# US 8,280,164 B2 · Producing object cutouts in topically related images

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 8,280,164 B2
- Document ID: US8280164B2
- Status: Granted
- Kind: B2
- Application number: 12/397,547
- Filed: 2009-03-04
- Published: 2010-09-09
- Granted: 2012-10-02
- Earliest claimed: 2009-03-04
- Assignees: Eastman Kodak Company
- Inventors: Jiebo Luo; Dhruv Batra; Andrew C. Gallagher
- Web page: https://patentgrep.com/patent/US8280164B2

## CPC

- G06T 11/60: Creating or editing images; Combining images with text
- G06T 7/11: Region-based segmentation
- G06T 7/143: involving probabilistic approaches, e.g. Markov random field [MRF] modelling
- G06V 10/26: Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06T 2207/20101: Interactive definition of point of interest, landmark or seed
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING

## Abstract

A method for extracting an object out of each image in a group of digital images that contain the object, includes providing a group of digital images, each containing the object with a background; selecting a seed image from the group of digital images, and displaying the seed image to a user; the user providing at least one marking for the seed image, which corresponds to a subset of pixels in the seed image that indicates whether the set of pixels belongs to a part of object of interest or a part of the background; producing from the seed image and the marking a statistical model that can be used for separating the object of interest from the background in the group of images; and applying the statistical model to each image in the group of digital images to produce a cutout of the object of interest from each digital image.

## Claims

### Claim 1

A method for extracting an object of interest out of each image in a group of digital images that contain the object of interest, comprising:
(a) providing a group of digital images, each containing the object of interest with a background;
(b) selecting a seed image from the group of digital images, and displaying the seed image to a user;
(c) the user providing at least one marking for the seed image, which corresponds to a subset of pixels in the seed image that indicates whether the set of pixels belongs to a part of object of interest or a part of the background;
(d) producing from the seed image and the marking a statistical model that can be used for separating the object of interest from the background in the group of images, wherein the statistical model includes:
(i) learned likelihood for a feature vector belonging to foreground and background; and
(ii) learned distances or affinities between features such that these distances reflect the input received via the user marking; and
(e) applying the statistical model to each image in the group of digital images to produce a cutout of the object of interest from each digital image.

### Claim 2 (depends on 1)

The method of claim 1, wherein step (a) further comprises:
(i) providing a collection of digital images with associated metadata;
(ii) analyzing either the metadata or image content or both to identify subsets of topically related digital images; and
(iii) producing the group of digital images from the subset of similar images.

### Claim 3 (depends on 2)

The method of claim 2, wherein:
(i) metadata includes information having date stamp, geotags, camera parameters (focal length), user notes or labels; and
(ii) image content is based on visual features including color features or texture features, or a combination of both.

### Claim 4 (depends on 1)

The method of claim 1, wherein the seed image is selected from the group of digital images by:
(i) randomly choosing an image out of the group; or
(ii) a heuristic that chooses the image with the most non-informative distribution of visual features among all images in the group.

### Claim 5 (depends on 1)

The method of claim 1, wherein:
(i) a marking includes a scribble, stroke, rectangle, or other designation of a subset of image pixels, some of which are marked as belonging to foreground, while others belonging to background; and
(ii) this marking is received via a user-interface on a display screen and via a user input devices.

### Claim 6 (depends on 1)

The method of claim 1 further includes computing a plurality of cues, individually or in combination, including uncertainty-based, scribble-based, or global image cues, to indicate areas in the seed image where the user marking is helpful.

### Claim 7 (depends on 1)

The method of claim 1 further includes using the object cutouts to produce an image product, including:
(i) producing a collage from the object cutouts in a group;
(ii) removing unwanted object and filling in the resulting blank space using texture synthesis or in-painting;
(iii) compositing one or more of the object cutouts on a new image background; or
(iv) producing samples of the object for training a detector for the object.
