Methods and Apparatus for Detecting Defects in Imaging Arrays by Image Analysis

• US 20080218610A1
• Filed: 10/02/2006
• Published: 09/11/2008
• Est. Priority Date: 09/30/2005
• Status: Active Grant

• Current Assignee: Simon Fraser University
• Original Assignee: Simon Fraser University
• Inventors: Jung, Cory, Koren, Zahava, Koren, Israel, Chapman, Glenn Harrison, Dudas, Jozsef
• Granted Patent: US 8,009,209 B2
• Time in Patent Office: Days
• US Class Current: 348/246
• CPC Class Codes: H04N 17/002 for television cameras
• CPC Class Codes: H04N 25/68 applied to defects

Abstract

Methods for detecting defective pixels in imaging arrays involve establishing probabilities that individual pixels are defective and updating those probabilities by analysing images acquired by the imaging arrays. Probabilities may be evaluated for each of two or more defect conditions. The methods may be used to detect defects such as stuck-low, stuck-high, high-sensitivity, low sensitivity, hot, and defect-free conditions. Other more complicated defect conditions can also be detected. Apparatus for detecting defective pixels may be integrated with a camera or other imaging device or provided separately.

Claims

1. A method for detecting defects in imaging arrays comprising arrays of pixels, the method comprising,

obtaining a set of images acquired by an imaging array, each image comprising a set of pixel values corresponding to pixels of the image;

taking each of a plurality of images in the set as a current image and obtaining for at least one of the pixels, for each of one or more of a plurality of defect types, a corresponding updated probability, the updated probability based at least in part upon;

a pixel value corresponding to the at least one pixel;

statistics of pixel values of other pixels in the current image; and

one or more prior probabilities that the one of the pixels has the corresponding defect type.

2. A method according to claim 1 wherein, for at least one of the defect types, updating the probability that the at least one pixel has the defect type comprises obtaining image statistics for the current image and basing the updated probability at least in part on a model of the defect type and the image statistics.

3. A method according to claim 2 wherein the image statistics comprise a proportion of pixels in the current image that have pixel values in a range that includes the pixel value for the at least one pixel.

4. A method according to claim 3 wherein, for a defect-free defect type the probability that the pixel has the defect free defect type is proportional to the proportion of pixels in the current image that have pixel values in a range that includes the pixel value for the at least one pixel.

5. A method according claim 4 wherein updating the probability that the pixel has the defect type comprises applying Bayesian statistical inference.

6. A method according to claim 5 wherein, for at least one of the defect types, the model comprises an imaging parameter associated with the current image and the model is based at least in part on the imaging parameter.

7. A method according to claim 6 wherein the imaging parameter comprises an exposure time.

8. A method according to claim 1 comprising:

for a plurality of the images including the current image, acquiring a statistical distribution of the pixel values in the image;

wherein obtaining the updated probability for at least one of the plurality of defect types for the current image is based at least in part on the statistical distribution corresponding to the current image and a model of the at least one of the plurality of defect types.

9. A method according to claim 8 wherein the at least one pixel comprises a plurality of pixels.

10. A method according to claim 9 wherein the plurality of pixels comprises substantially all of the pixels in the images.

11. A method according to claim 9 comprising, prior to evaluating the probability that the at least one pixel has a defect type:

specifying a model behaviour for each defect type; and

for each defect type, establishing an initial probability for each pixel that the pixel has the defect type.

12. A method according to claim 11 wherein specifying the model behaviour for a defect type comprises providing a mathematical function which relates the response of pixels having each defect type to a stimulus strength.

13. A method according to claim 12 wherein the mathematical function is a function of at least one imaging parameter in addition to the stimulus strength.

14. A method according to claim 13 wherein the at least one imaging parameter comprises an exposure time.

15. A method according to claim 11 wherein the model behaviour for at least one of the defect types comprises a noise model.

16. A method according to claim 2 comprising establishing an initial probability for each of the plurality of defect types.

17. A method according to claim 16 wherein the initial probability is different for each of the plurality of defect types.

