Registration of differently scaled images
View Patent ↗An image registration method involves computing horizontal and vertical integral projection vectors for first and second distorted or partially distorted images. The images are registered by applying a scale factor estimation between the first and second images on the horizontal and vertical integral projection vectors.
1. An image registration method, comprising:
acquiring first and second distorted or partially distorted images of scenes that each include a first object;
computing horizontal and vertical integral projection vectors for the first and second distorted or partially distorted images or distortion-corrected images, or both;
registering the first and second distorted or partially distorted images, or the distortion-corrected images, or both, including applying a scale factor estimation between the first and second images on the horizontal and vertical integral projection vectors, including:
resampling at least a portion of the first distorted or partially distorted image with multiple different resampling factors over a range of resampling factors to generate multiple resampled integral projection vectors;
computing for each of the different resampling factors an absolute sum of differences, or sum of squares, between the resampled integral projection vectors of the first image and corresponding integral projection vectors of the second image; and
determining the scaling factor between the first and second distorted or partially distorted images, including finding the resampling factor that corresponds to the minimum value of the absolute sums over the range of resampling factors; and
performing a further action using the registered first and second distorted or partially distorted images or the corrected images, or both.
2. The method of claim 1 , further comprising correcting distortion in the first and second distorted or partially distorted images prior to computing the horizontal or vertical integral projection vectors.
3. The method of claim 1 , wherein the computing of the horizontal and vertical integral projection vectors comprises summing columns and rows of image data, respectively.
4. The method of claim 1 , further comprising determining a distortion function for at least a portion of the first or second distorted or partially distorted image, or both, to compute said distortion-corrected images or to compute distorted integral projection vectors, or both.
5. The method of claim 4 , wherein the determining a distortion function comprises finding a model function that approximately fits the distortion of at least said portion of the first or second distorted or partially distorted image, or both.
6. The method of claim 5 , wherein the distortion function determined for at least said portion of the first or second distorted or partially distorted image, or both, comprises:
1 +k 1 r ; or
[1 +k 1 r] −1 ; or
1 +k 1 r 2 ; or
[1 +k 1 r 2 ] −1 ; or
1 +k 1 r+k 2 r 2 ; or
(1 +k 1 r )/(1 +k 2 r 2 ); or
combinations thereof, wherein r comprises a radial distortion function parameter and k 1 and k 2 comprise distortion function coefficients for the first or second distorted or partially distorted image, respectively.
7. The method of claim 1 , wherein the first object comprises a face.
8. One or more non-transitory processor-readable media having code embedded therein for programming one or more processors to perform an image registration method, wherein the method comprises:
acquiring first and second distorted or partially distorted images of scenes that each include a first object;
computing horizontal and vertical integral projection vectors for the first and second distorted or partially distorted images or distortion-corrected images, or both;
registering the first and second distorted or partially distorted images, or the distortion-corrected images, or both, including applying a scale factor estimation between the first and second images on the horizontal and vertical integral projection vectors including:
resampling at least a portion of the first distorted or partially distorted image with multiple different resampling factors over a range of resampling factors to generate multiple resampled integral projection vectors;
computing for each of the different resampling factors an absolute sum of differences, or sum of squares, between the resampled integral projection vectors of the first image and corresponding integral projection vectors of the second image; and
determining the scaling factor between the first and second distorted or partially distorted images, including finding the resampling factor that corresponds to the minimum value of the absolute sums over the range of resampling factors; and
performing a further action using the registered first and second distorted or partially distorted images or the corrected images, or both.
9. The one or more non-transitory processor-readable media of claim 8 , wherein the method further comprises correcting distortion in the first and second distorted or partially distorted images prior to computing the horizontal or vertical integral projection vectors.
10. The one or more non-transitory processor-readable media of claim 8 , wherein the computing of the horizontal and vertical integral projection vectors comprises summing columns and rows of image data, respectively.
11. The one or more non-transitory processor-readable media of claim 8 , wherein the method further comprises determining a distortion function for at least a portion of the first or second distorted or partially distorted image, or both, to compute said distortion-corrected images or to compute distorted integral projection vectors, or both.
12. The one or more non-transitory processor-readable media of claim 11 , wherein the determining a distortion function comprises finding a model function that approximately fits the distortion of at least said portion of the first or second distorted or partially distorted image, or both.
13. The one or more non-transitory processor-readable media of claim 12 , wherein the distortion function determined for at least said portion of the first or second distorted or partially distorted image, or both, comprises:
1 +k 1 r ; or
[1 +k 1 r] −1 ; or
1 +k 1 r 2 ; or
[1 +k 1 r 2 ] −1 ; or
1 +k 1 r+k 2 r 2 ; or
(1 +k 1 r )/(1 +k 2 r 2 ); or
combinations thereof, wherein r comprises a radial distortion function parameter and k 1 and k 2 comprise distortion function coefficients for the first or second distorted or partially distorted image, respectively.
14. The one or more non-transitory processor-readable media of claim 8 , wherein the first object comprises a face.
15. A digital image acquisition device, comprising:
a lens,
an image sensor,
a processor,
a memory having code embedded therein for programming the processor to perform an image registration method, wherein the method comprises:
acquiring first and second distorted or partially distorted images of scenes that each include a first object;
computing horizontal and vertical integral projection vectors for the first and second distorted or partially distorted images or distortion-corrected images, or both;
registering the first and second distorted or partially distorted images, or the distortion-corrected images, or both, including applying a scale factor estimation between the first and second images on the horizontal and vertical integral projection vectors, including:
resampling at least a portion of the first distorted or partially distorted image with multiple different resampling factors over a range of resampling factors to generate multiple resampled integral projection vectors;
computing for each of the different resampling factors an absolute sum of differences, or sum of squares, between the resampled integral projection vectors of the first image and corresponding integral projection vectors of the second image; and
determining the scaling factor between the first and second distorted or partially distorted images, including finding the resampling factor that corresponds to the minimum value of the absolute sums over the range of resampling factors; and
performing a further action using the registered first and second distorted or partially distorted images or the corrected images, or both.
16. The device of claim 15 , wherein the method further comprises correcting distortion in the first and second distorted or partially distorted images prior to computing the horizontal or vertical integral projection vectors.
17. The device of claim 15 , wherein the computing of the horizontal and vertical integral projection vectors comprises summing columns and rows of image data, respectively.
18. The device of claim 15 , wherein the method further comprises determining a distortion function for at least a portion of the first or second distorted or partially distorted image, or both, to compute said distortion-corrected images or to compute distorted integral projection vectors, or both.
19. The device of claim 18 , wherein the determining a distortion function comprises finding a model function that approximately fits the distortion of at least said portion of the first or second distorted or partially distorted image, or both.
20. The device of claim 19 , wherein the distortion function determined for at least said portion of the first or second distorted or partially distorted image, or both, comprises:
1 +k 1 r ; or
[1 +k 1 r] −1 ; or
1 +k 1 r 2 ; or
[1 +k 1 r 2 ] −1 ; or
1 +k 1 r+k 2 r 2 ; or
(1 +k 1 r )/(1 +k 2 r 2 ); or
combinations thereof, wherein r comprises a radial distortion function parameter and k 1 and k 2 comprise distortion function coefficients for the first or second distorted or partially distorted image, respectively.
21. The device of claim 15 , wherein the first object comprises a face.