Processing images for removal of artifacts

• US 5,828,725 A
• Filed: 06/23/1997
• Issued: 10/27/1998
• Est. Priority Date: 07/03/1996
• Status: Expired due to Fees

• Current Assignee: Eliav Medical Imaging Systems Limited
• Original Assignee: Eliav Medical Imaging Systems Limited
• Inventors: Levinson, Reuven
• Primary Examiner(s): Porta, David P.
• Assistant Examiner(s): Bruce, David Vernon
• Application Number:
  US08/880,529
• Time in Patent Office: 491 Days
• Field of Search: 378/62, 378/98, 378/901, 382/132, 382/275
• US Class Current: 378/98
• CPC Class Codes: A61B 6/5258 involving detection or reduction of artifacts or noise
• CPC Class Codes: G06T 5/20 using local operators

Abstract

A method and apparatus for processing images received on an image sensor of rectangular configuration bounded by four outer edges and including a plurality of pixel elements arranged in a rectangular matrix includes the following operations: A. locating an initial rectangular-matrix region consisting of pixel elements determined to be free of an artifact; B. sequentially testing each pixel element along two contiguous outer edges of the image sensor, starting from the initial rectangular-matrix region, to determine whether the pixel value of the tested pixel element is sufficiently large to indicate the probability of an artifact, and for each such pixel element found to have such a pixel value, replacing its pixel value with another having a relation to the pixel value of its closest neighbours found to be free of an artifact, thereby cleaning the respective pixel element; and C. cleaning in a similar manner each of the remaining pixel elements.

Claims

1. A method of processing images received on an image sensor of rectangular configuration bounded by four outer edges and including a plurality of pixel elements arranged in a rectangular matrix of horizontal rows and vertical columns, comprising the following operations:

A. locating an initial rectangular-matrix region consisting of pixel elements determined to be free of an artifact;

B. sequentially testing each pixel element along two contiguous outer edges of the image sensor, starting from said initial rectangular-matrix region, to determine whether the pixel value of the tested pixel element is sufficiently large to indicate the probability of an artifact, and for each such pixel element found to have such a pixel value, replacing its pixel value with another having a relation to the pixel value of its closest neighbours found to be free of an artifact, thereby cleaning the respective pixel element; and

C. testing each of the remaining pixel elements to determine whether its pixel value is sufficiently large to indicate the probability of an artifact, and for each such pixel element found to have such a pixel value, replacing its pixel value with another pixel value having a predetermined relationship to the pixel values of its closest neighbours, thereby cleaning the respective pixel element.

2. The method according to claim 1, wherein Operation A is performed by:

(a) calculating the standard deviation of the pixel values in a selected initial rectangular-matrix region of the image sensor;

(b) determining whether said calculated standard deviation exceeds a predetermined threshold;

(c) if not, determining said initial rectangular-matrix region to be free of an artifact; and

(d) if said calculated standard deviation is determined to exceed said threshold, incrementing the selected initial rectangular-matrix region by one pixel element, and repeating (a), (b) and (c) until an initial rectangular-matrix region is determined to be free of an artifact.

3. The method according to claim 2, wherein said selected initial rectangular-matrix region is at a corner of the image sensor.

4. The method according to claim 3, wherein said selected initial rectangular matrix region is at the left-hand corner of the image sensor.

5. The method according to claim 1, wherein said Operation B includes the steps:

(a) determining whether the pixel value of each tested pixel element bears a predetermined relationship with respect to clean neighbouring pixel elements, and

(b) if not, replacing the pixel value of the tested pixel element by a pixel value having another predetermined relationship with respect to its clean neighbouring pixel elements.

6. The method according to claim 5, wherein said predetermined relationship of step (a) is whether the pixel value of the tested pixel element is equal to or less than the pixel value of any one of a number of clean neighbouring pixel elements plus a predetermined threshold, and the predetermined relationship of step (b) is the average of said number of clean neighbouring pixel elements.

7. The method according to claim 6, wherein said number of clean neighbouring pixel elements in both step (a) and step (b) is "2".

8. The method according to claim 1, wherein said Operation C includes:

(a) determining, in a first step, whether the pixel value of each tested pixel element bears a first predetermined relationship to the four clean neighbouring pixel elements; and

if not

(b) determining, in a second test more detailed than the first test, whether the pixel value of the tested pixel element bears a second relationship to all eight of the neighbouring pixel elements; and

if not

(c) replacing the pixel value of the tested pixel element with the average of the pixel values of at least some of said eight neighbouring pixel elements involved in said second test.