18. A method according to claim 16 wherein the initial probability is a function of pixel location for one or more of the plurality of defect types.

19. A method according to claim 8 wherein the pixel value for the at least one pixel is in a range from zero to a maximum value, and wherein acquiring a statistical distribution of the pixel values in the image comprises generating a histogram relating possible pixel values to the probability that a pixel will have that output value.

20. A method of claim 1 comprising, after processing each of the images, testing a condition for termination of the method and processing another image if the termination condition is not satisfied.

21. A method according to claim 20 comprising, if the termination condition is satisfied, establishing a defect map based upon the updated probabilities for one or more of the defect types.

22. A method according to claim 21 wherein the defect map specifies the location and defect type of each defective pixel,

23. A method according to claim 22 wherein the termination condition is satisfied when a predetermined number of the images have been processed.

24. A method according to claim 22 wherein the termination condition is satisfied when a probability of one of the defect types is greater than a threshold and greater than probabilities of the other defect types and has remained so during processing of a predetermined number of the images.

25. A method according to claim 1 comprising, for each pixel in the imaging array, identifying a defect type having the highest probability that exceeds a predetermined threshold as a defect type for the pixel.

26. A method according to claim 1 comprising acquiring one or more dark-field images using the imaging array and including the one or more dark-field images in the set of images.

27. A method according to claim 1 wherein the defect types include stuck-low and stuck-high defect types.

28. A method according to claim 1 wherein the defect types include high-sensitivity and low-sensitivity defect types.

29. A method according to claim 1 wherein the plurality of defect types comprise one or more of stuck-low, stuck-high, low-sensitivity, high-sensitivity and hot-pixel defect types.

30. A method according to claim 2 comprising performing interpolation to obtain an estimated value of a given pixel based on the values of other pixels and comparing the estimated value for the given pixel to the pixel value for the given pixel.

31. A method according to claim 30 comprising establishing a histogram of differences between the pixel value of each pixel and the estimated value for the pixel.

32. A method according to claim 31 comprising basing the probability that a pixel has one of the defect types at least in part on the difference between the estimated value and pixel value for that pixel.

33. A method according to claim 30 wherein obtaining the updated probability that the pixel has the one defect type comprises applying a compensation function corresponding to the defect type to the pixel value to yield a compensated pixel value, comparing the compensated and estimated pixel values, and determining from image statistics a proportion of pixels in the image having pixel values differing from estimated values in a range that includes a result of the comparison.

34. A method according to claim 33 wherein comparing the compensated and estimated pixel values comprises computing a difference between the compensated and estimated pixel values.

35. A method according to claim 33 wherein performing interpolation comprises excluding one or more pixels identified as being defective from the interpolation.

36. A method according to claim 30 wherein the image statistics comprise statistics regarding differences between the estimated and pixel values for pixels in the image.

37. A method according to claim 1 comprising basing the probability that a pixel has one of the defect types at least in part on a result of a comparison of the pixel value for that pixel and a pixel value for another pixel in the image.

38. A method according to claim 37 comprising basing the probability that a pixel has the one defect type at least in part on a sequence of comparisons of the pixel value for that pixel and pixel values for a sequence of other pixels in the image.

39. A method according to claim 37 wherein obtaining the updated probability that the pixel has the one defect type comprises:

applying a compensation function corresponding to the defect type to the pixel value to yield a compensated pixel value;

for each comparison, comparing the compensated pixel value to the pixel value for the other pixel in the image; and

for each comparison, determining from image statistics a proportion of pixels in the image having pixel values differing from pixel values of corresponding other pixels in a range that includes a result of the comparison.

40. A method according to claim 39 wherein obtaining an updated probability that the pixel has the one defect type comprises applying Bayesian inference to the results of the comparisons.

41. A method according to claim 40 wherein applying Bayesian inference comprises, for a plurality of the comparisons computing a conditional probability for the one defect type based upon a mathematical model associated with the defect type.

42. A method according to one of claim 37 wherein obtaining the updated probability comprises combining the conditional probabilities from the plurality of comparisons and weighting one or more of the conditional probabilities differently from others.

