Apparatus and method for reconstructing an image using high-energy-based data
View Patent ↗An apparatus reconstructs an image using high energy-based data regarding a scene provides access to first image-capture data regarding the scene that is formed using a first image-capture modality and to second image-capture data that is formed using a second (different) image-capture modality. A control circuit executes an iterative image reconstruction process that establishes a first and a second image-representation channel, a fidelity error measure that measures inconsistency of the image as compared to first image-capture data and second image-capture data, and a prior-penalty term that scores the image based on a priori likelihood or desirability using, at least in part, for each of a plurality of pixels, a non-separable matrix-penalty of a Jacobian-matrix of the image at that pixel, such as nuclear norm. The control circuit further utilizes, in combination with the foregoing, an iterative process to generate a reconstructed image that at least approximately minimizes a combination of the fidelity error measure and the prior-penalty term.
1. An apparatus for reconstructing an image using high energy-based data regarding a scene, the apparatus comprising:
a data interface configured to provide access to:
first image-capture data regarding the scene formed using a first image-capture modality;
second image-capture data regarding the scene formed using a second image-capture modality that is different from the first image-capture modality;
a control circuit operably coupled to the data interface and configured to execute an iterative image reconstruction process by, at least in part:
establishing a first image-representation channel;
establishing a second image-representation channel;
establishing a fidelity error measure that, for any image containing first image-representation channel data and second image-representation channel data, measures inconsistency of the image as compared to the first image-capture data and the second image-capture data;
establishing a prior-penalty term that, for any image containing first image-representation channel data and second image-representation channel data, scores the image based on a priori likelihood or desirability using, at least in part, for each of a plurality of pixels for that image that contains the first image-representation channel data and the second image-representation channel data, a non-separable matrix-penalty of a Jacobian-matrix of the image at that pixel;
using an iterative process to generate a reconstructed image for which the reconstructed image at least approximately minimizes a combination of the fidelity error measure and the prior-penalty term.
2. The apparatus of claim 1 wherein the first image-capture modality comprises using a first level of x-ray energy and the second image-capture modality comprises using a second level of x-ray energy that is substantially different from the first level of x-ray energy.
3. The apparatus of claim 1 wherein the control circuit is configured to generate the reconstructed image by, at least in part, using the first image-representation channel and the second image-representation channel to evaluate individual pixels in the image-representation data with respect to statistical likelihood.
4. The apparatus of claim 1 wherein the control circuit is configured to evaluate the individual pixels by, at least in part, assessing a gradient for at least some of the individual pixels as compared to their corresponding neighboring pixels.
5. The apparatus of claim 4 wherein the gradient is an extended gradient.
6. The apparatus of claim 4 wherein the control circuit is configured to generate the reconstructed image by, at least in part, calculating a function of both the gradient as corresponds to the first image-representation channel for a given pixel and the gradient as corresponds to the second image-representation channel for the same given pixel.
7. The apparatus of claim 1 wherein the non-separable matrix-penalty of the Jacobian-matrix is a function of singular values of the Jacobian-matrix.
8. The apparatus of claim 7 wherein the function of the singular values of the Jacobian-matrix is a nuclear norm.
9. The apparatus of claim 1 further comprising:
third image-capture data regarding the scene formed using a third image-capture modality that is different from the first image-capture modality and the second image-capture modality;
wherein the fidelity error measure further measures the inconsistency of the image as compared to the third image-capture data.
10. The apparatus of claim 1 further comprising:
establishing a third image-representation channel;
wherein measuring the inconsistency of the image further comprises measuring the inconsistency of the image as compares to third image-representation channel data; and wherein scoring the image based on a priori likelihood or desirability further comprises scoring an image that further comprises third image-representation channel data.
11. The apparatus of claim 1 wherein the first image-representation channel comprises a first material basis and the second image-representation channel comprises a second material basis.
12. The apparatus of claim 1 wherein the control circuit is further configured to:
establish a first material basis;
establish a second material basis;
wherein the first image-representation channel describes image data as relates to the first material basis and the second image-representation channel describes image data as relates to the second material basis.
13. The apparatus of claim 1 wherein the iterative process comprises an iterative statistical reconstruction process.
14. The apparatus of claim 13 wherein the iterative statistical image reconstruction process comprises a computed tomography reconstruction process.
15. The apparatus of claim 13 wherein the iterative statistical image reconstruction process comprises a radiography-image reconstruction process.
16. A method by which a control circuit reconstructs an image using high energy-based data regarding a scene, the method including an iterative image reconstruction process that comprises:
by the control circuit:
accessing first image-capture data regarding the scene formed using a first image-capture modality;
accessing second image-capture data regarding the scene formed using a second image-capture modality that is different from the first image-capture modality;
establishing a first image-representation channel;
establishing a second image-representation channel;
establishing a fidelity error measure that, for any image containing first image-representation channel data and second image-representation channel data, measures inconsistency of the image as compared to the first image-capture data and the second image-capture data;
establishing a prior-penalty term that, for any image containing first image-representation channel data and second image-representation channel data, scores the image based on a priori likelihood or desirability using, at least in part, for each of a plurality of pixels for that image that contains the first image-representation channel data and the second image-representation channel data, a non-separable matrix-penalty of a Jacobian-matrix of the image at that pixel;
using an iterative process to generate a reconstructed image for which the reconstructed image at least approximately minimizes a combination of the fidelity error measure and the prior-penalty term.
17. The method of claim 16 wherein the non-separable matrix-penalty of the Jacobian-matrix is a function of singular values of the Jacobian-matrix.
18. The method of claim 17 wherein the function of the singular values of the Jacobian-matrix is a nuclear norm.
19. The method of claim 16 wherein the first image-representation channel comprises a first material basis and the second image-representation channel comprises a second material basis.
20. The method of claim 16 further comprising assessing a gradient for at least some of the individual pixels as compared to their corresponding neighboring pixels.
21. The method of claim 20 further comprising calculating a function of both the gradient as corresponds to the first image-representation channel for a given pixel and the gradient as corresponds to the second image-representation channel for the same given pixel.