Selective redaction of images

Patent metadata

Field	Value
Publication number	US-11468617-B2
Application number	US-202117197115-A
Country	US
Kind code	B2
Filing date	Mar 10, 2021
Priority date	Mar 10, 2021
Publication date	Oct 11, 2022
Grant date	Oct 11, 2022

Abstract

Selectively redacting an image by determining a set of attributes used by a machine learning model for an analysis, receiving image data detecting, by the one or more computer processors, a portion of the image data relevant to the analysis, the portion comprising at least some of the set of attributes, generating a synthetic portion from the portion, wherein the synthetic portion retains at least some of the attributes of the detected portion, replacing the portion with the synthetic portion, yielding redacted image data, and providing the redacted image data for analysis.

First claim

What is claimed is:

1. A computer implemented method for selectively redacting an image, the method comprising:
receiving image data, by one or more computer processors;
detecting, by the one or more computer processors, a portion of the image data relevant to an analysis, the portion comprising a set of attributes;
generating, by the one or more computer processors, a synthetic portion from the portion, wherein the synthetic portion retains at least some of the set of attributes of the portion;
detecting, by the one or more computer processors, a set of portions;

for each portion of the set of portions:
generating, by the one or more computer processors, an avatar for the portion;
applying, by the one or more computer processors, an overlay to each of the avatar and the portion, the overlay comprising a plurality of cells;
replacing, by the one or more computer processors, cells of the portion with cells of the avatar according to a transform, yielding a set of synthetic portions, each synthetic portion of the set of synthetic portions associated with one transform of a set of transforms;
analyzing, by the one or more computer processors, the set of synthetic portions using a machine learning model;
analyzing, by the one or more computer processors, the set of portions using the machine learning model;
determining, by the one or more computer processors, an accuracy according to the analysis of the set of synthetic portions and the analysis of the set of portions;
selecting, by the one or more computer processors, a transform from the set of transforms according to the accuracy; and
utilizing, by the one or more computer processors, the transform to generate synthetic portions from portions;
replacing, by the one or more computer processors, the portion with the synthetic portion, yielding redacted image data; and
providing, by the one or more computer processors, the redacted image data for the analysis.

2. The computer implemented method according to claim 1 , further comprising:
analyzing, by the one or more computer processors, the redacted image data using a machine learning model.

3. The computer implemented method according to claim 1 , wherein the image data comprises video data.

4. The computer implemented method according to claim 1 , further comprising:
generating, by the one or more computer processors, an avatar for the portion;
applying, by the one or more computer processors, an overlay to each of the avatar and the portion, the overlay comprising a plurality of cells; and
replacing, by the one or more computer processors, cells of the portion with cells of the avatar, yielding the synthetic portion.

5. The computer implemented method according to claim 4 , further comprising:
detecting, by the one or more computer processors, at least some of the set of attributes of the portion; and
adjusting, by the one or more computer processors, the avatar according to at least some of the detected set of attributes.

6. A computer program product for selectively redacting an image, the computer program product comprising one or more computer readable storage devices and collectively stored program instructions on the one or more computer readable storage devices, the stored program instructions comprising:
program instructions to receive image data;
program instructions to detect a portion of the image data relevant to an analysis, the portion comprising a set of attributes;
program instructions to detect a set of portions;
for each portion of the set of portions:
program instructions to generate an avatar for the portion;
program instructions to apply an overlay to each of the avatar and the portion, the overlay comprising a plurality of cells;
program instructions to replace cells of the portion with cells of the avatar according to a transform, yielding a set of synthetic portions, each synthetic portion of the set of synthetic portions associated with one transform of a set of transforms;
program instructions to analyze the set of synthetic portions using a machine learning model;
program instructions to analyze the set of portions using the machine learning model;
program instructions to determine an accuracy according to the analysis of the set of synthetic portions and the analysis of the set of portions;
program instructions to select a transform from the set of transforms according to the accuracy; and

program instructions to utilize the transform to generate synthetic portions from portions;
program instructions to generate a synthetic portion from the portion, wherein the synthetic portion retains at least some of the set of attributes of the portion;
program instructions to replace the portion with the synthetic portion, yielding redacted image data; and
program instructions to provide the redacted image data for the analysis.

7. The computer program product according to claim 6 , further comprising:
program instructions to analyze the redacted image data using a machine learning model.

8. The computer program product according to claim 6 , wherein the image data comprises video data.

9. The computer program product according to claim 6 , further comprising:
program instructions to generate an avatar for the portion;
program instructions to apply an overlay to each of the avatar and portion, the overlay comprising a plurality of cells; and
program instructions to replace cells of the portion with cells of the avatar, yielding the synthetic portion.

10. The computer program product according to claim 9 , further comprising:
program instructions to detect at least some of the set of attributes of the portion; and
program instructions to adjust the avatar according to at least some of the detected set of attributes.

11. A computer system for selectively redacting an image, the computer system comprising:
one or more computer processors;
one or more computer readable storage devices; and
stored program instructions on the one or more computer readable storage devices for execution by the one or more computer processors, the stored program instructions comprising:
program instructions to receive image data;
program instructions to detect a portion of the image data relevant to an analysis, the portion comprising a set of attributes;

program instructions to detect a set of portions;
for each portion of the set of portions:
program instructions to generate an avatar for the portion;
program instructions to apply an overlay to each of the avatar and the portion, the overlay comprising a plurality of cells;
program instructions to replace cells of the portion with cells of the avatar according to a transform, yielding a set of synthetic portions, each synthetic portion of the set of synthetic portions associated with one transform of a set of transforms;
program instructions to analyze the set of synthetic portions using a machine learning model;
program instructions to analyze the set of portions using the machine learning model;
program instructions to determine an accuracy according to the analysis of the set of synthetic portions and the analysis of the set of portions;
program instructions to select a transform from the set of transforms according to the accuracy; and
program instructions to utilize the transform to generate synthetic portions from portions;
program instructions to generate a synthetic portion from the portion, wherein the synthetic portion retains at least some of the set of attributes of the portion;
program instructions to replace the portion with the synthetic portion, yielding redacted image data; and
program instructions to provide the redacted image data for the

Assignees

• IBM

Inventors

• Pinel Florian
• Coldicott Peter Alan
• Yeh Jr Frank
• Bobbitt Russell Patrick

Classifications

• G06T11/00 Primary

  Two-dimensional [2D] image generation · CPC title

• G06T13/40 Primary

  of characters, e.g. humans, animals or virtual beings · CPC title

• G06N20/00

  Machine learning · CPC title