Device, method, and computer readable medium for restoring an image

Patent metadata

Field	Value
Publication number	US-9317905-B2
Application number	US-201214364834-A
Country	US
Kind code	B2
Filing date	Dec 10, 2012
Priority date	Dec 12, 2011
Publication date	Apr 19, 2016
Grant date	Apr 19, 2016

Abstract

An image processing device, which uses a dictionary which stores data of two or over sets which associate a blurred patch of a blurred image generated by blurring a predetermined image with a restoration patch of the predetermined image. The device includes a processor that processes instructions to generate a tentative restored image from the restoration patch selected based on values indicating similarities between plural input patches generated by dividing an inputted input image and the blurred patch. The processor also processes the instructions to generate a tentative patch by dividing the tentative restored image. The processor further processes the instructions to generate a restored image from the restoration patch selected based on a sum of the value indicating the similarity between the input patch and the blurred patch and a value indicating the similarity between the tentative patch and the restoration patch.

First claim

The invention claimed is:

1. An image processing device, which uses a dictionary which stores data of two or over sets which associate a blurred patch which is a patch of a blurred image generated by blurring a predetermined image with a restoration patch which is a patch of the predetermined image; comprising:
a memory storing instructions derived from a computer program; and
a processor configured to process the instructions to:
generate a tentative restored image from the restoration patch selected based on values indicating similarities between plural input patches generated by dividing an inputted input image and the blurred patch;
generate a tentative patch by dividing the tentative restored image; and
generate a restored image from the restoration patch selected based on a sum of the value indicating the similarity between the input patch and the blurred patch and a value indicating the similarity between the tentative patch and the restoration patch.

2. The image processing device according to claim 1 , wherein the processor is further configured to process the instructions to:
receive an image from outside;
generate plural input patches by dividing the received input image into blocks;
select the restoration patch for each input patch from the data of the sets stored in the dictionary based on values indicating the similarities of the images between the plural input patches and the blurred patch; and
generate the tentative restored image by composing the restoration patch for each selected input patch.

3. The image processing device according to claim 1 , wherein the value indicating the similarity is a value calculated based on a difference of brightness value of each pixel between the patches.

4. The image processing device according to claim 1 , wherein
the dictionary stores a blur parameter of the blurred patch together with the blurred patch and the restoration patch, and
the processor is further configured to process the instructions to generate the restored image from the restoration patch selected based on a sum of the value indicating the similarity between the input patch and the blurred patch, the value indicating the similarity between the tentative patch and the restoration patch, and a value indicating the similarity of the blur parameter between the tentative patch and the blurred patch.

5. The image processing device according to claim 1 , wherein the processor is further configured to process the instructions to generate the restoration image by selecting the restoration patch regarding an area in which the value indicating the similarity between the tentative patch and the restoration patch is a value which indicates lower similarity between the tentative patch and the restoration patch.

6. The image processing device according to claim 5 ,
wherein the processor is further configured to process the instructions to distinguish an area in which the value indicating the similarity between the tentative patch and the restoration patch is a value which indicates that a similarity between the tentative patch and the restoration patch is low from areas other than the area, and display it.

7. An image processing method comprising:
storing data of two or over sets which associate a blurred patch which is a patch generated by blurring a predetermined image with a restoration patch which is a patch of the predetermined image;
generating a tentative restored image from the restoration patch selected based on values indicating similarities between plural input patches generated by dividing an inputted input image and the blurred patch;
generating a tentative patch by dividing the tentative restored image; and
generating a restored image from the restoration patch selected based on a sum of the value indicating the similarity between the input patch and the blurred patch and a value indicating the similarity between the tentative patch and the restoration patch.

8. A non transitory computer readable medium embodying a program, said program causing an image processing device to perform a method, said method comprising:
storing data of two or over sets which associates a blurred patch which is a patch generated by blurring a predetermined image with a restoration patch which is a patch of the predetermined image;
generating a tentative restored image from the restoration patch selected based on values indicating a similarities between plural input patches generated by dividing an inputted input image and the blurred patch;
generating a tentative patch by dividing the tentative restored image; and
generating a restored image from the restoration patch selected based on a sum of the value indicating the similarity between the input patch and the blurred patch and a value indicating the similarity between the tentative patch and the restoration patch.

9. An image processing device, which uses a dictionary storing data of two or over sets which associate a blurred patch which is a patch of a blurred image generated by blurring a predetermined image with a restoration patch which is a patch of the predetermined image; comprising:
tentative restored image generation means for generating a tentative restored image from the restoration patch selected based on values indicating similarities between plural input patches generated by dividing an inputted input image and the blurred patch;
tentative patch generation means for generating a tentative patch by dividing the tentative restored image; and
restoration means for generating a restored image from the restoration patch selected based on a sum of the value indicating the similarity between the input patch and the blurred patch and a value indicating the similarity between the tentative patch and the restoration patch.

Assignees

• Shibata Takashi
• Iketani Akihiko
• Senda Shuji
• Nec Corp

Inventors

• Shibata Takashi
• Iketani Akihiko
• Senda Shuji

Classifications

• G06T2207/20182

  Noise reduction or smoothing in the temporal domain; Spatio-temporal filtering · CPC title

• G06T2207/20081

  Training; Learning · CPC title

• G06T5/50

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

• G06T5/003 Primary

  Physics · mapped topic

• H04N19/97

  Matching pursuit coding · CPC title