SYSTEM AND METHOD OF USING IMPRINT ANALYSIS IN PILL IDENTIFICATION

• US 20140355849A1
• Filed: 06/04/2013
• Published: 12/04/2014
• Est. Priority Date: 06/04/2013
• Status: Abandoned Application

• Current Assignee: Indicator Sciences, LLC
• Original Assignee: Medsnap, LLC
• Inventors: Brossette, Stephen E., Zheng, Ning, Hymel, Patrick A. Jr.
• Application Number:
  US13/909,821
• Publication Number:
  US 20140355849A1
• Time in Patent Office: Days
• US Class Current: 382/128
• CPC Class Codes: G06F 18/28 Determining representative reference patterns, e.g. by averaging or distorting; Generating dictionaries
• CPC Class Codes: G06V 10/7515 Shifting the patterns to accommodate for positional errors
• CPC Class Codes: G06V 10/772 Determining representative reference patterns, e.g. averaging or distorting patterns; Generating dictionaries
• CPC Class Codes: G06V 20/66 Trinkets, e.g. shirt buttons or jewellery items recognising microscopic objects G06V20/69
• CPC Class Codes: G06V 2201/06 Recognition of objects for industrial automation
• CPC Class Codes: G16H 20/10 relating to drugs or medications, e.g. for ensuring correct administration to patients
• CPC Class Codes: G16H 70/40 relating to drugs, e.g. their side effects or intended usage

Abstract

A system and method for identifying a pill by its imprint. A digital pill imprint image for the pill is obtained and compared with one or more composite imprint images in a database. Each of the composite imprint images is a composite of two or more digital pill imprint images of a single type of pill. The composite imprint images are formed by aligning and combining the two or more digital pill imprint images for each type of pill. A match score is determined as a result of the comparing of the digital pill imprint image with each of the composite imprint images. The match score represents a degree of overlap between the digital pill imprint image and each composite imprint image. The pill is identified based on the composite imprint image having the best match score.

Claims

1. A method of creating a database of pill imprint images, the method comprising:

obtaining two or more digital pill imprint images for each type of a plurality of types of pills;

aligning the two or more digital pill imprint images for each type of pill;

combining the two or more digital pill imprint images for each type of pill into a single composite imprint image for each type of pill; and

storing the composite imprint image in a database such that the composite imprint image is associated with the corresponding type of pill.

2. The method of claim 1, wherein obtaining two or more digital pill imprint images for each type of pill includes obtaining digital pill imprint images from different pills of the same type for each of the plurality of types of pills.

3. The method of claim 1, wherein digital pill imprint images for each type of pill are obtained by normalizing the results of using edge-finding and adaptive threshold techniques.

4. The method of claim 1, further comprising converting the digital pill imprint images for each type of pill into binary pill imprint images.

5. The method of claim 1, further comprising determining a center of each of the digital pill imprint images.

6. The method of claim 5, wherein the center of each digital pill imprint image is a geometric center of a smallest-area rectangle bounding a contour of the pill imaged in each digital pill imprint image.

7. The method of claim 5, wherein the center of each digital pill imprint image is a center of mass based on a contour of the pill imaged in each digital pill imprint image.

8. The method of claim 5, wherein aligning the two or more digital pill imprint images for each type of pill further comprises selecting a seed digital pill imprint image from the two or more digital pill imprint images for each type of pill and aligning the centers of the other digital pill imprint images corresponding to the same type of pill with the center of the seed digital pill imprint image.

9. The method of claim 8, further comprising rotating about the aligned pill centers each of the other digital pill imprint images with respect to the seed digital pill image in order to maximize the overlap of the digital pill imprint images.

10. The method of claim 9, further comprising, for each rotation, shifting each of the other digital pill imprint images with respect to the seed digital pill image in order to maximize the overlap of the digital pill imprint images.

11. The method of claim 10, wherein the degree of overlap of each of the other digital pill imprint images with the seed pill imprint image is determined as a match score using at least one of the following:

a sum of squared pixel-wise differences, a sum of pixel-wise log likelihoods, correlation, and a correlation coefficient.

12. The method of claim 11, further comprising determining a rotation and shift resulting in a best match score for each of the other digital pill imprint images with respect to the seed digital pill imprint image.

13. The method of claim 12, wherein combining the two or more digital pill imprint images includes adding the pixel values of the seed digital pill imprint image and each of the other digital pill imprint images after each of the other digital pill imprint images are rotated and shifted to obtain the best match score.

14. The method of claim 13, further comprising normalizing the added images to the single composite imprint image for each type of pill.

15. A method of identifying a pill by its imprint, the method comprising:

obtaining a digital pill imprint image for the pill;

comparing the digital pill imprint image with one or more composite imprint images in a database, each of the composite imprint images being a composite of two or more digital pill imprint images of a single type of pill, said comparing including determining a match score for each compared composite imprint image, said match score representing a degree of overlap between the digital pill imprint image and each composite imprint image; and

identifying the pill based on the composite imprint image having the best match score.

