System and method for synthetic data generation using dead leaves images

Patent metadata

Field	Value
Publication number	US-12347110-B2
Application number	US-202218060419-A
Country	US
Kind code	B2
Filing date	Nov 30, 2022
Priority date	Sep 9, 2022
Publication date	Jul 1, 2025
Grant date	Jul 1, 2025

Abstract

A method includes obtaining a raw image in a first image domain. The method also includes determining a color distribution and an amount of variation in the raw image. The method further includes using an iterative process to generate a dead leaf image from a blank image in the first image domain. The iterative process includes adding multiple circles and multiple sticks to the blank image until the dead leaf image is filled. The iterative process also includes blurring portions of the dead leaf image during at least one iteration of the iterative process. Textures of the multiple circles are blended based on the color distribution and the amount of variation in the raw image.

First claim

What is claimed is:

1. A method comprising:
obtaining a raw image in a first image domain;
determining a color distribution and an amount of variation in the raw image; and
using an iterative process to generate a dead leaf image from a blank image in the first image domain, the iterative process comprising:
adding multiple circles and multiple sticks to the blank image until the dead leaf image is filled; and
blurring portions of the dead leaf image during at least one iteration of the iterative process;

wherein textures of the multiple circles are blended based on the color distribution and the amount of variation in the raw image.

2. The method of claim 1 , wherein colors of the multiple sticks are assigned based on the color distribution in the raw image.

3. The method of claim 1 , wherein determining the color distribution in the raw image comprises:
filtering noise in the raw image to create a filtered raw image;
determining a color histogram of the filtered raw image; and
generating a color table based on the color histogram.

4. The method of claim 3 , wherein determining the amount of variation in the raw image comprises:
converting the raw image from the first image domain to a second image domain;
dividing the raw image in the second image domain into multiple patches;
determining a range of each channel for each of the multiple patches;
determining a histogram of the raw image using the determined ranges; and
generating a range table based on the histogram.

5. The method of claim 4 , further comprising:
sampling color information from the color table to determine the color distribution in the raw image;
sampling range information from the range table to determine the amount of variation in the raw image; and
blending the textures of the multiple circles using the color distribution and the amount of variation in the raw image.

6. The method of claim 1 , wherein the textures of the multiple circles are further blended based on one or more texture maps obtained from a texture dataset.

7. The method of claim 1 , wherein blurring the portions of the dead leaf image during the at least one iteration of the iterative process comprises:
during an N th iteration, blurring at least one portion of the dead leaf image using a Gaussian filter, wherein N is a specified integer.

8. An electronic device comprising:
at least one processing device configured to:
obtain a raw image in a first image domain;
determine a color distribution and an amount of variation in the raw image; and
use an iterative process to generate a dead leaf image from a blank image in the first image domain;

wherein, in the iterative process, the at least one processing device is configured to:
add multiple circles and multiple sticks to the blank image until the dead leaf image is filled; and
blur portions of the dead leaf image during at least one iteration of the iterative process; and

wherein the at least one processing device is configured to blend textures of the multiple circles based on the color distribution and the amount of variation in the raw image.

9. The electronic device of claim 8 , wherein the at least one processing device is configured to assign colors of the multiple sticks based on the color distribution in the raw image.

10. The electronic device of claim 8 , wherein, to determine the color distribution in the raw image, the at least one processing device is configured to:
filter noise in the raw image to create a filtered raw image;
determine a color histogram of the filtered raw image; and
generate a color table based on the color histogram.

11. The electronic device of claim 10 , wherein, to determine the amount of variation in the raw image, the at least one processing device is configured to:
convert the raw image from the first image domain to a second image domain;
divide the raw image in the second image domain into multiple patches;
determine a range of each channel for each of the multiple patches;
determine a histogram of the raw image using the determined ranges; and
generate a range table based on the histogram.

12. The electronic device of claim 11 , wherein the at least one processing device is further configured to:
sample color information from the color table to determine the color distribution in the raw image;
sample range information from the range table to determine the amount of variation in the raw image; and
blend the textures of the multiple circles using the color distribution and the amount of variation in the raw image.

13. The electronic device of claim 8 , wherein the at least one processing device is configured to blend the textures of the multiple circles based on one or more texture maps obtained from a texture dataset.

14. The electronic device of claim 8 , wherein, to blur the portions of the dead leaf image during the at least one iteration of the iterative process, the at least one processing device is configured to:
during an N th iteration, blur at least one portion of the dead leaf image using a Gaussian filter, wherein N is a specified integer.

15. A non-transitory machine-readable medium containing instructions that when executed cause at least one processor of an electronic device to:
obtain a raw image in a first image domain;
determine a color distribution and an amount of variation in the raw image; and
use an iterative process to generate a dead leaf image from a blank image in the first image domain;
wherein, during the iterative process, the instructions when executed cause the at least one processor to:
add multiple circles and multiple sticks to the blank image until the dead leaf image is filled; and
blur portions of the dead leaf image during at least one iteration of the iterative process; and

wherein the instructions when executed cause the at least one processor to blend textures of the multiple circles based on the color distribution and the amount of variation in the raw image.

16. The non-transitory machine-readable medium of claim 15 , wherein the instructions when executed cause the at least one processor to assign colors of the multiple sticks based on the color distribution in the raw image.

17. The non-transitory machine-readable medium of claim 15 , wherein the instructions that when executed cause the at least one processor to determine the color distribution in the raw image comprise instructions that when executed cause the at least one processor to:
filter noise in the raw image to create a filtered raw image;
determine a color histogram of the filtered raw image; and
generate a color table based on the color histogram.

18. The non-transitory machine-readable medium of claim 17 , wherein the instructions that when executed cause the at least one processor to determine the amount of variation in the raw image comprise instructions that when executed cause the at least one processor to:
convert the raw image from the first image domain to a second image domain;
divide the raw image in the second image domain into multiple patches;
determine a range of each channel for each of the multiple patches;
determine a histogram of the raw image using the determined ranges; and
generate a range table based on the histogram.

19. The non-transitory machine-readable medium of claim 18 , further containing instructions that when executed cause the at least one processor to:
sample color information from the color table to determine the color distribution in the raw image;
sample range information

Assignees

• Samsung Electronics Co Ltd

Inventors

• Yu Mingzhao
• Nadir Zeeshan
• Sheikh Hamid R
• Glotzbach John W

Classifications

• G06T11/23

  using straight lines or curves · CPC title

• H04N23/741 Primary

  by increasing the dynamic range of the image compared to the dynamic range of the electronic image sensors · CPC title

• G06T2207/10024

  Color image · CPC title

• G06T11/40

  Filling planar surfaces by adding surface attributes, e.g. adding colours or textures · CPC title

• G06T5/40

  using histogram techniques · CPC title