METHOD AND APPARATUS FOR RECONSTRUCTING A HIGH-RESOLUTION IMAGE BY USING MULTI-LAYER LOW-RESOLUTION IMAGES

• US 20110038558A1
• Filed: 08/13/2010
• Published: 02/17/2011
• Est. Priority Date: 08/13/2009
• Status: Active Grant

• Current Assignee: Samsung Electronics Co. Ltd.
• Original Assignee: Samsung Electronics Co. Ltd.
• Inventors: PARK, Jong-hyun, LEE, Hyun-seung, GIM, Gi-yeong
• Granted Patent: US 8,571,355 B2
• Time in Patent Office: Days
• US Class Current: 382/264
• CPC Class Codes: G06T 1/00 General purpose image data processing

Abstract

A method of reconstructing a high-resolution image by using multi-layer low-resolution images includes dividing a low-resolution image into a plurality of texture layer low-resolution images and a plurality of base layer low-resolution images; generating a texture layer high-resolution image by compositing the plurality of the texture layer low-resolution images and generating a base layer high-resolution image by compositing the plurality of the base layer low-resolution images; and outputting a high-resolution image by compositing the texture layer high-resolution image and the base layer high-resolution image.

Claims

1. A method of reconstructing a high-resolution image by using multi-layer low-resolution images, the method comprising:

dividing a low-resolution image into a plurality of texture layer low-resolution images and a plurality of base layer low-resolution images;

generating a texture layer high-resolution image by compositing the plurality of the texture layer low-resolution images and generating a base layer high-resolution image by compositing the plurality of the base layer low-resolution images; and

outputting a high-resolution image by compositing the texture layer high-resolution image and the base layer high-resolution image.

2. The method of claim 1, wherein the outputting of the high-resolution image further comprises restoring high-frequency components of the high-resolution image.

3. The method of claim 1, wherein the dividing of the low-resolution image comprises obtaining the plurality of the texture layer low-resolution images and the plurality of the base layer low-resolution images by dividing the low-resolution image into a plurality of low-resolution images and dividing each of the plurality of low-resolution images into a texture layer low-resolution image and a base layer low-resolution image.

4. The method of claim 1, wherein the dividing of the low-resolution image comprises obtaining the plurality of the base layer low-resolution images by excluding texture layer image components from the low-resolution image.

5. The method of claim 1, wherein the dividing of the low-resolution image comprises:

generating the plurality of the base layer low-resolution images by applying a filter for at least one of preserving edge components and smoothing an image, to the low-resolution image; and

generating difference images between the low-resolution image and the plurality of the base layer low-resolution images, as the plurality of the texture layer low-resolution images.

6. The method of claim 1, wherein the generating of the texture layer high-resolution image and the base layer high-resolution image comprises:

generating the texture layer high-resolution image by performing non-iterative filtering on the plurality of the texture layer low-resolution images; and

generating the base layer high-resolution image by performing non-iterative filtering on the plurality of the base layer low-resolution images.

7. The method of claim 6, wherein the generating of the texture layer high-resolution image and the base layer high-resolution image comprises:

generating the texture layer high-resolution image by performing weighted sum filtering on a first current frame of each of the plurality of the texture layer low-resolution images, and at least one first temporally neighboring frame of the first current frame and at least one first spatially neighboring frame of the first current frame; and

generating the base layer high-resolution image by performing weighted sum filtering on a second current frame of each of the plurality of the base layer low-resolution images, at least one second temporally neighboring frame of the second current frame and at least one second spatially neighboring frame of the second current frame.

8. The method of claim 7, wherein the at least one first temporally neighboring frame of the first current frame and the at least one spatially neighboring frame of the first current frame or the at least one second temporally neighboring frame of the second frame and the at least one second spatially neighboring frame of the second current frame are one of motion-estimated and motion-compensated frames with respect to the first current frame or and the second current frame.

9. The method of claim 7, wherein a weight used to perform the weighted sum filtering is determined based on at least one of spatial proximity, temporal proximity, luminance differences, proximity to a strong edge and reliability of motion estimation of neighboring frames with respect to one of the first current frame and the second current frame.

10. The method of claim 1, wherein the outputting of the high-resolution image comprises generating the high-resolution image by performing weighted average filtering on the texture layer high-resolution image and the base layer high-resolution image.

11. The method of claim 10, wherein the weighted average filtering is performed by applying a first weight for minimizing damage of texture layer image components of an original image, to the texture layer high-resolution image, and applying a second weight for smoothing the original image in consideration of an edge direction of the original image, to the base layer high-resolution image.

12. The method of claim 2, wherein the restoring of the high-frequency components comprises restoring the high-frequency components of the high-resolution image by performing non-iterative filtering on the high-resolution image.

