Image processing method for removing noise from an image

• US 8,059,910 B2
• Filed: 03/31/2006
• Issued: 11/15/2011
• Est. Priority Date: 03/31/2005
• Status: Active Grant

• Current Assignee: Nikon Corporation
• Original Assignee: Nikon Corporation
• Inventors: Ishiga, Kenichi
• Primary Examiner(s): Repko; Jason M
• Assistant Examiner(s): Thirugnanam; Gandhi
• Application Number:
  US11/886,325
• Publication Number:
  US 20080199100A1
• Time in Patent Office: 2,055 Days
• Field of Search: 382/263
• US Class Current: 382/263
• CPC Class Codes: G06T 2207/20016 Hierarchical, coarse-to-fine, multiscale or multiresolution image processing; Pyramid transform
• CPC Class Codes: G06T 2207/20064 Wavelet transform [DWT]
• CPC Class Codes: G06T 2207/20192 Edge enhancement; Edge preservation
• CPC Class Codes: G06T 5/10 using non-spatial domain filtering
• CPC Class Codes: G06T 5/20 using local operators
• CPC Class Codes: G06T 5/70 Denoising; Smoothing

Abstract

An image processing method for removing noise contained in an image includes: inputting an original image constituted with a plurality of pixels; generating a plurality of band-limited images, resolution levels of which are lowered in sequence, by executing a specific type of band pass filter processing on the original image; extracting a high-frequency component from each of the band-limited images by executing high-pass filter processing on the band-limited images; extracting a noise component contained in each of the band-limited images by separating the noise component through nonlinear conversion processing executed on the extracted high-frequency component; and removing the noise in the original image based upon the extracted noise component in each band-limited image.

Claims

1. An image processing method for removing noise contained in an image, comprising:

inputting an original image constituted with a plurality of pixels;

extracting a first high-frequency component through first high-pass filter processing executed on the original image;

extracting a second high-frequency component through second high-pass filter processing executed on the extracted first high-frequency component;

extracting a noise component by separating the noise component through nonlinear conversion processing executed on the extracted second high-frequency component; and

removing noise in the original image based upon the extracted noise component, wherein

the nonlinear conversion processing is attenuation processing whereby a non-noise component contained in the second high-frequency component is attenuated by multiplying the second high-frequency component by a weighted Gaussian coefficient having an argument of a value resulting from a comparison of the second high-frequency component with a predetermined threshold value.

2. An image processing method according to claim 1, wherein:

a specific type of bandpass filter processing including the first high-pass filter processing is executed on the original image to generate a plurality of high-frequency band-limited images corresponding to the first high-frequency component, and the plurality of high-frequency band-limited images have resolution levels that are lowered in sequence;

the second high-frequency component is extracted from each of the high-frequency band-limited images by individually executing the second high-pass filter processing on the plurality of high-frequency band-limited images having been generated;

the noise component contained in each of the high-frequency band-limited images is extracted by separating the noise component through the nonlinear conversion processing executed on the extracted second high-frequency component; and

the noise in the original image is removed based upon the noise component extracted from each of the high-frequency band-limited images.

3. An image processing method according to claim 2, wherein:

the plurality of high-frequency band-limited images are a set of high-frequency band images obtained through a multi-resolution transformation executed on the original image, and the plurality of high-frequency band-limited images have resolution levels that are lowered in sequence.

4. An image processing method according to claim 3, wherein:

the plurality of high-frequency band-limited images are a set of high-pass component sub-band images resulting from wavelet transformation.

5. An image processing method according to claim 2, wherein:

the second high-pass filter processing is executed on the plurality of high-frequency band-limited images by using filters of uniform size.

8. An image processing method according to claim 1, wherein:

a standard deviation value determined in a uniform image area of the original image is used as the predetermined threshold value.

9. A non-transitory computer-readable medium storing a computer program product, comprising:

an image processing program that enables a computer or an image processing apparatus to execute an image processing method according to claim 1.

10. An image processing apparatus, comprising:

a computer that executes an image processing method according to claim 1.

6. An image processing method for removing noise contained in an image, comprising:

obtaining the image; and

performing a processing operation on the image, comprising;

extracting a first high-frequency component by applying a high-pass filter to the obtained image;

extracting a noise component from the extracted first high-frequency component through non-linear conversion;

subtracting the extracted noise component from the image;

extracting a second high-frequency component from the image from which the noise component has been subtracted, by applying the high-pass filter with the same characteristics as in the extraction of the first high-frequency component;

extracting a correction component from the second high-frequency component through non-linear conversion having the same characteristics used in the extraction of the noise component; and

adding the extracted correction component to the image from which the noise component has been subtracted to generate a resultant image, wherein

the processing operation is repeated at least once, in which each repetition comprises performing the processing operation on the resultant image of a previous processing operation,

the nonlinear conversion processing in the extraction of both the noise and the correction component is attenuation processing whereby a non-noise component contained in the first high-frequency component is attenuated by multiplying the first high-frequency component by a weighted Gaussian coefficient having an argument of a value resulting from a comparison of the first high-frequency component with a predetermined threshold value, and

an edge component corresponding to overcorrection contained in the second high frequency component is attenuated by multiplying the second high frequency component by another weighted Gaussian coefficient having an argument of a value resulting from a comparison of the second high-frequency component with the identical threshold value as defined in the noise extraction processing.

7. An image processing method according to claim 6, wherein:

the image is constituted with a luminance component and a chrominance component; and

processing for adding the correction component to the image from which the noise component has been subtracted is executed for an image corresponding to the luminance component alone.