Image processing device, image processing method, and integrated circuit

• US 8,693,801 B2
• Filed: 09/15/2011
• Issued: 04/08/2014
• Est. Priority Date: 09/28/2010
• Status: Expired due to Fees

• Current Assignee: Sovereign Peak Ventures, LLC (Dominion Harbor Enterprises, LLC)
• Original Assignee: Panasonic Corporation (Panasonic Holdings Corporation)
• Inventors: Nakamura, Tsuyoshi, Tezuka, Tadanori
• Primary Examiner(s): Liu, Li
• Application Number:
  US13/512,190
• Publication Number:
  US 20120288194A1
• Time in Patent Office: 936 Days
• Field of Search: None
• US Class Current: 382/264
• CPC Class Codes: G06T 2207/20192 Edge enhancement; Edge preservation
• CPC Class Codes: G06T 5/20 using local operators
• CPC Class Codes: G06T 5/70 Denoising; Smoothing
• CPC Class Codes: H04N 1/4092 Edge or detail enhancement

Abstract

Provided is an image processing device capable of performing appropriate noise reduction without causing blurring of edges and details of the image. The image processing device: specifies, on a normal distribution representing a noise model of the image, a reference range having a value of a target pixel as a median; computes, with respect to a local region that is an image region containing the target pixel, the number of pixels having a value falling within a first sub-range of the reference range and the number of pixels having a value falling within a second sub-range of the reference range, the first sub-range corresponding to larger values than the median and the second sub-range corresponding to smaller values than the median; corrects one of the first sub-range and the second sub-range so that one of the sub-ranges in which a fewer number of pixels are distributed than in the other sub-range is narrower relative to the other sub-range; determines a selection range that is composed of the one of the first sub-range and the second sub-range as corrected and the other one of the first sub-range and the second sub-range as is; and performs noise reduction on an image region subjected to noise reduction and at least containing the local region by using values falling within the selection range from among the values of the respective pixels contained in the target region.

Claims

1. An image processing device for correcting pixel values of an image to reduce noise in the image, the image processing device comprising:

a determination unit configured to;

specify, on a normal distribution representing a noise model of the image, a reference range having a value of a target pixel as a median;

compute, with respect to a local region that is an image region containing the target pixel, the number of pixels having a value falling within a first sub-range of the reference range and the number of pixels having a value falling within a second sub-range of the reference range, the first sub-range corresponding to larger values than the median and the second sub-range corresponding to smaller values than the median;

correct one of the first sub-range and the second sub-range so that one of the sub-ranges in which a fewer number of pixels are distributed than in the other sub-range is narrower relative to the other sub-range; and

determine a selection range that is composed of the one of the first sub-range and the second sub-range as corrected and the other one of the first sub-range and the second sub-range as is; and

a generating unit configured to generate a corrected value for the target pixel by correcting values of pixels contained in a target region to reduce noise, the target region being an image region subjected to noise reduction and at least containing the local region, and the value correction involving use of values falling within the selection range from among the values of the respective pixels contained in the target region.

2. The image processing device according to claim 1, wherein

the target region is designated as an initial target region,

the image processing device further comprising:

an output unit configured to perform noise reduction on an enlarged target region based on the corrected target pixel value and output an image represented by pixel values resulting from the noise reduction, the enlarged target region being an image region subjected to noise reduction by the output unit and at least containing the initial target region.

3. The image processing device according to claim 2, wherein

the output unit is configured to further acquire a smoothness level of the enlarged target region based on a value of each pixel contained in the enlarged target region,

the image processing device further comprising:

an acquiring unit configured to acquire, by using the smoothness level, part of removed components, the removed components determined based on the image input from an external source to the image processing device and the image to be output by the output unit; and

an adding unit configured to add the acquired part of the removed components to the image to be output by the output unit.

4. The image processing device according to claim 2, wherein

each pixel of the image has a color difference component and a luminance component,

the noise model is a representation of a pixel value associated with a standard deviation,

the determination unit is configured to:

specify a standard deviation corresponding to the value of the target pixel from the noise model; and

specify the reference range based on (i) the standard deviation, (ii) a first value determined according to a luminance level of the target pixel, and (iii) a second value determined according to the smoothness level of the initial target region, and

the generating unit is configured to generate the corrected value for the target pixel, by correcting the color difference component values of pixels contained in the initial target region to reduce noise, and the value correction involving use of values falling within the selection range from among the color difference component values of the respective pixels contained in the initial target region.

5. The image processing device according to claim 4, wherein

the output unit is configured to:

specify a standard deviation corresponding to the corrected target pixel value from the noise model;

specify a corrected reference range having the corrected target pixel value as a median based on (i) the standard deviation corresponding to the corrected target pixel value, (ii) the first value, and (iii) a third value determined according to a gray level of a corrected target region that is an image region at least containing the initial target region;

compute, with respect to the corrected target region, the number of pixels having a value falling within a third sub-range of the corrected reference range and the number of pixels having a value falling within a fourth sub-range of the corrected reference range, the third sub-range corresponding to larger values than the corrected target pixel value and the fourth sub-range corresponding to smaller values than the corrected target pixel value;

correct one of the third sub-range and the fourth sub-range in which a fewer number of pixels are distributed than in the other sub-range to be narrower than the other sub-range;

determine a corrected selection range that is composed of the one of the third sub-range and the fourth sub-range as corrected and the other one of the third sub-range and the fourth sub-range as is; and

generate a pixel value for the target pixel of the image to be output, by reducing noise from the enlarged target region that is an image region at least containing the corrected target region, and the noise reduction of the new target region involving use of values falling within the corrected selection range from among the values of the respective pixels contained in the enlarged target region.

