Algorithm description on non-motion blur image generation project

• US 20070127574A1
• Filed: 12/05/2005
• Published: 06/07/2007
• Est. Priority Date: 12/05/2005
• Status: Active Grant

• Current Assignee: ArcSoft, Inc.
• Original Assignee: ArcSoft, Inc.
• Inventors: Jin, Yiqing, Wu, Donghui, Yao, Tianxiang
• Granted Patent: US 8,018,999 B2
• Time in Patent Office: Days
• US Class Current: 375/240.160
• CPC Class Codes: G06T 2207/20201 Motion blur correction
• CPC Class Codes: G06T 5/50 using two or more images, e.g. averaging or subtraction
• CPC Class Codes: G06T 5/73 Deblurring; Sharpening
• CPC Class Codes: G06T 7/20 Analysis of motion motion estimation for coding, decoding, compressing or decompressing digital video signals H04N19/43, H04N19/51
• CPC Class Codes: H04N 23/68 for stable pick-up of the scene, e.g. compensating for camera body vibrations
• CPC Class Codes: H04N 23/951 by using two or more images to influence resolution, frame rate or aspect ratio

Abstract

A method for simulating an image captured at a long exposure time (“simulated image”), includes (1) capturing each of first, second, and third images at a short exposure time, (2) determining a first relative motion between the first and the second images, (3) transforming the first image to remove the first relative motion, (4) determining a second relative motion between the third and the second images, (5) transforming the third image to remove the second relative motion, and (6) combining the first, the second, and the third images to form the simulated image. Relative motions between images are determined by matching blocks at multiple resolutions to determine corresponding points between the images. Transformation to remove relative motion is determined by fitting corresponding points between the images using a minimum square error (MSE) algorithm in a random sample consensus (RANSAC) framework.

Claims

1. A method for simulating an image captured at a long exposure time (“

simulated image”

), comprising;

capturing each of a first image, a second image, and a third image at a short exposure time;

matching a first plurality of blocks between the first and the second images to determine a first plurality of motion vectors between the first and the second images;

determining a first relative motion between the first and the second images from the first plurality of relative motion vectors;

transforming the first image to remove the first relative motion;

matching a second plurality of blocks between the third and the second images to determine a second plurality of motion vectors between the third and the second images;

determining a second relative motion between the third and the second images from the second plurality of relative motion vectors;

transforming the third image to remove the second relative motion; and

combining the first, the second, and the third images linearly to form the simulated image.

2. The method of claim 1, wherein the short exposure time is at most 1/25 second.

3. The method of claim 1, where in the long exposure time is at least ⅕

second.

4. The method of claim 1, wherein the first plurality of motion vectors is determined by:

down-sampling the first and the second images to a first resolution and a second resolution lower than the first resolution;

matching blocks between the first and the second images at the second resolution;

propagating locations of matching blocks at the second resolution to the first and the second images at the first resolution; and

using the propagated locations of the matching blocks at the second resolution as starting points, matching blocks between the first and the second images at the first resolution, wherein center points of the matching blocks are corresponding points between the first and the second images.

5. The method of claim 4, wherein said transforming the first image comprises determining a transformation comprising:

6. The method of claim 5, further comprising fitting the corresponding points into the transformation using a minimum square error (MSE) algorithm incorporated into a random sample consensus (RANSAC) framework to determine the motion parameters between the first and the second images.

7. The method of claim 1, further comprising scaling the simulated image by a factor.

8. The method of claim 7, wherein said combing the first, the second, and the third images linearly comprises: