Method and system for archival and retrieval of images based on the shape properties of identified segments

• US 6,801,661 B1
• Filed: 02/15/2001
• Issued: 10/05/2004
• Est. Priority Date: 02/15/2001
• Status: Active Grant

• Current Assignee: Monument Peak Ventures, LLC (Dominion Harbor Enterprises, LLC)
• Original Assignee: Eastman Kodak Company
• Inventors: Sotak, George E., Mehrotra, Rajiv
• Primary Examiner(s): COUSO, YON JUNG
• Application Number:
  US09/784,744
• Time in Patent Office: 1,328 Days
• Field of Search: 382/280, 382/224, 382/190, 382/195, 382/199, 382/261, 382/308, 382/276, 382/103, 382/203, 382/209, 382/141, 382/100, 382/305, 382/220, 348/86, 348/91
• US Class Current: 382/203
• CPC Class Codes: G06V 10/478 Contour-based spectral representations or scale-space representations, e.g. by Fourier analysis, wavelet analysis or curvature scale-space [CSS]
• CPC Class Codes: G06V 10/507 Summing image-intensity values; Histogram projection analysis

Abstract

A method for representing an image in terms of the shape properties of its identified segments of interest involves analyzing each segment of interest to automatically identify one or more of its perceptually significant components and then representing each perceptually significant component in terms of its shape properties, e.g., by means of Fourier descriptors. The image segment is then characterized as a composition of the shape properties of its perceptually significant components. By repeating the foregoing steps, the image is represented as a composition of its characterized segments. This technique can then be used to archive and retrieve images based on the shape properties of identified segments.

Claims

1. A method for representing an image in terms of the shape properties of its identified segments of interest, said method comprising the steps of:

(a) providing a digital image with identified segments of interest;

(b) analyzing a segment of interest to identify one or more of its perceptually significant components, wherein perceptually insignificant components are removed from the segment of interest;

(c) representing a perceptually significant component in terms of its shape properties, thereby providing a shape representation;

(d) characterizing an image segment as a composition of the shape representations of its perceptually significant components, thereby providing a characterized image segment; and

(e) repeating steps (b) to (d) to represent the image as a composition of its characterized segments;

wherein step (b) includes removing perceptually insignificant components from the boundary model of the identified segments with a morphological filter.

2. The method as claimed in claim 1 wherein step (b) comprises the steps of extracting a boundary model of the identified segments and decomposing the boundary model into perceptually significant components.

3. The method as claimed in claim 1 wherein the morphological filter is adaptive to the roughness of the shape boundary.

4. The method as claimed in claim 1 wherein step (c) comprises representing the perceptually significant component in terms of Fourier Descriptors.

5. The method as claimed in claim 4 wherein the Fourier Descriptors are normalized so as to be insensitive at least to rigid transformations.

6. The method as claimed in claim 1 as used for determining the similarity of two image segments, said method further comprising the steps of:

(a) providing first and second image segments;

(b) applying steps (b) to (e) of claim 1 to both the first and second segments to generate the respective representations; and

(c) computing the similarity between the first and second segments in terms of the similarity of their perceptually significant components.

7. The method as claimed in claim 6 wherein said computing the similarity between the first and second segments involves the comparison of a Euclidean distance metric for the respective representations.

8. A system representing an image in terms of the shape properties of its identified segments of interest, said system retrieving, from a database of image segments, images containing a segment similar to a query image segment, said system comprising:

means for analyzing each segment of interest to automatically identify one or more of its perceptually significant components, wherein perceptually insignificant components are removed from each segment of interest;

means for representing each perceptually significant component in terms of its shape properties, thereby providing a shape representation;

means for characterizing each image segment as a composition of the shape representations of its perceptually significant components, thereby providing a characterized image segment; and

means for representing the image as a composition of its characterized image segments;

means for providing an example digital image with the identified segments of interest and a database of archived images and their segment representations;

means for searching the database for candidate images that may contain segments with perceptually similar components to those of the example digital image;

means for determining overall similarity between example digital image and candidate images by computing the similarity between the segment representations of the example digital image and the segment representation of the candidate images in terms of the similarity of their perceptually significant components; and

means for retrieving and displaying the resulting candidate images.

9. A method for retrieving, from a database of image segments, images containing a segment similar to a query image segment, said method further comprising the steps of:

providing an example digital image with the identified segments of interest;

providing a database of archived images and their segment representations;

analyzing a segment of interest to identify one or more of its perceptually significant components, wherein perceptually insignificant components are removed from the segment of interest;

representing a perceptually significant component in terms of its shape properties, thereby providing a shape representation;

characterizing an image segment as a composition of the shape representations of its perceptually significant components, thereby providing a characterized image segment; and

repeating said analyzing, representing, and characterizing steps to represent the image as a composition of its characterized segments;

searching the database for candidate images that may contain segments with perceptually similar components to those of the example digital image;

determining overall similarity between example digital image and candidate images by computing the similarity between the segment representations of the example digital image and the segment representation of the candidate images in terms of the similarity of their perceptually significant components; and

retrieving and displaying the resulting candidate images.

