Method for antiscatter stationary grid artifacts detection and attenuation in digital radiographic images

1. A method for detecting and attenuating grid artifacts in a digital radiographic image comprising:

providing an input digital radiographic image;

processing the digital radiographic image with a detection algorithm to determine whether the digital radiographic image has grid artifacts;

if it does, then detecting grid characteristics of the grid artifacts;

designing a notch filter as a function of the grid characteristics;

identifying edges and spikes in the digital radiographic image;

eliminating the identified edges and spikes from the digital radiographic image to generate a first processed digital radiographic image;

applying the notch filter to the first processed digital radiographic image to generate a second processed digital radiographic image; and

restoring the identified edges and spikes to the second processed digital radiographic image to generate an output image having attenuated grid artifacts.

2. The method of claim 1 , wherein the detection algorithm is based on a 2-D dynamic correlation in both spatial and frequency domains.

3. The method of claim 1 , wherein the processing includes consequent dynamic analyzing image profiles in two dimensions both in spatial and in frequency domains in a predefined square sub-region of the input digital radiographic image (window) in each of horizontal and vertical directions.

4. The method of claim 3 , wherein the consequent dynamic analyzing includes calculating and analyzing auto-correlation (ACF) and cross-correlation (CCF) functions for 2 image profiles with predefined incremental step in the predefined window.

5. The method of claim 4 , wherein the analyzing of ACF and CCF includes:

obtaining Fourier power spectra of ACF and CCF 1 st derivatives;

averaging the obtained Fourier power spectra;

searching for the frequencies of the maxima in the right part of averaged Fourier spectra; and

checking that these frequencies of the maxima are the same values, within predefined accuracy.

6. The method of claim 5 , wherein the dynamic analyzing includes tracking successful matches of found frequency values treated as detected grid artifacts.

7. The method of claim 5 , wherein the detecting of grid frequency includes detecting a grid frequency and maximum magnitude in a current image profile Fourier spectrum within the bandwidth equal to the pre-defined accuracy at the found frequency for CCF.

8. The method of claim 5 , wherein the detecting of grid signal-to-noise (SNR) ratio includes:

calculating a standard deviation of signal frequencies within the bandwidth equal to the predefined accuracy at the frequency of 0.25 of Nyquist;

calculating and accumulating current image profile SNR as a ratio of the standard deviation and the grid maximum magnitude; and

calculating SNR averaged by total number of profiles with detected grid artifacts.

9. The method of claim 5 , wherein the detecting of grid frequency includes calculating the grid frequency from the frequencies of the maxima found in the right part of the CCF 1 st derivative power spectrum, averaged over all image profiles that contain detected grid artifacts.

10. The method of claim 1 , wherein the grid characteristics include at least one of the following: grid orientation, frequency, and signal-to-noise ratio.

11. A method for detecting and attenuating grid artifacts in a digital radiographic image comprising:

providing an input digital radiographic image comprised of rows of pixels;

processing the digital radiographic image with a detection algorithm to determine whether the digital radiographic image has grid artifacts;

if it does, then detecting grid characteristics of the grid artifacts;

designing a notch filter as a function of the grid characteristics;

identifying edges and spikes in one of the rows of pixels of the digital radiographic image;

eliminating the identified edges and spikes from the one of the rows of pixels to generate a first processed row of pixels;

applying the notch filter to the first processed row to generate a second processed row of pixels;

restoring the identified edges and spikes to the second processed row of pixels to generate an output pixel row;

repeating the steps of identifying, eliminating, applying, and restoring for another row of pixels of the digital radiographic image; and

generating an output image having attenuated grid artifacts from the output pixel rows.

12. The method of claim 11 , wherein the detection algorithm is based on a 2-D dynamic correlation in both spatial and frequency domains.

13. The method of claim 11 , wherein the processing includes consequent dynamic analyzing image profiles in two dimensions both in spatial and in frequency domains in a predefined square sub-region of the input digital radiographic image (window) in each of horizontal and vertical directions.

14. The method of claim 13 , wherein the consequent dynamic analyzing includes calculating and analyzing auto-correlation (ACF) and cross-correlation (CCF) functions for 2 image profiles with predefined incremental step in the predefined window.

15. The method of claim 14 , wherein the analyzing of ACF and CCF includes:

obtaining Fourier power spectra of ACF and CCF 1 st derivatives;

averaging the obtained Fourier power spectra;

searching for the frequencies of the maxima in the right part of averaged Fourier spectra; and

checking that these frequencies of the maxima are the same values, within predefined accuracy.

16. The method of claim 15 , wherein the dynamic analyzing includes tracking successful matches of found frequency values treated as detected grid artifacts.

17. The method of claim 15 , wherein the detecting of grid frequency includes detecting a grid frequency and maximum magnitude in a current image profile Fourier spectrum within the bandwidth equal to the pre-defined accuracy at the found frequency for CCF.

18. The method of claim 15 , wherein the detecting of grid signal-to-noise (SNR) ratio includes:

calculating a standard deviation of signal frequencies within the bandwidth equal to the predefined accuracy at the frequency of 0.25 of Nyquist;

calculating and accumulating current image profile SNR as a ratio of the standard deviation and the grid maximum magnitude; and

calculating SNR averaged by total number of profiles with detected grid artifacts.

19. The method of claim 15 , wherein the detecting of grid frequency includes calculating the grid frequency from the frequencies of the maxima found in the right part of the CCF 1 st derivative power spectrum, averaged over all image profiles that contain detected grid artifacts.

20. The method of claim 11 , wherein the grid characteristics include at least one of the following: grid orientation, frequency, and signal-to-noise ratio.