Method and apparatus with image processing

Patent metadata

Field	Value
Publication number	US-11538139-B2
Application number	US-202016951290-A
Country	US
Kind code	B2
Filing date	Nov 18, 2020
Priority date	Aug 7, 2020
Publication date	Dec 27, 2022
Grant date	Dec 27, 2022

Abstract

An image processing method includes: determining a source transmission map based on a dark channel map of an input image; determining transformed transmission maps by applying different filters to the determined source transmission map; generating haze-free images by removing haze from the input image based respectively on the determined transformed transmission maps; and generating an output image by blending the generated haze-free images.

First claim

What is claimed is:

1. An image processing method, comprising:
determining a source transmission map based on a dark channel map of an input image;
determining transformed transmission maps by applying different filters to the determined source transmission map;
generating haze-free images by removing haze from the input image based respectively on the determined transformed transmission maps; and
generating an output image,
wherein the haze-free images comprise:
a first haze-free image for determining a flat region of the output image; and
a second haze-free image for determining a strong edge region of the output image, and

wherein the generating of the output image comprises:
determining an edge map by detecting an edge in the input image;
generating first transformed haze-free images of different scales by performing a pyramid transformation on the first haze-free image;
generating second transformed haze-free images of different scales by performing another pyramid transformation on the second haze-free image;
generatinq the blended intermediate images by blending the first transformed haze-free images and the second transformed haze-free images based on the edge map; and
generating the output image based on the blended intermediate images.

2. The method of claim 1 , wherein the transformed transmission maps and the haze-free images are based on a size of a single mask of the source transmission map.

3. The method of claim 1 , wherein the filters comprise:
the first filter being configured to restore a texture; and
a second filter configured to suppress a halo artifact.

4. The method of claim 3 , wherein the first filter comprises a multi-directional kernel-based filter, and the second filter comprises a guided filter.

5. The method of claim 1 , wherein the generating of the output image comprises:
determining a blending weight based on edges in the input image; and
performing the blending based on the determined blending weight.

6. The method of claim 1 , wherein the first haze-free image comprises the flat region including a restored texture; and
the second haze-free image comprises the strong edge region including a suppressed halo artifact.

7. The method of claim 1 , wherein the blending comprises:
blending a first transformed haze-free image and a second transformed haze-free image by assigning a greater weight to an edge region in the second transformed haze-free image than a weight assigned to the edge region in the first transformed haze-free image, wherein the edge region is determined based on the edge map.

8. The method of claim 1 , wherein the generating of the output image comprises:
generating the output image by performing an inverted pyramid transformation on the blended intermediate images.

9. The method of claim 1 , wherein the determining of the edge map comprises:
determining the edge map by performing a morphological operation on the detected edge of the input image.

10. A non-transitory computer-readable storage medium storing instructions that, when executed by a processor, configure the processor to perform the method of claim 1 .

11. An image processing apparatus, comprising:
a processor configured to:
determine a source transmission map based on a dark channel map of an input image;
determine transformed transmission maps by applying different filters to the determined source transmission map;
generate haze-free images by removing haze from the input image based respectively on the determined transformed transmission maps; and
generate an output image,

wherein the haze-free images comprise:
a first haze-free image for determining a flat region of the output image; and
a second haze-free image for determining a strong edge region of the output image,

wherein, for the generating of the output image, the processor is configured to:
determine an edge map by detecting an edge in the input image;
generate first transformed haze-free images of different scales by performing a pyramid transformation on the first haze-free image;
generate second transformed haze-free images of different scales by performing another pyramid transformation on the second haze-free image;
generate blended intermediate images by blending the first transformed haze-free images and the second transformed haze-free images based on the edge map; and
generate the output image based on the blended intermediate images.

12. The apparatus of claim 11 , wherein the transformed transmission maps and the haze-free images are based on a size of a single mask of the source transmission map.

13. The apparatus of claim 11 , wherein the filters comprise:
the first filter being a multi-directional kernel-based filter configured to restore a texture; and
a guided filter configured to suppress a halo artifact.

14. The apparatus of claim 11 , wherein the processor is further configured to determine a blending weight, based on a detected edge in the input image, by performing a morphological operation on the detected edge.

15. The apparatus of claim 11 , further comprising:
a camera configured to generate the input image; and
a control system configured to control a vehicle based on a generated control instruction,
wherein the processor is configured to generate the control instruction based on the generated output image, and the apparatus is a vehicle control apparatus.

16. The apparatus of claim 11 further comprising a memory storing instructions that, when executed by the processor, configure the processor to perform the determining of the source transmission map, the determining of the transformed transmission maps, the generating of the haze-free images, and the generating of the output image.

17. A vehicle control apparatus, comprising:
a camera configured to generate an input image of surroundings of a vehicle;
a processor configured to
determine a source transmission map based on a dark channel map of the input image,
determine transformed transmission maps by applying different filters to the determined source transmission map,
generate haze-free maps by removing haze from the input image based respectively on the determined transformed transmission maps,
generate an output image based on the generated haze-free maps, and
generate a control instruction for traveling of the vehicle based on the generated output image; and

a control system configured to control the vehicle based on the generated control instruction,
wherein the haze-free maps comprise:
a first haze-free map for determining a flat region of the output image; and
a second haze-free map for determining a strong edge region of the output image,

wherein, for the generating of the output image, the processor is configured to:
determine an edge map by detecting an edge in the input image; and
generate first transformed haze-free maps of different scales by performing a pyramid transformation on the first haze-free map;
generate second transformed haze-free maps of different scales by performing a pyramid transformation on the second haze-free map; and
generate blended intermediate images by blending the first transformed haze-free maps and the second transformed haze-free maps based on the edge map; and
generate the output image based on the blended intermediate images.

18. The apparatus of claim 17 , wherein the filters comprise:
a multi-directional kernel-based filter configured to restore a texture; and
a guided filter configured to suppress a halo artifact.

19. The apparatus of cl

Assignees

• Samsung Electronics Co Ltd
• Univ Yonsei Iacf

Inventors

• Jee Seunghoon
• Kang Moon Gi
• Hong Soon Young
• Kim Min Sub
• Lee Sang Yoon

Classifications

• G06T5/50

  using two or more images, e.g. averaging or subtraction · CPC title

• G06T2207/20221

  Image fusion; Image merging · CPC title

• G06T7/13

  Edge detection · CPC title

• G06T2207/30252

  Vehicle exterior; Vicinity of vehicle · CPC title

• G06T5/20

  using local operators · CPC title