Method and apparatus for robust shape detection using a hit/miss transform

• US 5,790,691 A
• Filed: 03/12/1996
• Issued: 08/04/1998
• Est. Priority Date: 03/12/1996
• Status: Expired due to Fees

• Current Assignee: The Board of Regents of the University of Colorado (University of Colorado System)
• Original Assignee: The Board of Regents of the University of Colorado (University of Colorado System)
• Inventors: Narayanswamy, Ram, Johnson, Kristina M.
• Primary Examiner(s): JOHNS, ANDREW W
• Application Number:
  US08/614,852
• Time in Patent Office: 875 Days
• Field of Search: 382/128, 382/133, 382/205, 382/210, 382/211, 382/212, 382/213, 382/219, 382/308, 128/922
• US Class Current: 382/133
• CPC Class Codes: G06V 20/69 Microscopic objects, e.g. biological cells or cellular parts

Abstract

An apparatus and method for detecting anomalies morphologically detecting regions of interest. The apparatus and method include a region of interest (ROI) detection unit which morphologically detects anomalies in two dimensional input images or morphologically detects regions of interest based on structuring elements which can be set by a user using a modified hit and miss transform which includes a thresholding feature. The thresholding feature makes it possible to vary how much the shape of features within the input images can deviate from the shape of structuring elements and still be detected as a region of interest. One application of this involves detection of abnormal cells in a cervical Pap-smear.

Claims

1. An apparatus for detecting regions of interest in input data, comprising:

a hit correlator which includes a hit thresholder, said hit correlator receiving the input data, performing a hit transform on said input data using a hit kernel to yield hit data, thresholding the hit data and outputting resulting thresholded hit data;

a miss correlator which includes a miss thresholder, said miss correlator receiving the input data, performing a miss transform on said input data using a miss kernel to miss data, thresholding the miss data and outputting resulting thresholded miss data, wherein the hit and miss kernels are adapted to together yield hit and miss data corresponding to a range of shapes and/or sizes; and

a logically ANDing unit that logically ANDs said resulting thresholded hit data and said resulting thresholded miss data and outputs resulting region of interest data.

2. The apparatus as claimed in claim 1, wherein said input data is modulated onto optical radiation and said hit correlator comprises a spatial light modulator.

3. The apparatus as claimed in claim 2, wherein said input data is modulated onto optical radiation and said hit correlator comprises an electrically addressable spatial light modulator.

4. The apparatus as claimed in claim 2, wherein said input data is modulated onto optical radiation and said hit correlator comprises an optically addressable spatial light modulator.

5. The apparatus as claimed in claim 1, wherein said input data is modulated onto optical radiation and said miss correlator comprises a spatial light modulator.

6. The apparatus as claimed in claim 5, wherein said input data is modulated onto optical radiation and said miss correlator comprises an electrically addressable spatial light modulator.

7. The apparatus as claimed in claim 5, wherein said input data is modulated onto optical radiation and said miss correlator comprises an optically addressable spatial light modulator.

8. The apparatus of claim 1, further comprising a post-processing unit that performs image processing on the resulting region of interest data.

9. The apparatus of claim 1, wherein a size and/or shape of the hit kernel is adjustable.

10. The apparatus of claim 1, wherein a size and/or shape of the miss kernel is adjustable.

11. The apparatus of claim 1, wherein a size and/or shape of both the hit and the miss kernel are adjusted.

12. A system for identifying regions of interest, comprising:

a data acquisition unit for acquiring image data;

a data preprocessing unit for receiving said image data and processing said image data to yield processed image data;

a region of interest processing unit for receiving and performing a modified hit/miss transform on said processed image data, and for outputting resulting region of interest data, whereby said modified hit/miss transform incorporates a hit transform and a hit kernel that yield hit data, a miss transform and a miss kernel that yield miss data, and thresholding, wherein the hit and miss kernels are adapted to together yield hit and miss data corresponding to a range of shapes and/or sizes.

13. The system as claimed in claim 12, wherein said region of interest processing unit comprises a hit correlator and thresholder, which receives said processed image data and processes the hit transform and a hit threshold on said processed image data based on the hit kernel and a hit threshold value.

14. The system as claimed in claim 12, wherein said region of interest processing unit comprises a miss correlator and thresholder, which receives said processed image data and processes the miss transform and a miss threshold on said processed image data based on the miss kernel and a miss threshold value.

15. The system as claimed in claim 12, wherein said region of interest processing unit comprises:

a hit correlator and thresholder, which receives said processed image data and processes the hit transform and a hit threshold on said processed image data based on the hit kernel and a hit threshold value to yield a correlated and thresholded hit image; and

a miss correlator and thresholder, which receives said processed image data and processes the miss transform and a miss threshold on said processed image data based on the miss kernel and a miss threshold value to yield a correlated and thresholded miss image.

16. The system as claimed in claim 15, wherein said region of interest processing unit further comprises a logical ANDing unit coupled to said hit correlator and thresholder and to said miss correlator and thresholder which receives and logically ANDs the correlated and thresholded hit image and the correlated and thresholded miss image to yield the resulting region of interest data.

17. The system of claim 12, wherein the region of interest processing unit comprises a post-processing unit that performs image processing on the resulting region of interest data.