13. The method of claim 2, wherein the restoring of the high-frequency components comprises generating a high-resolution image by performing weighted sum filtering on at least one of the high-resolution image, a result of performing first filtering for restoring damaged high-frequency components of the high-resolution image, on the high-resolution image, and a result of performing second filtering for preventing noise amplification, on the high-resolution image.

14. The method of claim 13, wherein the first filtering is performed by using at least one bandpass filter that operates according to band characteristics of frequency bands.

15. The method of claim 13, wherein the second filtering is performed by using at least one bilateral filter that operates according to band characteristics of frequency bands.

16. The method of claim 1, wherein the low-resolution image is obtained by magnifying an original image.

17. The method of claim 5, wherein the filter for preserving edge components and smoothing an image comprises a bilateral filter.

25. A computer readable recording medium having recorded thereon a computer program for executing the method of claim 1.

18. An apparatus for reconstructing a high-resolution image by using multi-layer low-resolution images, the apparatus comprising:

a low-resolution image layer division unit for dividing a low-resolution image into a plurality of texture layer low-resolution images and a plurality of base layer low-resolution images;

a texture layer high-resolution image generation unit for generating a texture layer high-resolution image by compositing the plurality of the texture layer low-resolution images;

a base layer high-resolution image generation unit for generating a base layer high-resolution image by compositing the plurality of the base layer low-resolution images; and

a high-resolution image composition unit for outputting a high-resolution image by compositing the texture layer high-resolution image and the base layer high-resolution image.

19. The apparatus of claim 18, wherein the high-resolution image composition unit comprises a high-frequency component restoration unit for restoring high-frequency components of the high-resolution image.

20. The apparatus of claim 18, further comprising a motion estimation unit for estimating motion information between a current frame of the low-resolution image and continuous neighboring frames of the current frame,

wherein the texture layer high-resolution image generation unit generates the texture layer high-resolution image by performing weighted sum filtering on a first current frame of each of the plurality of the texture layer low-resolution images, and at least one first temporally neighboring frame of the first current frame and at least one first spatially neighboring frame of the first current frame, by using the neighboring frames motion-compensated based on the motion information, and

wherein the base layer high-resolution image generation unit generates the base layer high-resolution image by performing weighted sum filtering on a second current frame of each of the plurality of the base layer low-resolution images, and at least one temporally neighboring frame of the second current frame and at least one spatially neighboring frame of the second current frame, by using neighboring frames motion-compensated based on the motion information.

21. The apparatus of claim 18, wherein the high-resolution image composition unit generates the high-resolution image by performing weighted average filtering on the texture layer high-resolution image and the base layer high-resolution image.

22. The apparatus of claim 19, wherein the high-frequency component restoration unit generates the high-resolution image by performing weighted sum filtering on at least one of the high-resolution image, a result of performing first filtering for restoring damaged high-frequency components of the high-resolution image, on the high-resolution image, and a result of performing second filtering for preventing noise amplification, on the high-resolution image.

23. The apparatus of claim 18, wherein at least one of the texture layer high-resolution image generation unit, the base layer high-resolution image generation unit and the high-resolution image composition unit performs non-iterative weighted sum filtering on input images.

24. An apparatus for reconstructing a high-resolution image by using multi-layer low-resolution images, the apparatus comprising:

a low-resolution image multi-layer division unit for dividing an input low-resolution image into a plurality of individual layer low-resolution images with respect to each of a plurality of layers;

an individual layer high-resolution image generation unit for generating an individual layer high-resolution image by compositing the plurality of the individual layer low-resolution images with respect to each of the plurality of the layers; and

a multi-layer high-resolution image composition unit for outputting a high-resolution image by compositing a plurality of individual layer high-resolution images generated with respect to the plurality of layers.

26. A method of reconstructing a high-resolution image by using multi-layer low-resolution images, said method comprising:

dividing a low-resolution image into at least one texture layer low-resolution image and at least one base layer low-resolution image;

generating a texture layer high-resolution image based on said at least one texture layer low-resolution image;

generating a base layer high-resolution image based on said at least one base layer low-resolution image;

generating the high-resolution image based on said texture layer high-resolution image and said base layer high-resolution image.

27. The method according to claim 26, wherein at least one of the base layer high-resolution image and texture layer high-resolution image is generated by performing non-iterative filtering.

28. An apparatus for reconstructing a high-resolution image by using multi-layer low-resolution images, the apparatus comprising:

a low-resolution image layer division unit for dividing a low-resolution image into at least one texture layer low-resolution and at least one base layer low-resolution image;

a texture layer high-resolution image generation unit for generating a texture layer high-resolution image based on said at least one texture layer low-resolution image;

a base layer high-resolution image generation unit for generating a base layer high-resolution image based on said at least one base layer low-resolution image; and

a high-resolution image composition unit for generating the high-resolution image based on said texture layer high-resolution image and said base layer high-resolution image.

29. The apparatus according to claim 28, wherein at least one of the base layer high-resolution image and texture layer high-resolution image is generated by performing non-iterative filtering.