10. The method as claimed in claim 9 further comprising the step of ranking the similarity and wherein said retrieving and displaying rank orders the resulting candidate images.

11. The method as claimed in claim 9 wherein said step of providing a database includes providing a multi-dimensional index to the archived images and wherein said characterized segments of said digital image are limited in dimensionality to correspond to the index.

12. A method for representing an image in terms of the shape properties of its identified segments of interest, said method comprising the steps of:

(a) extracting a boundary model of an identified segment of interest;

(b) decomposing the boundary model of the segment of interest into relevant components;

(c) morphologically filtering the relevant components to remove perceptually insignificant components of the segment of interest, thereby identifying one or more of its perceptually significant components;

(d) representing a perceptually significant component in terms of its shape properties, thereby providing a shape representation;

(e) characterizing an image segment as a composition of the shape representations of its perceptually significant components, thereby providing a characterized image segment; and

(f) repeating steps (b) to (e) to represent the image as a composition of its characterized segments.

13. The method as claimed in claim 12 wherein the filtering of step (c) is adaptive to the roughness of the shape boundary.

14. The method as claimed in claim 12 wherein step (d) comprises representing the perceptually significant component in terms of Fourier Descriptors.

15. The method as claimed in claim 14 wherein the Fourier Descriptors are normalized so as to be insensitive at least to rigid transformations.

16. A computer program product for representing an image in terms of the shape properties of its identified segments of interest comprising:

a computer readable storage medium having a computer program stored thereon for performing the steps of;

(a) providing a digital image with the identified segments of interest;

(b) analyzing a segment of interest to automatically identify one or more of its perceptually significant components, wherein perceptually insignificant components are removed from the segment of interest;

(c) representing a perceptually significant component in terms of its shape properties, thereby providing a shape representation;

(d) characterizing an image segment as a composition of the shape representations of its perceptually significant components, thereby providing a characterized image segment; and

(e) repeating steps (b) to (d) to represent the image as a composition of its characterized segments;

wherein step (b) includes removing perceptually insignificant components from the boundary model of the identified segments with a morphological filter.

17. The computer program product as claimed in claim 16 as used for determining the similarity of two image segments, said computer program product further comprising the steps of:

(a) providing first and second image segments;

(b) applying steps (b) to (e) of claim 16 to both the first and second segments to generate the respective representations; and

(c) computing the similarity between the first and second segments in terms of the similarity of their perceptually significant components.

18. A computer program product for retrieving, from a database of image segments, images containing a segment similar to a query image segment, comprising:

a computer readable storage medium having a computer program stored thereon for performing the steps of;

providing an example digital image with the identified segments of interest;

providing a database of archived images and their segment representations;

analyzing a segment of interest to identify one or more of its perceptually significant components, wherein perceptually insignificant components are removed from the segment of interest;

representing a perceptually significant component in terms of its shape properties, thereby providing a shape representation;

characterizing an image segment as a composition of the shape representations of its perceptually significant components, thereby providing a characterized image segment; and

repeating said analyzing, representing, and characterizing steps to represent the image as a composition of its characterized segments;

searching the database for candidate images that may contain segments with perceptually similar components to those of the example digital image;

determining overall similarity between example digital image and candidate images by computing the similarity between the segment representations of the example digital image and the segment representation of the candidate images in terms of the similarity of their perceptually significant components; and

retrieving and displaying the resulting candidate images.

19. A system representing an image in terms of the shape properties of its identified segments of interest, said system comprising:

means for analyzing each segment of interest to automatically identify one or more of its perceptually significant components, wherein perceptually insignificant components are removed from each segment of interest;

means for representing each perceptually significant component in terms of its shape properties, thereby providing a shape representation;

means for characterizing each image segment as a composition of the shape representations of its perceptually significant components, thereby providing a characterized image segment; and

means for representing the image as a composition of its characterized image segments;

wherein said means for analyzing includes means for removing perceptually insignificant components from the boundary model of the identified segments with a morphological filter.

20. A system for determining the similarity of two image segments that are represented according to claim 19 in terms of the shape representations of their identified segments of interest, said system comprising:

means for obtaining first and second image segments; and

means for computing the similarity between the first and second segments in terms of the similarity of their perceptually significant components.