Method for automatically classifying images into events

• US 20040208365A1
• Filed: 10/29/2003
• Published: 10/21/2004
• Est. Priority Date: 04/15/2003
• Status: Active Grant

• Current Assignee: Monument Peak Ventures, LLC (Dominion Harbor Enterprises, LLC)
• Original Assignee: Eastman Kodak Company
• Inventors: Loui, Alexander C., Pavie, Eric
• Granted Patent: US 7,298,895 B2
• Time in Patent Office: Days
• US Class Current: 382/171
• CPC Class Codes: G06F 16/58 Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
• CPC Class Codes: G06F 16/5838 using colour
• CPC Class Codes: G06F 16/5854 using shape and object relationship
• CPC Class Codes: G06F 18/23 Clustering techniques
• CPC Class Codes: G06V 10/758 Involving statistics of pixels or of feature values, e.g. histogram matching
• CPC Class Codes: G06V 20/00 Scenes; Scene-specific elements control of digital cameras H04N23/60
• CPC Class Codes: G06V 20/13 Satellite images
• CPC Class Codes: G06V 2201/10 Recognition assisted with metadata

Abstract

A method for obtaining and automatically classifying images into events comprising the steps of: (a) obtaining a group of images from a digital source, wherein the images are in chronological order; (b) transferring the group of images to a computer system; said computer system (c) clustering the images into smaller groups based on chronological image similarity of nearby images by computing histograms of the images and comparing histogram intersection values obtained therefrom with one or more thresholds, whereby the clustering based on chronological image similarity is done in at least one stage by comparing each image with its direct neighboring images; and (d) evaluating the clustered images against a final condition related to at least one of a predetermined group maximum for the number of smaller groups and a predetermined maximum number of isolated pictures, whereby the smaller groups are classified as events if the final condition is met.

Claims

1. A method for obtaining and automatically classifying images into events, the method comprising the steps of:

(a) obtaining a group of images from a digital source, wherein images are in chronological order;

(b) transferring the group of images to a computer system;

said computer system

(c) clustering the images into smaller groups based on chronological image similarity of nearby images by computing histograms of the images and comparing histogram intersection values obtained therefrom with one or more thresholds, whereby the clustering based on chronological image similarity is done in at least one stage by comparing each image with its direct neighboring images; and

(d) evaluating the clustered images against a final condition related to at least one of a predetermined group maximum for the number of smaller groups and a predetermined maximum number of isolated pictures, whereby the smaller groups are classified as events if the final condition is met.

2. The method as in claim 1, wherein the clustering based on chronological image similarity in step (c) is done in at least two stages by comparing each image with its direct neighboring images and secondly with its second neighboring images.

3. The method as in claim 1, wherein step (c) includes clustering the images into smaller groups based on chronological image similarity of nearby images by using a block-based histogram technique.

4. The method as in claim 3, wherein the block-based histogram technique comprises analyzing the events for content by dividing the images into a plurality of blocks and grouping the images into subject grouping based on block-based histogram correlation which includes computing a color histogram of each block and computing a histogram intersection value which determines the similarity between blocks, thereby refining and improving the overall classification and subject grouping of the events.

5. The method as in claim 1, whereby if the final conditions are not initially met, further comprising the step of decrementing one or more of the thresholds and iterating through steps (c) and (d) until the final conditions are met.

6. The method as in claim 5, wherein the number of iterations is limited.

7. The method as in claim 1, wherein the digital source is a digital camera.

8. The method as in claim 1, wherein the digital source is a film scanner for scanning a photographic film.

9. The method as in claim 1, wherein the digital source is a CD ROM.

10. The method as in claim 1, wherein the digital source is one or more image files from a user hard disk.