Back to results
US 10,380,447 B1Granted
Providing regions of interest in an image
AssigneeGoogle Inc.
InventorsKrzysztof Kulewski, Aravind Krishnaswamy, Sevket Derin Babacan
Life of this patent
Estimating the term…
Abstract
Implementations generally relate to providing regions of interest in an image. In some implementations, a method includes receiving an image. The method further includes determining one or more image features. The method further includes grouping the one or more image features into one or more regions of interest.
Claims20
A computer-implemented method comprising:
- receiving an image;
- determining two or more image features in the image, wherein the two or more image features include one or more faces and one or more non-facial image features;
- grouping the two or more image features into one or more regions that separate the one or more faces from the one or more non-facial image features such that each feature of the two or more image features is located in one of the one or more regions, wherein the grouping includes:
- determining a respective initial region for each image feature of the two or more image features, wherein the respective initial region includes all pixels of the image feature;
- determining that at least a first initial region border intersects a second image feature border within a second initial region;
- and
- in response to determining that the first initial region border and the second image feature border intersect, determining a merged region within a single image that includes a first portion having a first image feature of the first initial region and a second portion having a second image feature of the second initial region, wherein the first image feature corresponds to different pixels from pixels of the second image feature;
- providing a user interface on a display screen, wherein the user interface includes the image and enables a user to select at least one region of interest that includes at least one of the one or more faces from the one or more regions;
- and
- in response to user selection of the at least one region of interest, providing a modified image that includes corresponding pixels for the at least one region of interest with the at least one of the one or more faces and excludes other regions of the one or more regions with the one or more non-facial image features.
7 more elements
A non-transitory computer-readable storage medium carrying program instructions thereon, the instructions when executed by one or more processors cause the one or more processors to perform operations comprising:
- receiving an image;
- determining two or more image features in the image, wherein the two or more image features include one or more faces and one or more non-facial image features;
- grouping the two or more image features into one or more regions that separate the one or more faces from the one or more non-facial image features such that each feature of the two or more image features is located in one of the one or more regions, wherein the grouping includes:
- determining a respective initial region for each image feature of the two or more image features, wherein the respective initial region includes all pixels of the image feature;
- and
- merging at least two of the initial regions based on one or more predetermined merging criteria to generate a merged region within a single image containing a portion of each of the at least two initial regions, wherein the merged region contains the two or more image features, and wherein the one or more predetermined merging criteria include determining that a first initial region border intersects a second image feature border within a second initial region;
- providing a user interface on a display screen, wherein the user interface includes the image and enables user input to select at least one of the one or more regions that includes at least one of the one or more faces;
- and
- causing a modified image to be displayed based on the at least one selected region, wherein the modified image includes corresponding pixels for the at least one selected region with the at least one of the one or more faces and no other regions of the one or more regions.
6 more elements
A system comprising:
- one or more processors;
- and
- logic encoded in one or more tangible media for execution by the one or more processors and when executed operable to perform operations comprising:
- receiving a single image;
- determining one or more image features in the single image;
- grouping the one or more image features into one or more regions within the single image such that each region includes a plurality of pixels of the single image;
- determining that a border of a first region of the one or more regions intersects a border of a particular image feature of the one or more image features;
- merging the first region with a second region of the one or more regions that includes the particular image feature to generate a merged region within the single image, such that the merged region includes each pixel of the first region and each pixel of the second region such that the merged region includes the particular image feature as a whole;
- selecting at least one of the one or more regions that satisfies one or more predetermined criteria;
- determining an importance score for each image feature based on one or more predetermined factors;
- modifying the single image to include corresponding pixels for the at least one of the one or more regions based on the importance scores, wherein the modifying comprises at least one of:
- cropping the single image based on the at least one of the one or more regions such that regions that are not selected are excluded;
- and
- applying one or more filters to the single image that emphasize the at least one of the one or more regions, wherein applying the one or more filters comprises multiplying pixels inside the at least one of the one or more regions by a predetermined multiple to increase values of the pixels;
- and
- providing the modified single image in a user interface on a display screen.
13 more elements
Dependencies are inferred from the claim text.