9. The method according to claim 8, wherein, in testing all said pixel elements in Operation C except for those in the remaining two contiguous outer edges of the image sensor,

(a) said first test in step (a) is whether the pixel value of the tested pixel element is equal to or less than the average of the four clean neighbouring pixel elements plus a predetermined threshold;

(b) said second test in step (b) is whether the pixel value of the tested pixel elements is equal to or less than the average of the four lowest-value pixel elements of all eight neighbouring pixel elements plus a predetermined threshold;

(c) and in testing the pixel elements in Operation C in said remaining two contiguous outer edges of the image sensor, said first and second predetermined relations are as set forth in (a) and (b) above, except that the relationship in (a) involves three clean neighbouring pixel elements, and the relationship in (b) involves five neighbouring pixel elements.

10. The method according to claim 1, wherein said initial rectangular-matrix region is a matrix of 5×

5 pixel elements.

11. Imaging apparatus comprising:

an image sensor of rectangular configuration bounded by four outer edges and including a plurality of pixel elements arranged in a rectangular matrix of horizontal rows and vertical columns;

and a data processor programmed to perform the following operations;

A. locating an initial rectangular-matrix region consisting of pixel elements determined to be free of an artifact;

B. sequentially testing each pixel element along two contiguous outer edges of the image sensor, starting from said initial rectangular-matrix region, to determine whether the pixel value of the tested pixel element is sufficiently large to indicate the probability of an artifact, and for each such pixel element found to have such a pixel value, replacing its pixel value with another having a relation to the pixel value of its closest neighbours found to be free of an artifact, thereby cleaning the respective pixel element; and

C. testing each of the remaining pixel elements to determine whether its pixel value is sufficiently large to indicate the probability of an artifact, and for each such pixel element found to have such a pixel value, replacing its pixel value with another pixel value having a predetermined relationship to the pixel values of its closest neighbours, thereby cleaning the respective pixel element.

12. The apparatus according to claim 11, wherein said data processor is programmed to perform said Operation A by:

(a) calculating the standard deviation of the pixel values in a selected initial rectangular-matrix region of the image sensor;

(b) determining whether said calculated standard deviation exceeds a predetermined threshold;

(c) if not, determining said initial rectangular-matrix region to be free of an artifact; and

(d) if said calculated standard deviation is determined to exceed said threshold, incrementing the selected initial rectangular-matrix region by one pixel element, and repeating (a), (b) and (c) until an initial rectangular-matrix region is determined to be free of an artifact.

13. The apparatus according to claim 12, wherein said selected initial rectangular-matrix region is at a corner of the image sensor.

14. The apparatus according to claim 13, wherein said selected initial rectangular matrix region is at the left-hand corner of the image sensor.

15. The apparatus according to claim 11, wherein said data processor is programmed to perform Operation B by:

(a) determining whether the pixel value of each tested pixel element bears a predetermined relationship with respect to clean neighbouring pixel elements, and

(b) if not, replacing the pixel value of the tested pixel element by a pixel value having another predetermined relationship with respect to its clean neighbouring pixel elements.

16. The appartus according to claim 15, wherein said predetermined relationship of step (a) is whether the pixel value of the tested pixel element is equal to or less than the pixel value of any one of a number of clean neighbouring pixel elements plus a predetermined threshold, and the predetermined relationship of step (b) is the average of said number of clean neighbouring pixel elements.

17. The apparatus according to claim 16, wherein said number of clean neighbouring pixel elements in both step (a) and step (b) is "2".

18. The apparatus according to claim 11, wherein said data processor is programmed to perform said Operation C by:

(a) determining, in a first step, whether the pixel value of each tested pixel element bears a first predetermined relationship to the four clean neighbouring pixel elements; and

if not

(b) determining, in a second test more detailed than the first test, whether the pixel value of the tested pixel element bears a second relationship to all eight of the neighbouring pixel elements; and

if not

(c) replacing the pixel value of the tested pixel element with the average of the pixel values of at least some of said eight neighbouring pixel elements involved in said second test.

19. The apparatus according to claim 18, wherein, in testing all said pixel elements in Operation C except for those in the remaining two contiguous outer edges of the image sensor,

(a) said first test in step (a) is whether the pixel value of the tested pixel element is equal to or less than the average of the four clean neighbouring pixel elements plus a predetermined threshold;

(b) said second test in step (b) is whether the pixel value of the tested pixel elements is equal to or less than the average of the four lowest-value pixel elements of all eight neighbouring pixel elements plus a predetermined threshold;

(c) and in testing the pixel elements in Operation C in said remaining two contiguous outer edges of the image sensor, said first and second predetermined relations are as set forth in (a) and (b) above, except that the relationship in (a) involves three clean neighbouring pixel elements, and the relationship in (b) involves five neighbouring pixel elements.

20. The apparatus according to claim 11, wherein said imaging apparatus is X-ray imaging apparatus.