Image stitching and related method therefor
A method to stitch images includes the steps of: determining a template window in the first digital image and a target window in the second digital image and extracting selected features from the selected windows and a template in the template window; extracting selected features from within the template at a first resolution; matching the selected features in the target window; extracting selected features from within the template and target window at a second resolution higher than the first resolution; matching the selected features in the target window; performing a first evaluation of the second estimate of the stitching location; blending the second estimate of the stitching location; performing an evaluation of the stitching location; stitching the first digital image and a second digital image using the stitching location to create a stitched image; and saving the stitched image to memory. A system to perform the method is also described.
1. A method to stitch images, comprising: determining a first Region of Interest (ROI) in a first digital image and a second ROI in a second digital image; determining a template window in said first digital image and a target window in said second digital image, wherein said template window comprises a template; extracting selected features from within said template at a first resolution; matching said selected features of said template at the first resolution to features in said target window to make a first estimate of a stitching location along a stitching edge; based on the first estimate of the stitching location, extracting selected features from within said template at a second resolution higher than said first resolution; matching said selected features of said template in said first digital image at the second resolution to features in said target window in said second digital image to make a second estimate of said stitching location along the stitching edge; stitching said first digital image and said second digital image using said stitching location to create a stitched image; and saving said stitched image to memory.
2. The method of claim 1 , further comprising performing an evaluation of said second estimate of said stitching location.
3. The method of claim 1 , wherein said template and said template window have substantially a same size.
4. The method of claim 1 , wherein said template is equal to or smaller than said template window in size.
5. The method of claim 1 , further comprising smoothing an image within said template and an image within said target window by use of a Gaussian kernel.
6. The method of claim 1 , wherein said acts of extracting selected features comprise extracting selected one dimensional (1D) features, or extracting selected two dimensional (2D) features.
7. The method of claim 6 , wherein at least one of said acts of extracting selected features is performed based on at least one Radon transform.
8. The method of claim 6 , wherein at least one of said acts of extracting selected features is performed based on one or more of a gray-scale feature, an edge feature, and a range feature.
9. The method of claim 6 , wherein at least one of said acts of extracting selected features is performed based on one or more of a selected gray mean feature, a selected edge mean feature, and a selected range mean feature.
10. The method of claim 6 , wherein at least one of said acts of extracting selected features is performed based on one or more of a selected gray sum feature, a selected edge sum feature, and a selected range sum feature.
11. The method of claim 6 , wherein at least one of said acts of extracting selected features is performed based on one or more of a thresholded gray feature, a thresholded edge feature, and a thresholded range feature.
12. The method of claim 6 , wherein at least one of said acts of extracting selected features is performed based on one or more of a gray-scale feature, a smoothed gray-scale feature, a selected gray mean feature, a selected gray sum feature, and a thresholded gray feature.
13. The method of claim 6 , wherein at least one of said acts of extracting selected features is performed based on one or more of an edge feature, a selected edge mean feature, a selected edge sum feature, and a thresholded edge feature.
14. The method of claim 6 , wherein at least one of said acts of extracting selected features is performed based on one or more of a range feature, a selected range feature, a selected range mean feature, a selected range sum feature, and a thresholded range feature.
15. The method of claim 1 , wherein at least one of said acts of matching is performed using a template matching algorithm.
16. The method of claim 15 , wherein said template matching algorithm comprises at least one mean square distance computation.
17. The method of claim 15 , wherein said template matching algorithm comprises at least one least squares computation.
18. The method of claim 15 , wherein said template matching algorithm comprises at least one Euclidian distance computation.
19. The method of claim 1 , wherein said stitching location falls substantially along a row or column.
20. The method of claim 2 , wherein said act of performing an evaluation comprises determining at least one 2D city block distance.
21. The method of claim 2 , wherein said act of performing an evaluation comprises determining at least one 1D city block distance.
22. The method of claim 2 , wherein said act of performing an evaluation comprises determining at least one 2D normalized city block distance.
23. The method of claim 2 , wherein said act of performing an evaluation comprises determining at least one 2D normalized cross correlation.
24. The method of claim 2 , wherein said act of performing an evaluation comprises determining at least one 1D normalized city block distance.
25. The method of claim 2 , wherein said act of performing an evaluation comprises determining at least one 1D normalized cross correlation.
26. The method of claim 2 , wherein said act of performing an evaluation is performed based on a predetermined similarity distance obtained from one or more stitched images deemed by visual inspection to have an acceptable stitched quality.
27. The method of claim 1 , further comprising performing a blending procedure using said second estimate of said stitching location using a blending algorithm:
alfa=(1* i )/(Numbe_Of_Rows_Overlapped_Region)
OVERLAP( i,j )=(1−alfa)*OVERLAP1( i,j )+alfa*OVERLAP2( i,j )
For_all_( i,j ).
28. A method to stitch images, comprising: determining a first Region of Interest (ROI) in a first digital image and a second ROI in a second digital image; determining a template window in said first digital image and a target window in said second digital image, wherein said template window comprises a template; step for extracting selected features from within said template at a first resolution; matching said selected features of said template at said first resolution to features in said target window to make a first estimate of a stitching location along a stitching edge; step for extracting selected features from within said template at a second resolution higher than said first resolution; matching said selected features of said template at said second resolution to features in said target window at a second resolution to make a second estimate of said stitching location along the stitching edge; stitching said first digital image and a second digital image using said stitching location to create a stitched image; and saving said stitched image to memory.
29. An imaging system comprising: a source for providing imaging radiation, the source having a shutter; a detector for receiving the imaging radiation; and an imaging computer component communicatively coupled to said detector and configured to run an autostitching algorithm to stitch images together, wherein said autostitching algorithm is configured to read a first digital image and a second digital image, determine a first Region of Interest (ROI) in said first digital image and a second ROI in said second digital image, determine a template window in said first digital image and a target window in said second digital image, wherein said template window comprises a template, extract selected features from within said template at a first resolution, match said selected features of said template at said first resolution to features in said target window to make a first estimate of a stitching location along a stitching edge, based on the first estimate of the stitching location, extract selected features from within said template at a second resolution higher than said first resolution, match said selected features of said template at said second resolution to features in said target window to make a second estimate of said stitching location along the stitching edge, stitch said first digital image and a second digital image using said stitching location to create a stitched image, and save said stitched image to memory.
30. The medical imaging system of claim 29 , wherein said source comprises an X-ray source.
31. The method of claim 2 , further comprising iteratively performing: said act of extracting selected features to said act of performing said evaluation of said second estimate, until said evaluation of said second estimate yields a stitching performance at or better than a pre-determined stitching performance level.
32. The method of claim 28 , further comprising performing an evaluation of said second estimate of said stitching location.
33. The medical imaging system of claim 29 , further comprising perform an evaluation of said second estimate of said stitching location.