Method for the reduction of image content redundancy in large image libraries

• US 20070260639A1
• Filed: 05/03/2006
• Published: 11/08/2007
• Est. Priority Date: 05/03/2006
• Status: Active Grant

• Current Assignee: UT-Battelle LLC
• Original Assignee: UT-Battelle LLC
• Inventors: Tobin, Kenneth, Karnowski, Thomas
• Granted Patent: US 7,672,976 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 16/5838 using colour
• CPC Class Codes: Y10S 707/99948 Application of database or data structure, e.g. distributed, multimedia, or image

Abstract

A method of increasing information content for content-based image retrieval (CBIR) systems includes the steps of providing a CBIR database, the database having an index for a plurality of stored digital images using a plurality of feature vectors, the feature vectors corresponding to distinct descriptive characteristics of the images. A visual similarity parameter value is calculated based on a degree of visual similarity between features vectors of an incoming image being considered for entry into the database and feature vectors associated with a most similar of the stored images. Based on said visual similarity parameter value it is determined whether to store or how long to store the feature vectors associated with the incoming image in the database.

Claims

1. A method of increasing information content for content-based image retrieval (CBIR) systems, comprising the steps of:

providing a CBIR database, said database comprising an index for a plurality of stored digital images using a plurality of feature vectors, said feature vectors corresponding to distinct descriptive characteristics of said images;

calculating a visual similarity parameter value based on a degree of visual similarity between features vectors of an incoming image being considered for entry into said database and feature vectors associated with a most similar of said stored images, and

determining whether to store or how long to store said feature vectors associated with said incoming image in said database based on said visual similarity parameter value.

2. The method of claim 1, wherein said visual similarity parameter is based on a distance or a divergence.

3. The method of claim 1, wherein said similarity parameter comprises a Euclidean, L-norm, Mahalanobis or quadradic form distance. Minkowski-form distance.

4. The method of claim 1, wherein said similarity parameter comprises a Kullback-Leiber or Jeffrey divergence.

5. The method of claim 1, further comprising the step of defining a threshold value, wherein:

if said visual similarity parameter value is above said threshold value said feature vectors associated with said incoming image is denied entry into said database, and

if said similarity parameter value is less than said threshold said feature vectors associated with said incoming image is entered into said database.

6. The method of claim 1, further comprising the step of defining a plurality of threshold values, wherein said plurality of threshold values are used to define ranges of said similarity parameter values which are paired with durations for storage lifetimes in said database for said feature vectors associated with incoming image.

7. The method of claim 1, wherein said images are in-process manufacturing images or in-process images from an on-going preclinical or clinical study involving a plurality of patients in a biological or medical environment.

8. The method of claim 7, wherein said in-process manufacturing images are biological, medical or semiconductor images.

9. The method of claim 1, further comprising the step of indexing said plurality of feature vectors derived from said images using an image clustering method to produce a hierarchical search tree for indexing said plurality of feature vectors.

10. The method of claim 9, wherein said image clustering method comprises an unsupervised clustering method.

11. The method of claim 9, wherein said image clustering method maps said plurality of said feature vectors associated with said images into a set of groups based on similar image content, with the number of groups numbering less than the number of said images.

12. The method of claim 1, wherein said database includes stored information describing specific processes or process steps associated with each of said images.

13. A content-based image retrieval (CBIR) system, comprising:

computer apparatus programmed with a routine set of instructions stored in a fixed medium, said computer apparatus comprising;

structure for extracting a plurality of feature vectors from a digital image, said feature vectors corresponding to particular descriptive characteristics of said image;

structure for implementing a clustering method to index said feature vectors into a hierarchical search tree;

a database storing said feature vectors associated with a plurality of said digital images communicably coupled to an image storage manager module, said image storage module identifying a visual similarity parameter value based on a degree of visual similarity between features vectors of an incoming digital image being considered for entry into said database and feature vectors of a most similar of said plurality of said digital images, and determining whether to store or how long to store said incoming said feature vectors associated with said incoming image in said database based on said visual similarity parameter value, and

structure for retrieving at least one of said stored images, said structure for retrieving extracting at least one image having feature vectors comparably related to said feature vectors associated with said query image.

14. The system of claim 13, wherein said image storage manager module includes at least one stored threshold value, wherein:

if said visual similarity parameter value is above said threshold value said feature vectors associated with said incoming image is denied entry into said database, and

if said similarity parameter less than said threshold value said feature vectors associated with said incoming image is entered into said database.

15. The system of claim 13, wherein said image clustering comprises unsupervised clustering.