Image comparison by asymmetric dynamic warping

• US 8,405,681 B2
• Filed: 06/24/2009
• Issued: 03/26/2013
• Est. Priority Date: 06/24/2009
• Status: Expired due to Fees

• Current Assignee: Ariel Shamir, Michael Rubinstein, The Interdisciplinary Center Herzliya
• Original Assignee: Ariel Shamir, Michael Rubinstein, The Interdisciplinary Center Herzliya
• Inventors: Shamir, Ariel, Rubinstein, Michael
• Primary Examiner(s): Sams, Michelle L
• Application Number:
  US12/490,344
• Publication Number:
  US 20100328345A1
• Time in Patent Office: 1,371 Days
• Field of Search: 345/698, 345/699, 345/660, 345/672, 345/649, 345/619, 345/620
• US Class Current: 345/647
• CPC Class Codes: G06T 2207/10016 Video; Image sequence
• CPC Class Codes: G06T 2207/20021 Dividing image into blocks, subimages or windows
• CPC Class Codes: G06T 3/04 Context-preserving transformations, e.g. by using an importance map panospheric to cylindrical image transformations G06T3/12
• CPC Class Codes: G06T 7/30 Determination of transform parameters for the alignment of images, i.e. image registration

Abstract

A method for image comparison includes reading from a memory in a computerized image processor a source image and a target image. The source and target images are segmented, using the image processor, into respective pluralities of source and target arrays, respectively including source and target array elements, each array element including a matrix of one or more pixels from a respective image, each of the source arrays having a respective position in the source image, and each of the target arrays corresponding respectively to one of the source arrays based on the respective position. An asymmetric warping process is applied between each of the source arrays and each of the respectively corresponding target arrays so as to compute respective array similarity scores. The respective array similarity scores are combined so as to compute an image similarity score of the target image with respect to the source image.

Claims

1. A method for image comparison, comprising:

reading from a memory in a computerized image processor a source image and a target image;

segmenting the source and target images, using the image processor, into respective pluralities of source and target arrays, respectively comprising source and target array elements, each array element comprising a matrix of one or more pixels from a respective image, each of the source arrays having a respective position in the source image, and each of the target arrays corresponding respectively to one of the source arrays based on the respective position;

applying an asymmetric warping process between each of the source arrays and each of the respectively corresponding target arrays so as to compute respective array similarity scores; and

combining the respective array similarity scores so as to compute an image similarity score of the target image with respect to the source image.

2. The method according to claim 1, wherein the source image has a source size, and wherein the target image has a target size that is different from the source size.

3. The method according to claim 1, wherein the method is applied to compute respective image similarity scores of a plurality of target images with respect to the source image, and wherein the method comprises selecting and displaying, using the computerized image processor, at least one of the target images responsively to the image similarity scores.

4. The method according to claim 3, wherein the method comprises generating the plurality of target images by a process of image retargeting using different retargeting operations, and wherein selecting the at least one of the target images comprises selecting one of the retargeting operations in order to generate a target image that is most similar to the source image.

5. The method according to claim 4, wherein the steps of generating the plurality of the target images and selecting one of the retargeting operations are repeated one or more times, thereby defining a sequence of the retargeting operations.

6. The method according to claim 1, wherein applying the asymmetric warping process comprises computing a bi-directional warping image similarity measure.

7. The method according to claim 6, wherein computing the bi-directional warping image similarity measure comprises:

calculating a first score by combining the respective array similarity scores computed in comparing the source image to the target image;

calculating a second score by combining the respective array similarity scores computed in comparing the target image to the source image; and

calculating the bi-directional warping image similarity measure by combining the first score and the second score.

8. A computer program product for image processing, the computer program product comprising:

a non-transitory computer usable medium having computer usable code embodied therewith, the computer usable program code comprising;

computer usable code configured for reading from a memory a source image and a target image;

computer usable code configured for segmenting the source and target images into respective pluralities of source and target arrays;

respectively comprising source and target array elements, each array element comprising a matrix of one or more pixels from a respective image, each of the source arrays having a respective position in the source image, and each of the target arrays corresponding respectively to one of the source arrays based on the respective position;

computer usable code configured for applying an asymmetric warping process between each of the source arrays and each of the corresponding target arrays so as to compute respective array similarity scores; and

computer usable code configured for combining the respective array similarity scores so as to compute a respective image similarity scores of the target images with respect to the source image.

10. The product according to claim 8, wherein the source image has a source size, and wherein the target image has a target size that is different from the source size.

11. The product according to claim 8, wherein the product is applied to compute respective image similarity scores of a plurality of target images with respect to the source image, and wherein the product comprises computer usable code for selecting and displaying at least one of the target images responsively to the image similarity scores.

12. The product according to claim 11, wherein the product comprises computer usable code for generating the plurality of target images by a process of image retargeting using different retargeting operations, and wherein selecting the at least one of the target images comprises selecting one of the retargeting operations in order to generate a target image that is most similar to the source image.

13. The product according to claim 12, wherein the steps of generating the plurality of the target images and selecting one of the retargeting operations are repeated one or more times, thereby defining a sequence of the retargeting operations.

14. The product according to claim 8, wherein applying the asymmetric warping process comprises computing a bi-directional warping image similarity measure.

15. The product according to claim 14, wherein computing the bi-directional warping image similarity measure comprises:

calculating a first score by combining the respective array similarity scores computed in comparing the source image to the target image;

calculating a second score by combining the respective array similarity scores computed in comparing the target image to the source image; and

calculating the bi-directional warping image similarity measure by combining the first score and the second score.

9. An image comparison apparatus, comprising:

a memory; and

an image processor, which is configured to

read from the memory a source image and a target image;

segment the source and target images into respective pluralities of source and target arrays, respectively comprising source and target array elements, each array element comprising a matrix of one or more pixels from a respective image, each of the source arrays having a respective position in the source image, and each of the target arrays corresponding respectively to one of the source arrays based on the respective position;

apply an asymmetric warping process between each of the source arrays and each of the respectively corresponding target arrays so as to compute respective array similarity scores; and

combine the respective array similarity scores so as to compute an image similarity score of the target image with respect to the source image.

16. The apparatus according to claim 9, wherein the source image has a source size, and wherein the target image has a target size that is different from the source size.

17. The apparatus according to claim 9, wherein the image processor is configured to compute respective image similarity scores of a plurality of target images with respect to the source image, and to select and display at least one of the target images responsively to the image similarity scores.

18. The apparatus according to claim 17, wherein the image processor is configured to generate the plurality of target images by a process of image retargeting using different retargeting operations, and to select one of the retargeting operations in order to generate a target image that is most similar to the source image.

19. The apparatus according to claim 18, wherein the image processor is configured to repeat the steps of generating the plurality of the target images and selecting one of the retargeting operations one or more times, thereby defining a sequence of the retargeting operations.

20. The apparatus according to claim 9, wherein applying the asymmetric warping process comprises computing a bi-directional warping image similarity measure.