Reducing noise in digital images

• US 7,813,586 B2
• Filed: 08/07/2006
• Issued: 10/12/2010
• Est. Priority Date: 08/07/2006
• Status: Active Grant

• Current Assignee: STRATA Skin Sciences, Inc.
• Original Assignee: Mela Sciences, Inc.
• Inventors: Kabelev, Nikolai, Gutkowicz-Krusin, Dina
• Primary Examiner(s): Bhatnagar; Anand
• Assistant Examiner(s): TORRES, JOSE
• Application Number:
  US11/500,197
• Publication Number:
  US 20080031537A1
• Time in Patent Office: 1,527 Days
• Field of Search: 382/254, 382/275, 348/281, 348/241, 348/243
• US Class Current: 382/275
• CPC Class Codes: G06T 2207/20008 Globally adaptive
• CPC Class Codes: G06T 2207/20076 Probabilistic image processing
• CPC Class Codes: G06T 5/70 Denoising; Smoothing
• CPC Class Codes: H04N 25/63 applied to dark current
• CPC Class Codes: H04N 25/67 applied to fixed-pattern noise, e.g. non-uniformity of response
• CPC Class Codes: H04N 25/677 for reducing the column or line fixed pattern noise
• CPC Class Codes: H04N 5/21 Circuitry for suppressing or minimising disturbance, e.g. moiré or halo

Abstract

A target digital image is received from an image sensor. The image is contaminated by noise of unknown magnitude that is represented by a reference digital image. A process is applied that uses statistical analysis of the target digital image and of the reference digital image to estimate a magnitude of the noise for at least some pixels of the target digital image.

Claims

1. A method comprising:

determining an unknown amplitude A of noise that contaminates at least some of the pixels of a digital image T of an arbitrary target, the amplitude being possibly different in different digital images, the determination being based on the digital image T and on a reference digital image that is representative of the contaminating noise and obtained by the same sensor,

the determination being made from a decorrelation condition;

Correlation (T−

A*{the reference digital image}, {the reference digital image})=0,

the determination being made over pixels in some regions of T and the reference digital image, to estimate the amplitude of the noise for each of at least some pixels of T.

2. The method of claim 1 in which the sensor comprises a CMOS sensor.

3. The method of claim 1 in which the sensor comprises a CCD sensor.

4. The method of claim 1 in which the noise comprises dark current noise.

5. The method of claim 4 in which a magnitude of the dark current noise is estimated for every pixel of T to use in producing a dark current corrected version of T.

6. The method of claim 4 in which the reference digital image is based on a dark current digital image that is substantially free of vertical patterns and has been generated from a grey digital image and a black digital image acquired respectively using different exposure times.

7. The method of claim 4 in which the reference digital image is based on a corrected dark current digital image that has been processed to reduce the effect of low-frequency spatial trends across the pixels of the sensor.

8. The method of claim 4 in which the reference digital image is based on a de-correlation of a black digital image with a de-trended dark current digital image.

9. The method of claim 4 in which the reference digital image is obtained from the sensor without external light.

10. The method of claim 5 in which the dark current noise amplitude is estimated without requiring information about a temperature of the sensor or a duration of exposure.

11. The method of claim 5 also including reducing noise in every pixel of T using the estimated dark current noise magnitudes for all of the pixels to produce a noise reduced T.

12. The method of claim 11 also including providing the noise reduced T to a processor for use in analyzing features of an image captured by the sensor.

13. The method of claim 12 in which the images include possibly malignant lesions.

14. The method of claim 1 in which determining A comprises executing program instructions.

15. The method of claim 1 also comprising subtracting fixed vertical patterns, pixel by pixel, from T, to produce a vertical pattern corrected T.

16. The method of claim 15 also comprising applying a dark current removal function to the vertical pattern corrected T to produce a dark current corrected T.

17. The method of claim 16 also comprising applying an offset estimation and subtraction function to the dark current corrected T to remove offset.

18. The method of claim 1 in which the image sensor comprises a monochromatic sensor.

19. The method of claim 1 in which the image sensor comprises a sensor that images in two or more different spectral bands simultaneously.

20. The method of claim 19 in which the image sensor images in red, green, and blue bands (RGB).

21. The method of claim 1 in which the image sensor produces images in the visible or infrared ranges.

22. The method of claim 1 in which the image sensor produces images using X-ray or ultrasound radiation.

23. An apparatus comprising

a software processor configured to

determine an unknown amplitude A of noise that contaminates at least some of the pixels of a digital image T of an arbitrary target, the amplitude being possibly different in different digital images, the determination being based on the digital image T and on a reference digital image that is representative of the contaminating noise and obtained by the same sensor,

the determination being made from a decorrelation condition:

Correlation (T−

A*{the reference digital image}, {the reference digital image})=0,

the determination being made over pixels in some regions of T and the reference digital image, to estimate the amplitude of the noise for each of at least some pixels of T.

24. The apparatus of claim 23 also comprising a device to analyze T, in which noise has been reduced using the estimated noise amplitude, to determine a presence or absence of a possibly malignant lesion.

25. An apparatus comprising

means for receiving pixels of a digital image T, and

means for determining an unknown amplitude A of noise that contaminates at least some of the pixels of a digital image T of an arbitrary target, the amplitude being possibly different in different digital images, the determination being based on the digital image T and on a reference digital image that is representative of the contaminating noise and obtained by the same sensor,

the determination being made from a decorrelation condition:

Correlation (T−

A*{the reference digital image}, {the reference digital image})=0,

the determination being made over pixels in some regions of T and the reference digital image, to estimate the amplitude of the noise for each of at least some pixels of T.