45. A method according to claim 1 comprising performing the method separately on each of a number of different blocks in the imaging array.

46. A method according to claim 1 comprising performing the method separately on each of a number of different or patterns of pixels in the imaging array.

47. A method according to claim 1 comprising comparing images in the set of images to ensure that the images are sufficiently different from one another.

61. Data processing apparatus comprising:

at least one data processor;

a data store containing a plurality of images obtained by an imaging array;

a program store containing software instructions which, when executed by the data processor, cause the data processor to execute a method according to claim 1 wherein the set of images comprises the plurality of images contained in the data store.

62. Apparatus according to claim 61 wherein the apparatus comprises a camera.

63. Apparatus according to claim 62 wherein the apparatus is configured to operate on new images as the new images are acquired.

64. Apparatus according to claim 61 comprising an editable defect map.

66. A method according to claim 1 wherein the corresponding defect type is characterized by one or more of:

a high sensitivity, a low sensitivity and an offset.

43. (canceled)

44. (canceled)

48. A method for categorizing a pixel of an imaging array into one of a plurality of defect types, the method comprising:

acquiring an image taken by the imaging array;

obtaining an image statistic for the image; and

,

determining a probability that the pixel has one of the defect types based at least in part upon a pixel value corresponding to the pixel in the image and the image statistic.

49. A method according to claim 48 wherein the image is one of a plurality of images acquired by the imaging array and the method comprises repeating the steps of claim 48 for each of the plurality of images and determining an updated probability that the pixel has the one of the defect types based upon the probability determined for each of the plurality of images.

50. A method according to claim 49 wherein determining an updated probability that the pixel has the one of the defect types based upon the probability determined for each of the images comprises applying an inference method to the to the probability determined for each of the plurality of images.

51. A method according to claim 50 wherein the inference method comprises a Bayesian inference method.

52. A method according to claim 48 wherein the image statistic comprises a proportion of pixels in the image or an area within the image having a value in a range of one or more values that includes the pixel value of the pixel.

53. A method according to claim 48 wherein the image statistic comprises a proportion of pixels in the image or an area within the image for which a difference between the pixel value and an interpolated estimate of the pixel value has a value in a range that includes the difference between the pixel value of the pixel and an interpolated estimate of the pixel value of the pixel.

54. A method for detecting defects in imaging arrays comprising an array of pixels, the method comprising:

acquiring a set of images taken by the imaging array during normal operation,

for each of the images, acquiring a statistical distribution of the pixel output values in the image; and

for each of a plurality of pixels, evaluating a probability that the pixel has a defect condition based at least in part on the pixel value and the statistical distribution.

55. A method according to claim 54 comprising establishing a plurality of pixel defect types to be identified and for each of the pixels, for each of the defect types, evaluating a probability that the pixel has the defect type.

56. A method according to claim 55 comprising, prior to evaluating the probability that the pixel has a defect condition:

specifying a model behaviour for each defect type;

for each defect type, establishing an initial probability for each pixel that the pixel has the defect type.

57. A method according to claim 56 wherein specifying the model behaviour for each defect type comprises providing a mathematical function which relates the response of each defect type to an intensity measured by an element of the imaging array.

58. A method according to claim 57 wherein the mathematical function is a function of at least one imaging parameter in addition to the intensity.

59. A method according to claim 58 wherein the at least one imaging parameter comprises an exposure time.

60. (canceled)

65. A method for determining sensitivities of elements in an imaging array comprising an array of imaging elements, the method comprising:

providing a plurality of sensitivity models, each corresponding to a different sensitivity;

acquiring a set of images taken by the imaging array, each image comprising an array of pixels;

for each of the images, acquiring a statistical distribution of pixel values in the image;

for each of the images creating a predicted statistical distribution for each of the sensitivity models based at least in part upon the statistical distribution and the sensitivity models; and

,

for each of the images obtaining a conditional probability that a pixel matches each of the sensitivity models based at least in part on the pixel value of the pixel and the predicted statistical distribution corresponding to the sensitivity model.