Apparatus, method and computer-readable medium removing noise of color image

• US 8,718,361 B2
• Filed: 10/14/2010
• Issued: 05/06/2014
• Est. Priority Date: 02/05/2010
• Status: Expired due to Fees

• Current Assignee: Samsung Electronics Co. Ltd.
• Original Assignee: Samsung Electronics Co. Ltd.
• Inventors: Moon, Young Su, Cha, Joon Hyuk, Lee, Shi Hwa, Jeon, Young Sun
• Primary Examiner(s): Chen, Wenpeng
• Application Number:
  US12/904,793
• Publication Number:
  US 20110194763A1
• Time in Patent Office: 1,300 Days
• Field of Search: None
• US Class Current: 382/167
• CPC Class Codes: G06T 2207/10012 Stereo images
• CPC Class Codes: G06T 5/70 Denoising; Smoothing

Abstract

Provided is an apparatus, method and computer-readable medium removing noise of an image. The apparatus may include a channel image correction unit to correct remaining channel images excluding a currently processed channel image, from among a plurality of channel images, using the currently processed channel image, a noise removal unit to remove noise of the currently processed channel image, using the corrected remaining channel images and the currently processed channel image, and a color image reconstruction unit to reconstruct a color image in which noise is removed, by combining the plurality of channel images in which noise is removed when the noise of the plurality of channel images is removed.

Claims

1. An apparatus removing noise of an image, comprising:

a color image division unit to divide an inputted color image into a plurality of channel images including a first channel image and a second channel image;

a channel image correction unit to correct the second channel image using the first channel image;

a noise removal unit to remove noise of the first channel image, using the corrected second channel image and the first channel image; and

a color image reconstruction unit to reconstruct a color image in which noise is removed, based on the noise-removed first channel image and the corrected second channel image,

wherein the first channel image is a current channel image to be processed, and

wherein the channel image correction unit corrects the second channel images to replace an image corresponding to a low frequency band from among the second channel images with an image corresponding to a low frequency band from among the first channel images.

2. The apparatus of claim 1, further comprising:

a frequency band separation unit to separate each of the plurality of channel images to a low frequency band image and a high frequency band image to generate a plurality of Laplacian pyramids corresponding to the plurality of channel images, wherein the channel image correction unit corrects low frequency bands of the plurality of Laplacian pyramids corresponding to the second channel images to output the corrected plurality of Laplacian pyramids.

3. The apparatus of claim 2, wherein the channel image correction unit corrects a part or all of the low frequency bands of the plurality of Laplacian pyramids.

4. The apparatus of claim 3, further comprising:

a channel image restoration unit to restore, into the plurality of channel images, the corrected plurality of Laplacian pyramids corresponding to the second channel images.

5. The apparatus of claim 1, wherein the noise removal unit sets, in the first channel image and the corrected second channel images, a search region where a pattern matching is performed, with respect to a current pixel in which noise is removed, and removes the noise of the current pixel using a cumulative result of intensity where a weight is applied to neighboring pixels of the current pixel and a cumulative result of the weight.

6. The apparatus of claim 5, wherein the noise removal unit divides the cumulative result of the intensity into the cumulative result of the weight to remove the noise of the current pixel.

7. A method of removing noise of an image, comprising:

dividing, an inputted color image into a plurality of channel images including a first channel image and a second channel image;

correcting, by at least one processor, the second channel image using the first channel image;

removing, by the at least one processor, noise of the first channel image, using the corrected second channel images and the first channel image; and

reconstructing, by the at least one processor, a color image in which noise is removed, based on the noise-removed first channel image and the corrected second channel,

wherein the first channel image is a current channel image to be processed, and

wherein the correcting corrects the second channel images to replace an image corresponding to a low frequency band from among the second channel images with an image corresponding to a low frequency band from among the first channel images.

8. The method of claim 7, further comprising:

generating a plurality of Laplacian pyramids corresponding to the plurality of channel images by separating each of the plurality of channel images to a low frequency band image and a high frequency band image,

wherein the correcting corrects low frequency bands of the plurality of Laplacian pyramids corresponding to the second channel images to output the corrected plurality of Laplacian pyramids.

9. The method of claim 8, wherein the correcting corrects a part or all of the low frequency bands of the plurality of Laplacian pyramids.

10. The method of claim 9, further comprising:

restoring, into the plurality of channel images, the corrected plurality of Laplacian pyramids corresponding to the second channel images.

11. The method of claim 7, wherein the removing sets, in the first channel and the corrected second channel images, a search region where a pattern matching is performed, with respect to a current pixel in which noise is removed, and removes the noise of the current pixel using a cumulative result of intensity where a weight is applied to neighboring pixels of the current pixel and a cumulative result of the weight.

12. At least one non-transitory computer-readable medium comprising computer readable instructions that control at least one processor to implement a method, comprising:

dividing, an inputted color image into a plurality of channel images including a first channel image and a second channel image;

correcting the second channel image using the first channel image;

removing noise of the first channel image, using the corrected second channel image and the first channel image; and

reconstructing a color image in which noise is removed, based on the noise-removed first channel image and the corrected second channel image,

wherein the first channel image is a current channel image to be processed, and

wherein the correcting corrects the second channel images to replace an image corresponding to a low frequency band from among the second channel images with an image corresponding to a low frequency band from among the first channel images.