18. The apparatus of claim 12, wherein a size and or/shape of the hit kernel is adjustable.

19. The apparatus of claim 12, wherein a size and/or shape of the miss kernel is adjustable.

20. The apparatus of claim 12, wherein a size and/or shape of both the hit and the miss kernel are adjustable.

21. A method for detecting regions of interest in input data, comprising:

receiving the input data;

performing a hit transform on said input data using a hit kernel to yield hit data;

performing a miss transform on said input data using a miss kernel to yield miss data, wherein the hit and miss kernels are adapted to together yield hit and miss data corresponding to a range of shapes and/or sizes;

thresholding the hit and miss data to yield resulting thresholded hit data and resulting thresholded miss data; and

logically ANDing said resulting thresholding hit data and said resulting thresholded miss data and outputting resulting region of interest data.

22. The method of claim 21, further comprising the step of performing image processing the resulting region of interest data.

23. The method of claim 21, wherein the hit and miss kernels are adapted by adjusting a size and/or shape of the hit kernel.

24. The method of claim 21, wherein the hit and miss kernels are adapted by adjusting a size and/or shape of the miss kernel.

25. The method of claim 21, wherein the hit and miss kernels are adapted by adjusting a size and/or shape of both the hit and miss kernels.

26. A method for identifying regions of interest, comprising:

acquiring image data based on radiation scattered from a view;

performing a hit correlation using a hit kernel to yield hit transform data;

thresholding said hit transform data to yield modified hit transform data of the view;

performing a miss correlation using a miss kernel to yield miss transform data;

thresholding said miss transform data to yield modified miss transform data of the view; and

outputting data approximately common to said modified hit transform data and said modified miss transform data to yield resulting region of interest data, wherein the hit and miss kernels are adapted to together yield hit and miss transform data corresponding to a range of shapes and/or sizes.

27. The method as claimed in claim 26, wherein said thresholding said hit transform data step comprises retaining only said hit transform data greater than a hit threshold level.

28. The method as claimed in claim 26, wherein said thresholding said miss transform data step comprises retaining only said miss transform data greater than a miss threshold level.

29. The method as claimed in claim 26, wherein said step of thresholding said hit transform data comprises retaining only said hit transform data greater than a hit threshold level, and wherein said thresholding said miss transform data step comprises retaining only said miss transform data greater than a miss threshold level.

30. The method of claim 26, further comprising the step of performing image processing on the resulting region of interest data.

31. The method of claim 26, wherein the hit and miss kernels are adapted by adjusting a size and/or shape of the hit kernel.

32. The method of claim 26, wherein the hit and miss kernels are adapted by adjusting a size and/or shape of the miss kernel.

33. The method of claim 26, wherein the hit and miss kernels are adapted by adjusting a size and/or shape of both the hit and miss kernels.

34. A method for identifying regions of interest, comprising:

acquiring image data based on radiation scattered from a view;

performing a hit correlation using a hit kernel to yield hit transform data;

thresholding said hit transform data to yield modified hit transform data of the view;

performing a miss correlation using a miss kernel to yield miss transform data;

thresholding said miss transform data to yield modified miss transform data of the view; and

logically ANDing said modified hit transform data and said modified miss transform data to yield resulting region of interest data, wherein the hit and miss kernels are adapted to together yield hit and miss transform data corresponding to a range of shapes and/or sizes.

35. The method of claim 34, further comprising the step of performing image processing on the resulting region of interest data.

36. An apparatus for detecting regions of interest in a Pap-smear slide, comprising:

an optical hit correlator which includes a hit thresholder, said optical hit correlator receiving optical radiation with input data modulated thereon by passing through the Pap-smear slide, performing a hit transform on said optical radiation using a hit kernel to yield optical radiation modulated with hit data, thresholding the hit-data-modulated optical radiation and outputting optical radiation with resulting thresholded hit data modulated thereon;

an optical miss correlator which includes a miss thresholder, said optical miss correlator also receiving said optical radiation with input data modulated thereon, performing a miss transform on said optical radiation using a miss kernel to yield a optical radiation modulated with miss data, thresholding the miss-data-modulated optical radiation and outputting optical radiation with resulting thresholded miss data modulated thereon, wherein the hit and miss kernels are adapted to together yield hit-data-modulated and miss-data-modulated optical radiation corresponding to a range of shapes and/or sizes; and

a logical ANDing unit that logically ANDs said optical radiation with resulting thresholded hit data modulated thereon and said optical radiation with resulting thresholded miss data modulated thereon and outputs optical radiation with resulting region of interest data modulated thereon.

37. The apparatus of claim 36, further comprising a post-processing unit that performs image processing on the resulting region of interest data.

38. A method for detecting regions of interest in a Pap-smear slide, comprising:

passing optical radiation through the Pap-smear slide to yield modulated optical radiation with input data modulated thereon;

receiving and performing a hit transform on said optical radiation using a hit kernel to yield optical radiation modulated with hit data, thresholding the hit-data-modulated optical radiation and outputting optical radiation with thresholded hit data modulated thereon;

receiving and performing a miss transform on said optical radiation using a miss kernel to yield optical radiation modulated with miss data, thresholding the miss-data-modulated optical radiation and outputting optical radiation with thresholded miss data modulated thereon, wherein the hit and miss kernels are adapted to together yield hit-data-modulated and miss-data-modulated optical radiation corresponding to a range of shapes and/or sizes; and

logically ANDing said thresholded hit-data-modulated optical radiation and said thresholded miss-data-modulated optical radiation and outputting optical radiation with resulting region of interest data modulated thereon.

39. The method of claim 38, further comprising the step of performing image processing on the resulting region of interest data.