Method and Product for Detecting Abnormalities

• US 20090022391A1
• Filed: 06/22/2006
• Published: 01/22/2009
• Est. Priority Date: 07/01/2005
• Status: Active Grant

• Current Assignee: Abb Oy (ABB Limited)
• Original Assignee: Abb Oy (ABB Limited)
• Inventors: Huotilainen, Tommi
• Granted Patent: US 8,000,501 B2
• Time in Patent Office: Days
• US Class Current: 382/141
• CPC Class Codes: G01N 21/8983 for testing textile webs, i.e. woven material
• CPC Class Codes: G06T 2207/30124 Fabrics; Textile; Paper
• CPC Class Codes: G06T 7/0004 Industrial image inspection

Abstract

A new method for processing image data in order to detect abnormalities in a web is provided. The web is monitored by at least one camera, whereby an image comprising of plurality of pixels is generated. The data of the image is stored in a memory. Image data is filtered by a processor by creating a filtered image data by weighting the image data and at least one of earlier image data and earlier filtered image data; and combining the weighted image data and at least one of the weighted earlier image data and the weighted earlier filtered image data; and controlling filtering by at least one nonlinear algorithm; and thresholding the created filtered image data.

Claims

1. A method for processing image data in order to detect abnormalities in a web, wherein the web is monitored by at least one camera, whereby at least one image comprising of plurality of pixels is generated for creating the image data, and the image data is stored in a memory, wherein the image data (x(n)) is filtered by a processor for creating a filtered image data (y(n)), wherein the current filtered image data (y(n)) is obtained by weighting the current image data (x(n), a₀, b₀) and at least one of image data (x(n)) from emphasis angles and filtered image data (y(n)) from emphasis angles;

and by combining the obtained weighted current image data and the obtained at least one of weighted image data from emphasis angles and weighted filtered image data from emphasis angles, wherein the emphasis angles are dependent on the result of previous filtering; and

wherein the filtering of the current image data (x(n)) is controlled by at least one nonlinear algorithm; and

followed by thresholding the created filtered image data (y(n)).

2. A method according to claim 1, wherein the image data (x(n)) and/or filtered image data (y(n)) is weighted from any emphasis angle of 0°

-135°

.

3. A method according to claim 1, wherein the image data (x(n)) and/or filtered image data (y(n)) is weighted from any emphasis angle of 0°

or 45°

or 90°

or 135°

.

4. A method according to claim 2, wherein the emphasis angle range of 0°

-180°

is covered with two parallel filters.

5. A method according to claim 1, wherein the emphasis angle of the image data (x(n)) and/or filtered image data (y(n)) is independent from emphasis angles of previous filtering.

6. A method according to claim 1, wherein the emphasis angle of the image data (x(n)) and/or filtered image data (y(n)) is dependent on emphasis angles of previous filtering.

7. A method according to claim 1, wherein the emphasis angle of the image data (x(n)) and/or filtered image data (y(n)) is chosen from within a predetermined sector of permitted emphasis angles or is chosen from within the sector of permitted emphasis angles of previous filtering.

8. A method according to claim 1, wherein the emphasis angle of the image data (x(n)) and/or filtered image data (y(n)) is chosen from within a sector of permitted emphasis angles, and the sector is more narrow than the sector of permitted emphasis angles of previous filtering.

9. A method according to claim 1, wherein the image data (x(n)) and/or filtered image data (y(n)) is from the local neighbourhood of the current image data (x(n)) and/or current filtered image data (y(n)).

10. A method according to claim 1, wherein the filtering is weighted based on image data (x(n)) of previous filtering.

11. A method according to claim 1, wherein the filtering is weighted based on filtered image data (y(n)) of previous filtering.

12. A method according to claim 1, wherein the filtering is weighted based on at least one parameter.

13. A method according to claim 1, wherein the method further comprises utilizing of side information data and the side information data is stored in a memory.

14. A method according to claim 13, wherein the nonlinear adaptive algorithm includes side information data.

15. A method according to claim 13, wherein the filtering is controlled based on side information data from the previous stage of the filtering process.

16. A method according to claim 13, wherein the filtering is controlled based on side information data from another filter of the sequential and/or parallel and/or matrix filter structure.

17. A method according to claim 13, wherein side information data is created based on the image data (x(n)) and/or filtered image data (y(n)).

18. A method according to claim 13, wherein the side information data is created based on the side information data from the previous stages of filtering processes.

19. A method according to claim 13, wherein the side information data is created based on at least one parameter.

20. A method according to claim 13, wherein the side information data is created based on the side information data of previous filter.

21. A method according to claim 13, wherein the side information data is fixed to a predetermined value.

22. A method according to claim 13, wherein the side information data includes the direction of the image data (x(n)) from previous image data locations and/or the direction of the filtered image data (y(n)) from previous filtered image data locations.

23. A method according to claim 1, wherein the filtering is controlled based on a nonlinear adaptive algorithm utilizing at least one of maximum and minimum and median function.

24. A method according to claim 1, wherein the web monitoring camera is a line scan camera.

25. A method according to claim 1, wherein the web monitoring camera is a matrix camera.

26. A method according to claim 1, wherein the image is preprocessed before generating the image data (x(n)).

27. A method according to claim 26, wherein preprocessing rotates and/or reverses the image before generating the image data (x(n)).

28. A method according to claim 27, wherein preprocessing reverses the image in the cross direction of the web before generating the image data (x(n)).

29. A computer program product stored on computer operable media for processing image data in order to detect abnormalities in a web, wherein the web is monitored by at least one camera, whereby at least one image comprising of plurality of pixels is generated for creating the image data (x(n)), and the image data (x(n)) is stored in a memory, wherein the image data (x(n)) is filtered by a processor for creating a filtered image data (y(n)), wherein the current filtered image data (y(n)) is obtained by weighting the current image data (x(n), a₀, b₀) and at least one of image data (x(n)) from emphasis angles and filtered image data (y(n)) from emphasis angles;

and by combining the obtained weighted current image data and the obtained at least one of weighted image data from emphasis angles and weighted filtered image data from emphasis angles, wherein the emphasis angles are dependent on the result of previous filtering; and

wherein the filtering of the current image data (x(n)) is controlled by at least one nonlinear algorithm; and

followed by thresholding the created filtered image data (y(n)).