16. The method of claim 15, wherein said comparing includes determining a center of the digital pill imprint image.

17. The method of claim 16, wherein the center of the digital pill imprint image is a geometric center of a smallest-area rectangle bounding a contour of the pill imaged in the digital pill imprint image.

18. The method of claim 16, wherein the center of the digital pill imprint image is a center of mass based on a contour of the pill imaged in the digital pill imprint image.

19. The method of claim 16, wherein comparing further comprises aligning the center of the digital pill image with centers of the one or more composite imprint images.

20. The method of claim 19, further comprising, for each composite imprint image with which the digital pill imprint image is aligned, rotating the digital pill imprint image about the aligned pill centers in order to maximize the overlap of the digital pill imprint image with each of the composite imprint images.

21. The method of claim 20, further comprising, for each rotation of the digital pill imprint image, shifting the digital pill imprint image with respect to each composite imprint image in order to maximize the overlap of the digital pill imprint image with each of the composite imprint images.

22. The method of claim 21, wherein the degree of overlap of the digital pill imprint image with each of the composite imprint images is determined as a match score using at least one of the following:

a sum of squared pixel-wise differences, a sum of pixel-wise log likelihoods, correlation, and a correlation coefficient.

23. The method of claim 22, further comprising determining a best match score for each composite imprint image with respect to the digital pill imprint image.

24. A storage device comprising:

at least a portion of a database which associates a plurality of different types of pills with a corresponding plurality of composite imprint images, each of the plurality of composite imprint images in the storage device being comprised of;

a combination of two or more digital pill imprint images added together.

25. The storage device of claim 24, wherein the two or more digital pill imprint images of each composite imprint image are aligned about their centers to form the corresponding composite imprint image.

26. The storage device of claim 25, wherein the center of each digital pill imprint image is a geometric center of a smallest-area rectangle bounding a contour of the pill imaged in each digital pill imprint image.

27. The storage device of claim 25, wherein the center of each digital pill imprint image is a center of mass based on a contour of the pill imaged in each digital pill imprint image.

28. The storage device of claim 25, wherein one of the two or more digital pill imprint images of each composite imprint image is a seed pill imprint image and the other digital pill imprint images are aligned with the seed pill imprint image by rotating and shifting the other digital pill imprint images about the aligned pill centers in order to maximize the overlap of the digital pill imprint images.

29. The storage device of claim 28, wherein the degree of overlap of each of the other digital pill imprint images with the seed pill imprint image is quantified as a match score using at least one of the following:

a sum of squared pixel-wise differences, a sum of pixel-wise log likelihoods, correlation, and a correlation coefficient.

30. The storage device of claim 29, wherein the two or more digital pill imprint images are combined when each of the other digital pill imprint images are aligned in correspondence with a best match score with respect to the seed pill imprint image.

31. The storage device of claim 24, wherein the storage device is a component in a mobile device.

32. The storage device of claim 24, wherein the database is configured to be accessed by a mobile device.

33. A system configured to identify a pill by its imprint, the system comprising:

a digital camera; and

a processor for receiving from the digital camera a digital image including at least one pill to be identified and for determining a digital pill imprint image for at least one of the at least one pill to be identified, the processor configured to use an imprint matching module to compare the digital pill imprint image with one or more composite imprint images in a database, each of the composite imprint images being a composite of two or more digital pill imprint images of a single type of pill.

34. The system of claim 33, wherein said imprint matching module is configured to determine a match score for each compared composite imprint image, said match score representing a degree of overlap between the digital pill imprint image and each composite imprint image.

35. The system of claim 34, wherein said imprint matching module is configured to determine a center of the digital pill imprint image.

36. The system of claim 35, wherein the center of the digital pill imprint image is a geometric center of a smallest-area rectangle bounding a contour of the pill imaged in the digital pill imprint image.

37. The system of claim 35, wherein the center of the digital pill imprint image is a center of mass based on a contour of the pill imaged in the digital pill imprint image.

38. The system of claim 35, wherein said imprint matching module is further configured to align the center of the digital pill image with centers of the one or more composite imprint images.

39. The system of claim 38, wherein said imprint matching module is further configured to, for each composite imprint image with which the digital pill image is aligned, rotate and shift the digital pill image about the aligned pill centers in order to maximize the overlap of the digital pill image with each of the composite imprint images.

40. The system of claim 39, wherein the degree of overlap of the digital pill image with each of the composite imprint images is determined as a match score using at least one of the following:

a sum of squared pixel-wise differences, a sum of pixel-wise log likelihoods, correlation, and a correlation coefficient.

41. The system of claim 40, wherein said imprint matching module is further configured to determine a best match score for each composite imprint image with respect to the digital pill image.