Method and system for dual energy image registration

1. A method for registering a set of dual energy images comprising first and second images, comprising:

generating first and second Gaussian pyramids for said first and second images, respectively, each of said first and second Gaussian pyramids comprising a plurality of pyramid images, each corresponding to one of a plurality of pyramid levels; initializing an optical flow value for a first pyramid level;

sequentially updating the optical flow value for each pyramid level based on the corresponding pyramid images of the first and second Gaussian pyramids based on an optimization function comprising a similarity measure and a regularizer; and

registering said first and second images based on a final optical flow value.

2. The method of claim 1 , further comprising:

preprocessing said first and second images to detect edges in said first and second images prior to said step of generating first and second Gaussian pyramids.

3. The method of claim 2 , wherein said step of preprocessing said first and second images comprises:

normalizing a magnitude of gradients in said first and second images;

filtering said first and second images using a binarized Laplacian of Gaussian (LoG) filter; and

smoothing said first and second images using one of a Gaussian filter and a nonlinear isotropic diffusion filter.

4. The method of claim 2 , wherein said step of preprocessing said first and second images comprises:

preprocessing said first and second images using at least one of a normalization method, a normalized gradient magnitude method, a Harris corner detector, a normalized Laplacian of Gaussian (LoG) filter, a binarized LoG filter, a Gaussian filter, and a nonlinear isotropic diffusion filter.

5. The method of claim 1 , wherein said step of initializing an optical flow value for a first pyramid level comprises:

initializing an initial value of said optical flow value as zero for said first pyramid level.

6. The method of claim 1 , wherein said step of sequentially updating the optical flow value for each pyramid level based on the corresponding pyramid images of the first and second Gaussian pyramids comprises, for each pyramid level:

iteratively updating an initial value of said optical flow value to optimize the optical flow value for the pyramid images corresponding to the pyramid level based on variational gradients of said similarity measure and said regularizer.

7. The method of claim 6 , wherein the initial value of said optical flow value for the first pyramid level is determined in said step of initializing an optical flow value for a first pyramid level, and the initial of the optical flow value for each subsequent pyramid level is the updated optical flow value for a previous pyramid level.

8. The method of claim 1 , wherein said similarity measure is one of a sum squared difference similarity measure, a mutual information similarity measure, and a cross correlation similarity measure.

9. The method of claim 1 , wherein said regularizer is one of a Laplacian regularizer and a Nagel-Enkelmann regularizer.

10. The method of claim 1 , wherein said optimization function is an Euler-Lagrange equation comprising a first term representing said similarity measure and a second term representing said regularizer.

11. The method of claim 1 , further comprising, prior to said step of sequentially updating the optical flow value for each pyramid level based on the corresponding pyramid images of the first and second Gaussian pyramids, the steps of:

selecting said similarity measure as one of a sum squared difference similarity measure, a mutual information similarity measure, and a cross correlation similarity measure;

selecting said regularizer as one of a Laplacian regularizer and a Nagel-Enkelmann regularizer; and

discretizing said similarity measure and said regularizer.

12. The method of claim 1 , wherein said step of registering said first and second images based on a final optical flow value comprises:

deforming said second image based on,said final optical flow value to compensate for motion between said first and second images.

13. The method of claim 1 , further comprising:

weighting and subtracting said first and second images to generate a soft tissue image and a bone image.

14. An apparatus for registering a set of dual energy images comprising first and second images, comprising:

means for generating first and second Gaussian pyramids for said first and second images, respectively, each of said first and second Gaussian pyramids comprising a plurality of pyramid images, each corresponding to one of a plurality of pyramid levels;

means for initializing an optical flow value for a first pyramid level; means for sequentially updating the optical flow value for each pyramid level based on the corresponding pyramid images of the first and second Gaussian pyramids based on an optimization function comprising a similarity measure and a regularizer; and

means for registering said first and second images based on a final optical flow value.

15. The apparatus of claim 14 , further comprising:

means for preprocessing said first and second images to detect edges in said first and second images.

16. The apparatus of claim 14 , wherein said means for sequentially updating the optical flow value for each pyramid level based on the corresponding pyramid images of the first and second Gaussian pyramids comprises:

means for iteratively updating an initial value of said optical flow value for each pyramid level to optimize the optical flow value for the pyramid images corresponding the pyramid level based on variational gradients of said similarity measure and said regularizer.

17. The apparatus of claim 16 , wherein the initial value of said optical flow value for the first pyramid level is zero, and the initial of the optical flow value for each subsequent pyramid level is the updated optical flow value for a previous pyramid level.

18. The apparatus of claim 14 , wherein said optimization function is an Euler-Lagrange equation comprising a first term representing said similarity measure and a second term representing said regularizer.

19. The apparatus of claim 14 , wherein said means for registering said first and second images based on a final optical flow value comprises:

means for deforming said second image based on said final optical flow value to compensate for motion between said first and second images.

20. A non-transitory computer readable medium encoded with computer executable instructions for registering a set of dual energy images comprising first and second images, said computer executable instructions defining steps comprising:

generating first and second Gaussian pyramids for said first and second images, respectively, each of said first and second Gaussian pyramids comprising a plurality of pyramid images, each corresponding to one of a plurality of pyramid levels; initializing an optical flow value for a first pyramid level;

sequentially updating the optical flow value for each pyramid level based on the corresponding pyramid images of the first and second Gaussian pyramids based on an optimization function comprising a similarity measure and a regularizer; and

registering said first and second images based on a final optical flow value.

21. The computer readable medium of claim 20 , further comprising computer executable instructions defining the step of:

preprocessing said first and second images to detect edges in said first and second images prior to said step of generating first and second Gaussian pyramids.

22. The computer readable medium of claim 21 , wherein the computer executable instructions defining the step of preprocessing said first and second images comprise computer executable instructions defining the steps of:

normalizing a magnitude of gradients in said first and second images;

filtering said first and second images using a binarized Laplacian of Gaussian (LoG) filter; and

smoothing said first and second images using one of a Gaussian filter and a nonlinear isotropic diffusion filter.

23. The computer readable medium of claim 20 , wherein the computer executable instructions defining the step of sequentially updating the optical flow value for each pyramid level based on the corresponding pyramid images of the first and second Gaussian pyramids comprises computer executable instructions defining, for each pyramid level, the step of:

iteratively updating an initial value of said optical flow value to optimize the optical flow value for the pyramid images corresponding to the pyramid level based on variational gradients of said similarity measure and said regularizer.

24. The computer readable medium of claim 23 , wherein the initial value of said optical flow value for the first pyramid level is determined in said step of initializing an optical flow value for a first pyramid level, and the initial of the optical flow value for each subsequent pyramid level is the updated optical flow value for a previous pyramid level.

25. The computer readable medium of claim 20 , further comprising computer executable instructions defining the steps of:

selecting said similarity measure as one of a sum squared difference similarity measure, a mutual information similarity measure, and a cross correlation similarity measure;

selecting said regularizer as one of a Laplacian regularizer and a Nagel-Enkelmann regularizer; and

discretizing said similarity measure and said regularizer.