Correction of spatial artifacts in radiographic images

• US 9,752,995 B2
• Filed: 06/07/2012
• Issued: 09/05/2017
• Est. Priority Date: 06/07/2012
• Status: Active Grant

• Current Assignee: Varex Imaging Corporation
• Original Assignee: Varex Imaging Corporation
• Inventors: Holt, Kevin Matthew
• Primary Examiner(s): Mazumder, Tapas
• Application Number:
  US13/490,935
• Publication Number:
  US 20130328919A1
• Time in Patent Office: 1,916 Days
• CPC Class Codes: G01N 2223/3307 source and detector fixed; object moves
• CPC Class Codes: G01N 2223/401 image processing
• CPC Class Codes: G01N 2223/631 large structures, walls
• CPC Class Codes: G01N 23/04 and forming images of the material
• CPC Class Codes: G09G 2340/10 Mixing of images, i.e. displayed pixel being the result of an operation, e.g. adding, on the corresponding input pixels
• CPC Class Codes: G09G 2340/125 wherein one of the images is motion video

Abstract

The present invention encompasses methods, apparati, and computer-readable media for correcting spatial artifacts in composite radiographic images of an object (1). A method embodiment of the present invention comprises the steps of generating the composite image from a plurality of views of the object (1); estimating a shift profile; and re-generating the image by shifting the views using the shift profile.

Claims

1. A method for correcting spatial artifacts within a composite radiographic image of an object, said method comprising the steps of:

generating the composite image by assembling a plurality of views of the object imaged by a radiographic scanner, wherein each view represents a different time, relative motion exists between the radiographic scanner and the object, and each view corresponds to a portion of data used to generate the composite image and where each view is no more than two pixels wide in a horizontal direction and many pixels tall in a vertical direction to cover entire height of the object;

estimating a correctional shift profile comprising tailored correctional shifts for at least two of the views;

correcting the views using the corresponding correctional shifts; and

re-generating the image by assembling the corrected views.

2. The method of claim 1 wherein the correctional shift profile is represented by s[k], where k is a view number representing a time when the corresponding view was obtained.

3. The method of claim 1 further comprising performing, between the estimating step and the re-generating step, the step of applying a smoothing filter to the correctional shift profile.

4. The method of claim 1 wherein the re-generating step is accomplished by interpolation.

5. The method of claim 4 wherein the interpolation is from the group of interpolations consisting of bi-linear, cubic B-spline, and sinc.

6. The method of claim 1 wherein the step of estimating a correctional shift profile utilizes an optimization algorithm.

7. The method of claim 1 wherein the composite radiographic image is produced by x-rays.

8. The method of claim 1 wherein each view is one pixel wide and many pixels tall, the re-generating step comprises concatenating corrected views, and the concatenation is performed horizontally.

9. The method of claim 1 wherein more than one image is re-generated using the same correctional shift profile.

10. The method of claim 1 wherein fewer than all of the views are corrected.

11. The method of claim 10 comprising the following steps:

examining the image to identify one or more artifact-related features;

identifying one or more subsets of views relative to locations of the artifact-related feature(s); and

correcting only the identified subset(s) of views.

12. The method of claim 11 where the artifact-related features are wheels on a car or truck.

13. The method of claim 1 wherein the step of re-generating the image comprises concatenating the corrected views.

14. The method of claim 1 wherein the composite image is a two dimensional (2D) image, and each view is a portion of the 2D image.

15. The method of claim 1 wherein the correctional shifts correct a spatial artifact, and the spatial artifact is a wiggle that appears along one direction.

16. A method for correcting spatial artifacts within a composite radiographic image of an object, said method comprising the steps of:

generating the composite image by assembling a plurality of views of the object imaged by a radiographic scanner, wherein each view represents a different time, relative motion exists between the radiographic scanner and the object, and each view corresponds to a portion of data used to generate the composite image;

estimating a correctional shift profile comprising tailored correctional shifts for at least two of the views;

correcting the views using the corresponding correctional shifts; and

re-generating the image by assembling the corrected views;

wherein the step of estimating a correctional shift profile utilizes an optimization algorithm; and

the optimization algorithm is applied to find the correctional shift profile that minimizes a sum of a set of penalty functions such that the optimization algorithm at least approximately satisfies a set of hard constraints, wherein each hard constraint uses a Fourier transform to constrain high frequency components of the shift profile to be within a preselected value.

17. The method of claim 16 wherein the penalty function(s) comprise(s) at least one function from the group consisting of a squared error function, an absolute error function, and a Huber function.

18. The method of claim 16 wherein:

each penalty function and each constraint is from one of two classes of criteria;

a shift criterion class and an image criterion class;

the shift criterion class governs the shift profile expressed as a discrete function of time;

the image criterion class governs that the re-generated image is relatively free of wiggles; and

criteria from both classes are employed by the optimization algorithm.

19. The method of claim 18, wherein each class is represented by at least one penalty function or constraint.

20. The method of claim 18 wherein the shift criterion class is enforced by at least one step from the following:

using finite differences;

encouraging derivatives of the shift profile to be small;

hard penalizing a Fourier or other transform;

soft penalizing a Fourier or other transform;

using group sparsity.

21. The method of claim 18 wherein the image criterion class is satisfied by at least one step from the following:

performing an edge detection on the image, producing a corrected image, and encouraging edges in the corrected image to have a desired shape;

calculating a displacement field and encouraging some component of the shift profile to be similar to the displacement field;

encouraging the corrected image to be smooth.

22. The method of claim 21 wherein the step of encouraging the corrected image to be smooth comprises at least one of:

encouraging total energy in the gradient of the corrected image to be small;

using a Total Variation semi-norm;

using a Huber function.

23. The method of claim 16 wherein the optimization is solved using a convex programming solver.

24. At least one non-transitory computer-readable medium containing computer program instructions for correcting spatial artifacts within a composite radiographic image of an object, said instructions performing the steps of:

generating the composite image by assembling a plurality of views of the object imaged by a radiographic scanner, wherein each view represents a different time, relative motion exists between the radiographic scanner and the object, and each view corresponds to a portion of data used to generate the composite image and where each view is no more than two pixels wide in a horizontal direction and many pixels tall in a vertical direction to cover entire height of the object;

estimating a correctional shift profile comprising tailored correctional shifts to be applied to at least two of the views;

correcting the views using the corresponding correctional shifts; and

re-generating the image by assembling the corrected views.

25. Apparatus for correcting spatial artifacts within a composite radiographic image of an object, said apparatus comprising:

a radiographic image scanner adapted to produce a plurality of views of the object, wherein each view represents a different time, relative motion exists between the radiographic scanner and the object, and each view corresponds to a portion of data used to generate the composite image and where each view is no more than two pixels wide in a horizontal direction and many pixels tall in a vertical direction to cover entire height of the object;

coupled to the radiographic image scanner, means for estimating a correctional shift profile comprising tailored correctional shifts to be applied to at least two of the views;

coupled to the estimating means, means for correcting the views using the corresponding correctional shifts; and

coupled to the correcting means, means for assembling the corrected views to form a corrected image.

26. The apparatus of claim 25 wherein the estimating means includes an optimization algorithm that is applied to find the shift profile that minimizes a sum of a set of penalty functions such that the optimization algorithm at least approximately satisfies a set of hard constraints.