Method, apparatus and program for image processing

Patent metadata

Field	Value
Publication number	US-2025029215-A1
Application number	US-202418774004-A
Country	US
Kind code	A1
Filing date	Jul 16, 2024
Priority date	Jul 17, 2023
Publication date	Jan 23, 2025
Grant date	—

Abstract

A method of image processing. The method comprises obtaining a set of image data, the set being associated with one or more parameters representative of one or more image capture characteristics for the set and comprising pixel intensity values representing image pixels having respective pixel locations in an image. The method comprises, for a given pixel intensity value in the set: determining an estimated noise value based on at least: the one or more parameters associated with the set, and a representative intensity value derived from one or more pixel intensity values in the set. The method comprises associating the estimated noise value with the given pixel intensity value.

First claim

1 . A method of image processing comprising:
obtaining a set of image data, the set being associated with one or more parameters representative of one or more image capture characteristics for the set and comprising pixel intensity values representing image pixels having respective pixel locations in an image; and for a given pixel intensity value in the set:
determining an estimated noise value based on at least:
the one or more parameters associated with the set, and
a representative intensity value derived from one or more pixel intensity values in the set; and

associating the estimated noise value with the given pixel intensity value.

2 . The method of claim 1 , comprising at least one of:
based on the estimated noise value, determining whether a particular pixel intensity value in the set of image data comprises a compromised pixel intensity value; based on the estimated noise value, modifying at least one pixel intensity value in the set of image data; or providing the estimated noise value and at least a portion of the set of image data to a computer vision system.

3 . The method of claim 2 , comprising said modifying of at least one pixel intensity value in the set of image data, wherein said modifying comprises, based on the estimated noise value, either:
denoising the at least one pixel intensity value; or correcting the at least one pixel intensity value, wherein the at least one pixel intensity value comprises a defective pixel intensity value.

4 . The method of claim 1 , wherein the representative intensity value is derived from at least a plurality of proximate pixel intensity values in the set, the plurality of proximate pixel intensity values representing image pixels at pixel locations near the pixel location of a given image pixel that has the given pixel intensity value.

5 . The method of claim 4 , wherein the plurality of proximate pixel intensity values comprises a plurality of adjacent pixel intensity values representing image pixels at pixel locations adjacent to the pixel location of the given image pixel.

6 . The method of claim 4 , wherein
the representative intensity value is derived using a configured weighting determining a contribution from one of said proximate pixel intensity values, and a contribution from the given pixel intensity value to the representative value has an effective weighting which is lower than the configured weighting.

7 . The method of claim 6 , wherein the representative intensity value has substantially no effective contribution from the given pixel intensity value.

8 . The method of claim 4 , wherein
the representative intensity value is derived using a configured weighting determining a contribution from the given pixel intensity value, and a contribution from one of said proximate pixel intensity values to the representative intensity value has an effective weighting which is lower than the configured weighting.

9 . The method of claim 1 , wherein the representative intensity value is the given pixel intensity value.

10 . The method of claim 1 , wherein the one or more parameters comprise a parameter representing an exposure value used during image capture, and the estimated noise value is determined based on the parameter representing the exposure value.

11 . The method of claim 10 , wherein the estimated noise value is determined based on a square root of a function of the parameter representing the exposure value, and the parameter representing the exposure value is proportional to the exposure value.

12 . The method of claim 1 , wherein
the set of image data has been captured using a pixel array associated with the one or more image capture characteristics, the one or more parameters comprise a pixel array noise parameter representing noise dependent on the image capture characteristics, and the estimated noise value is determined based on the pixel array noise parameter.

13 . The method of claim 12 , wherein the pixel array noise parameter is representative of read noise and/or dark current occurring in the respective pixel array.

14 . The method of claim 1 , wherein the estimated noise value is determined based on a square root of a function of the representative intensity value, the function being a linear function of the representative intensity value or a linear function of the reciprocal of the representative intensity value.

15 . The method of claim 14 , wherein:
the estimated noise value is an estimated noise value for the given pixel intensity value, and determining the estimated noise value comprises using a first function σ=√{square root over ((a1+b)} or an approximation of the first function, wherein σ is the estimated noise value and the one or more parameters comprise a and b, or alternatively the estimated noise value is an estimated noise value for a square root of the given pixel intensity value, and determining the estimated noise value comprises using a second function σ′=√{square root over ((c/I+d))} or an approximation of the second function, wherein σ′ is the estimated noise value and the one or more parameters comprise c and d, wherein, in either alternative, I is the representative intensity value.

16 . The method of claim 15 , wherein the estimated noise value is an estimated noise value for the square root of the given pixel intensity value, and determining the estimated noise value comprises selecting a limiting noise value if the representative intensity value meets a high magnitude condition.

17 . The method of claim 1 , comprising obtaining a plurality of sets of image data including the set, each set being associated with one or more parameters representative of one or more, respective, image capture characteristics for the set and comprising pixel intensity values representing image pixels having respective pixel locations in an image.

18 . The method of claim 17 , comprising generating a composite image comprising a plurality of output pixels, the plurality of output pixels being associated with a plurality of output pixel intensity values, the output pixel intensity values having been determined from pixel intensity values of at least two of the plurality of sets.

19 . A non-transitory computer-readable storage medium comprising instructions which, when executed by a computer, cause the computer to:
obtain a set of image data, the set being associated with one or more parameters representative of one or more image capture characteristics for the set and comprising pixel intensity values representing image pixels having respective pixel locations in an image; and for a given pixel intensity value in the set:
determine an estimated noise value based on at least:
the one or more parameters associated with the set, and
a representative intensity value derived from one or more pixel intensity values in the set; and

associate the estimated noise value with the given pixel intensity value.

20 . An image processing system comprising a data processor and data storage, the data storage comprising instructions which, when executed by the data processor, cause the data processor to:
obtain a set of image data, the set being associated with one or more parameters representative of one or more image capture characteristics for the set and comprising pixel intensity values representing image pixels having respective pixel locations in an image; and for a given pixel intensity value in the set:
determine an estimated noise value based on at least:
the one or more parameters

Assignees

• Advanced Risc Mach Ltd

Inventors

• Talagala Dumidu Sanjaya
• Hanwell David

Classifications

• G06V10/30

  Noise filtering · CPC title

• G06T7/0002 Primary

  Inspection of images, e.g. flaw detection · CPC title

• G06T2207/30168

  Image quality inspection · CPC title

• G06V10/60

  relating to illumination properties, e.g. using a reflectance or lighting model · CPC title

• G06T5/70 Primary

  Denoising; Smoothing · CPC title