# US 2013/0156305 A1 · System and method for the detection of anomalies in an image

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

- Number: US 2013/0156305 A1
- Document ID: US20130156305A1
- Status: Published application
- Kind: A1
- Application number: 13/660,672
- Filed: 2012-10-25
- Published: 2013-06-20
- Earliest claimed: 2009-06-23
- Assignees: Los Alamos National Security, LLC
- Inventors: LAKSHMAN PRASAD; SRIRAM SWAMINARAYAN
- Web page: https://patentgrep.com/patent/US20130156305A1

## CPC

- G06T 7/11: Region-based segmentation
- G06T 7/12: Edge-based segmentation
- G06T 7/162: involving graph-based methods
- G06T 7/181: involving edge growing; involving edge linking
- G06T 7/187: involving region growing; involving region merging; involving connected component labelling
- 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 2207/10024: Color image
- G06T 2207/10032: Satellite or aerial image; Remote sensing
- G06T 2207/20016: Hierarchical, coarse-to-fine, multiscale or multiresolution image processing; Pyramid transform
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING

## Abstract

Preferred aspects of the present invention can include receiving a digital image at a processor; segmenting the digital image into a hierarchy of feature layers comprising one or more fine-scale features defining a foreground object embedded in one or more coarser-scale features defining a background to the one or more fine-scale features in the segmentation hierarchy; detecting a first fine-scale foreground feature as an anomaly with respect to a first background feature within which it is embedded; and constructing an anomalous feature layer by synthesizing spatially contiguous anomalous fine-scale features. Additional preferred aspects of the present invention can include detecting non-pervasive changes between sets of images in response at least in part to one or more difference images between the sets of images.

## Claims

### Claim 1

A method comprising:
receiving a digital image at a processor;
segmenting the digital image into a hierarchy of feature layers comprising one or more fine-scale features defining a foreground object embedded in one or more coarser-scale features defining a background to the one or more fine-scale features in the segmentation hierarchy;
detecting a first fine-scale foreground feature as an anomaly with respect to a first background feature within which it is embedded; and
constructing an anomalous feature layer by synthesizing spatially contiguous anomalous fine-scale features.

### Claim 2 (depends on 1)

The method of claim 1, wherein segmenting the image comprises hierarchically segmenting the image into multiple feature layers, each comprising two or more polygons.

### Claim 3 (depends on 1)

The method of claim 1, further comprising: identifying for each fine-scale feature one or more background features as ancestors in the segmentation hierarchy, and identifying for each coarser-scale feature one or more foreground features as descendants in the segmentation hierarchy.

### Claim 4 (depends on 3)

The method of claim 3, further comprising constructing a statistical distribution of finer-scale descendant polygons for each coarser-scale background polygon in response to the spectral attributes of the descendant polygons.

### Claim 5 (depends on 1)

The method of claim 1, wherein detecting an anomaly in the background feature comprises detecting a statistically deviant spectral value in the distribution of spectral values of the descendant polygons of the background feature.

### Claim 6 (depends on 5)

The method of claim 5, further comprising characterizing all descendant polygons, corresponding to the statistically deviant spectral value identified in the distribution of the spectral values of descendant polygons of the background feature, as anomalous with respect to the background feature.

### Claim 7 (depends on 6)

The method of claim 6, wherein detecting an anomalous descendant feature in a background feature comprises detecting a statistically deviant spectral value in the distribution of the spectral values of the descendants at a finer scale layer of the background feature.

### Claim 8 (depends on 7)

The method of claim 7, wherein the outlier in the spectral distribution comprises a spectral value that lies at a statistical distance greater than a specified value from the mean value of the spectral distribution.

### Claim 9 (depends on 1)

The method of claim 1, wherein constructing the anomaly layer from the detected anomalies comprises agglomerating anomalous polygons using a segmentation method.

### Claim 10 (depends on 9)

The method of claim 9, wherein agglomerating anomalous polygons to form the anomaly layer further comprises preserving an anomaly detected in one of the background polygons.

### Claim 11 (depends on 1)

The method of claim 1, further comprising determining a non-pervasive anomaly in response to a comparison of a first image to a second image.

### Claim 12 (depends on 11)

The method of claim 11, wherein the comparison of the first image to the second image comprises a difference image of the first image and the second image.

### Claim 13 (depends on 12)

The method of claim 12, further comprising for the difference image:
segmenting the difference image into a hierarchy of polygonal feature layers comprising fine-scale foreground layers and progressively coarser-scale background layers;
detecting a non-pervasive anomaly with respect to the background layers in the segmentation of the difference image; and
constructing a change layer such that the detected non-pervasive anomaly of the difference image is included as a change in the change layer.

### Claim 14 (depends on 13)

The method of claim 13, wherein detecting a non-pervasive anomaly with respect to the various background polygonal features in the segmentation of the difference image comprises detecting a statistically deviant spectral value in the distribution of spectral values of descendants at a finer-scale layer of a background polygon in a hierarchical segmentation of the difference image.

### Claim 15 (depends on 14)

The method of claim 14, wherein detecting a statistically deviant spectral value comprises identifying a fine-scale polygon corresponding to the outlier spectral value in the spectral distribution of the background polygon in the hierarchical segmentation of the difference image.

### Claim 16 (depends on 14)

The method of claim 14, wherein detecting an outlier in the spectral distribution of the background polygon in the hierarchical segmentation of the difference image comprises finding an element of the distribution that is at a statistical distance greater than a prescribed value from the mean value of the spectral distribution.

### Claim 17 (depends on 15)

The method of claim 15, further comprising constructing a statistical distribution of the background polygon using a joint average spectral value of a region of each of the first image and the second image corresponding to a predetermined descendant polygon.
