Automatic selection of optimum algorithms for high dynamic range image processing based on scene classification

Patent metadata

Field	Value
Publication number	US-9852499-B2
Application number	US-201314105652-A
Country	US
Kind code	B2
Filing date	Dec 13, 2013
Priority date	Dec 13, 2013
Publication date	Dec 26, 2017
Grant date	Dec 26, 2017

Abstract

A method for processing high dynamic range (HDR) images by selecting preferred tone mapping operators and gamut mapping algorithms based on scene classification. Scenes are classified into indoor scenes, outdoor scenes, and scenes with people, and tone mapping operators and gamut mapping algorithms are selected on that basis. Prior to scene classification, the multiple images taken at various exposure values are fused into a low dynamic range (LDR) image using an exposure fusing algorithm, and scene classification is performed using the fused LDR image. Then, the HDR image generated from the multiple images are tone mapped into a LDR image using the selected tone mapping operator and then gamut mapped to the color space of the output device such as printer.

First claim

What is claimed is:

1. A method implemented in a data processing apparatus for processing a set of multiple input images of a scene taken at various exposure values, comprising:
(a) using one or more of the multiple input images, classifying the scene into one of a plurality of scene classes, wherein the plurality of scene classes either include a class of scenes that contains a significant presence of people, a class of outdoor scenes and a class of indoor scenes, or include a class of scenes that contains a significant presence of people, a class of daylight scenes and a class of night scenes,
wherein step (a) comprises:
(a1) selecting a subset of two or more images from the multiple input images;
(a2) down-sampling the selected images;
(a3) after down-sampling, fusing the selected images into a single fused image, by generating a weight map for each selected image and combining the selected images using the weight maps, the weight maps being generated based on only one of saturation, contrast, and well-exposedness of each pixel in each selected image; and
(a4) classifying the fused image into one of the plurality of scene classes;

(b) based on the scene class determined in step (a), selecting one of a plurality of pre-stored tone mapping operators;
(c) merging the multiple input images to generate a high dynamic range (HDR) image;
(d) tone mapping the HDR image using the tone mapping operator selected in step (b) to generate a low dynamic range (LDR) image;
(e) based on the scene class determined in step (a), selecting one of a plurality of pre-stored gamut mapping algorithms; and
(f) using the gamut mapping algorithm selected in step (e), converting the LDR image generated in step (d) from a color space of the image to a color space of an output device.

2. The method of claim 1 , wherein step (a4) comprises:
detecting faces or areas having skin tones and predefined shapes in the fused image;
if any face or area having skin tone and predefined shapes is detected, classifying the scene as a first class which is scenes with significant presence of people; and
otherwise, classifying the scene as either a second class which is outdoor scenes or a third class which is indoor scenes by using correlated color temperatures of the fused image.

3. The method of claim 1 , wherein step (a4) comprises:
detecting faces or areas having skin tones and predefined shapes in the fused image;
if any face or area having skin tone and predefined shapes is detected, classifying the scene as a first class which is scenes with significant presence of people; and
otherwise, classifying the scene as either a second class which is daylight scenes or a third class which is night scenes using a histogram of pixel intensities of the fused image.

4. The method of claim 1 , wherein the subset includes all of the multiple input images.

5. The method of claim 1 , wherein the color space of the image is an RGB color space and the color space of the output device is a CMYK color space.

6. The method of claim 1 , wherein the subset includes fewer than all of the multiple input images.

7. A computer program product comprising a computer usable non-transitory medium having a computer readable program code embedded therein for controlling a data processing apparatus, the computer readable program code configured to cause the data processing apparatus to execute a process for processing a set of multiple input images of a scene taken at various exposure values, the process comprising:
(a) using one or more of the multiple input images, classifying the scene into one of a plurality of scene classes, wherein the plurality of scene classes either include a class of scenes that contains a significant presence of people, a class of outdoor scenes and a class of indoor scenes, or include a class of scenes that contains a significant presence of people, a class of daylight scenes and a class of night scenes,
wherein step (a) comprises:
(a1) selecting a subset of two or more images from the multiple input images;
(a2) down-sampling the selected images;
(a3) after down-sampling, fusing the selected images into a single fused image, by generating a weight map for each selected image and combining the selected images using the weight maps, the weight maps being generated based on only one of saturation, contrast, and well-exposedness of each pixel in each selected image; and
(a4) classifying the fused image into one of the plurality of scene classes;

(b) based on the scene class determined in step (a), selecting one of a plurality of pre-stored tone mapping operators;
(c) merging the multiple input images to generate a high dynamic range (HDR) image;
(d) tone mapping the HDR image using the tone mapping operator selected in step (b) to generate a low dynamic range (LDR) image;
(e) based on the scene class determined in step (a), selecting one of a plurality of pre-stored gamut mapping algorithms; and
(f) using the gamut mapping algorithm selected in step (e), converting the LDR image generated in step (d) from a color space of the image to a color space of an output device.

8. The computer program product of claim 7 , wherein step (a4) comprises:
detecting faces or areas having skin tones and predefined shapes in the fused image;
if any face or area having skin tone and predefined shapes is detected, classifying the scene as a first class which is scenes with significant presence of people; and
otherwise, classifying the scene as either a second class which is outdoor scenes or a third class which is indoor scenes by using correlated color temperatures of the fused image.

9. The computer program product of claim 7 , wherein step (a4) comprises:
detecting faces or areas having skin tones and predefined shapes in the fused image;
if any face or area having skin tone and predefined shapes is detected, classifying the scene as a first class which is scenes with significant presence of people; and
otherwise, classifying the scene as either a second class which is daylight scenes or a third class which is night scenes using a histogram of pixel intensities of the fused image.

10. The computer program product of claim 7 , wherein the subset includes all of the multiple input images.

11. The computer program product of claim 7 , wherein the color space of the image is an RGB color space and the color space of the output device is a CMYK color space.

12. A digital camera comprising the computer usable non-transitory medium of claim 7 , the digital camera further comprising:
an imaging section for obtaining images; and
a control section for controlling the imaging section to obtain the set of multiple images having different exposure levels.

13. The computer program product of claim 7 , wherein the subset includes fewer than all of the multiple input images.

Assignees

• Konica Minolta Laboratory Usa Inc

Inventors

• Ming Wei
• Zhan Xiaonong

Classifications

• G06T2207/10016

  Video; Image sequence · CPC title

• G06T2207/20208

  High dynamic range [HDR] image processing · CPC title

• G06T5/50 Primary

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

• G06T5/009

  Physics · mapped topic

• G06T2207/10144

  Varying exposure · CPC title