6. The image processing device according to claim 5, wherein

the color difference component has a first color difference component and a second color difference component,

the determining unit includes:

a first determiner configured to determine a first selection range based on a first color difference component value of a pixel of the image; and

a second determiner configured to determine a second selection range based on a second color difference component value of a pixel of the image,

the generating unit includes;

a first generator configured to generate a first corrected value for the first color difference component value of the target pixel by correcting the first color difference component values of pixels contained in the initial target region to reduce noise, the correction of the first color difference component values involving use of values falling within the first selection range from among the first color difference component values of the respective pixels contained in the initial target region; and

a second generator configured to generate a second corrected value for the second color difference component value of the target pixel by correcting the second color difference component values of pixels contained in the initial target region to reduce noise, the correction of the second color difference component values involving use of values falling within the second selection range from among the second color difference component values of the respective pixels contained in the initial target region, and

the output unit includes;

a first output sub-unit configured to reduce noise from the enlarged target region based on the first corrected value of the target pixel and output an image represented by pixel values resulting from the noise reduction in terms of the first color difference components; and

a second output sub-unit configured to reduce noise from the enlarged target region based on the second corrected value of the target pixel and output an image represented pixel values resulting from the noise reduction in terms of the second color difference component;

the image processing device further comprising;

a calibrating unit configured to calibrate the first color difference component value of each pixel in the image to be output by the first output sub-unit and the second color difference component value of each pixel in the image to be output by the second output sub-unit, the calibration being made by applying norm processing to the first and second color difference component values.

7. The image processing device according to claim 2, wherein

the output unit is configured to:

specify, on the normal distribution representing the noise model of the image, a new reference range that is narrower than the reference range and having the corrected target pixel value as a new median;

determine, for each pixel contained in the enlarged target region, a new selection range based on the new median and the new reference range and in cooperation with the determination unit; and

perform noise reduction using values falling within the new selection range from among the values of the respective pixels contained in the enlarged target region.

8. An image processing method for an image processing device to correct pixel values of an image to reduce noise, the image processing method comprising:

a determining step of;

specifying, on a normal distribution representing a noise model of the image, a reference range having a value of a target pixel as a median;

computing, with respect to a local region that is an image region containing the target pixel, the number of pixels having a value falling within a first sub-range of the reference range and the number of pixels having a value falling within a second sub-range of the reference range, the first sub-range corresponding to larger values than the median and the second sub-range corresponding to smaller values than the median;

correcting one of the first sub-range and the second sub-range so that one of the sub-ranges in which a fewer number of pixels are distributed than in the other sub-range is narrower relative to the other sub-range; and

determining a selection range that is composed of the one of the first sub-range and the second sub-range as corrected and the other one of the first sub-range and the second sub-range as is; and

a generating step of generating a corrected value for the target pixel by correcting values of pixels contained in a target region to reduce noise, the target region being an image region subjected to noise reduction and at least containing the local region, and the value correction involving use of values falling within the selection range from among the values of the respective pixels contained in the target region.

9. An integrated circuit to be used in an image processing device that corrects pixel values of an image to reduce noise, the integrated circuit comprising:

a determination unit configured to;

specify, on a normal distribution representing a noise model of the image, a reference range having a value of a target pixel as a median;

compute, with respect to a local region that is an image region containing the target pixel, the number of pixels having a value falling within a first sub-range of the reference range and the number of pixels having a value falling within a second sub-range of the reference range, the first sub-range corresponding to larger values than the median and the second sub-range corresponding to smaller values than the median;

correct one of the first sub-range and the second sub-range so that one of the sub-ranges in which a fewer number of pixels are distributed than in the other sub-range is narrower relative to the other sub-range; and

determine a selection range that is composed of the one of the first sub-range and the second sub-range as corrected and the other one of the first sub-range and the second sub-range as is; and

a generating unit configured to generate a corrected value for the target pixel by correcting values of pixels contained in a target region to reduce noise, the target region being an image region subjected to noise reduction and at least containing the local region, and the value correction involving use of values falling within the selection range from among the values of the respective pixels contained in the target region.

10. An image processing device for correcting pixel values of an image to reduce noise, the image processing device comprising:

a first noise reduction unit configured to;

specify, on a normal distribution representing a noise model of the image, a first reference range having a value of a target pixel as a median; and

generate noise reduced pixel values by performing noise reduction processing on an initial target region that is an image region subjected to noise reduction and containing the target pixel, the noise reduction of the initial target region involving use of values falling within the first reference range from among values of respective pixels contained in the target region; and

a second noise reduction unit configured to;

specify, on the normal distribution representing the noise model of the image, a second reference range narrower than the first reference range and having, as a median, the noise reduced pixel value generated for the target pixel; and

generate noise reduced pixel values by performing noise reduction processing on an enlarged target region that is an image region subjected to noise reduction and at least containing the initial target region, the noise reduction of the enlarged target region involving use of values falling within the second reference range from among values of respective pixels contained in the target region.