of vertical and horizontal sizes of the blur information corresponding to each direction included in the created histogram, as a threshold value; and
applying the set threshold value to the gradient map to truncate the gradient.

### Claim 5 (depends on 4)

The method of claim 4, wherein applying the set threshold value to the gradient map to truncate the gradient comprises setting a gradient smaller than the threshold value to about zero.

### Claim 6 (depends on 4)

The method of claim 4, wherein the blur information estimation step comprises estimating the blur information based on the blurred image and the truncated gradient.

### Claim 7 (depends on 4)

The method of claim 4, wherein the blur information estimation step comprises estimating the blur information using an energy function including only image derivatives without a pixel value.

### Claim 8 (depends on 1)

The method of claim 1, wherein the image estimation step and the blur information estimation step are iteratively performed while changing resolutions of the blurred image and the estimated non-blurred image.

### Claim 9

A recording medium having a program of instructions recorded thereon, wherein the program is tangibly embodied and is readable and executable by a digital processing apparatus for deblurring a blurred image, the program performing:
receiving a blurred image;
an image estimation step of estimating a non-blurred image from the blurred image;
a blur information estimation step of estimating blur information from the blurred image and the estimated non-blurred image; and
a deblurring step of deblurring the blurred image based on the blurred image and the estimated blur information,
wherein the image estimation step and the blur information estimation step are iteratively performed.