Systems and methods for identifying salient images

Patent metadata

Field	Value
Publication number	US-10380452-B1
Application number	US-201715594429-A
Country	US
Kind code	B1
Filing date	May 12, 2017
Priority date	May 12, 2017
Publication date	Aug 13, 2019
Grant date	Aug 13, 2019

Abstract

Image information defining an image may be accessed. The image may include one or more salient objects. A saliency map may be generated based on the image information. The saliency map may include one or more regions corresponding to the one or more salient objects. The one or more regions may be characterized by different levels of intensity than other regions of the saliency map. One or more salient regions around the one or more salient objects may be identified based on the saliency map. A saliency metric for the image may be generated based on one or more of (1) sizes of the one or more salient regions; (2) an amount of the one or more salient regions; and/or (3) histograms within the one or more salient regions.

First claim

What is claimed is:

1. A system that identifies salient images, the system comprising:
one of more physical processors configured by machine-readable instructions to:
access image information defining an image, the image including salient objects;
generate a saliency map based on the image information, the saliency map including regions corresponding to the salient objects, the regions characterized by different levels of intensity than other regions of the saliency map;
identify salient regions around the salient objects based on the saliency map, individual salient regions including individual salient objects and at least two of the salient regions having different sizes; and
generate a saliency metric for the image based on (1) a comparison of individual sizes of the salient regions; and (2) a number of the salient regions.

2. The system of claim 1 , wherein the saliency map is generated using a single color space or multiple color spaces.

3. The system of claim 1 , wherein the saliency map is generated using connectivity of pixels within the image to a boundary of the image.

4. The system of claim 1 , wherein the one or more physical processors are, to generate the saliency map based on the image information, further configured by the machine-readable instruction to generate a distance map per a channel of one or more color spaces, the distance map characterized by a path cost function.

5. The system of claim 1 , wherein the comparison of the individual sizes of the salient regions includes a comparison of relative sizes of two biggest salient regions, and the number of the salient regions is compared with a threshold number.

6. The system of claim 1 , wherein the saliency metric is generated further based on histograms within the salient regions indicating that a threshold percentage of pixels inside the salient regions is characterized by a high intensity or a low intensity, the high intensity including top 4 percent of an intensity range of the image and the low intensity including bottom 4 percent of the intensity range of the image.

7. The system of claim 1 , wherein the salient regions are identified by bounding boxes, the bounding boxes extracted using a flood-fill algorithm.

8. The system of claim 7 , wherein the flood-fill algorithm is started at a given pixel of the saliency map based on the given pixel having an intensity value meeting a threshold value.

9. The system of claim 1 , wherein the saliency metric for the image is further based on a linear classification of the image.

10. A method for identifying salient images, the method performed by a computing system including one or more physical processors, the method comprising:
accessing, by the one or more physical processors, image information defining an image, the image including salient objects;
generating, by the one or more physical processors, a saliency map based on the image information, the saliency map including regions corresponding to the salient objects, the regions characterized by different levels of intensity than other regions of the saliency map;
identifying, by the one or more physical processors, salient regions around the salient objects based on the saliency map, individual salient regions including individual salient objects and at least two of the salient regions having different sizes; and
generating, by the one or more physical processors, a saliency metric for the image based on (1) a comparison of individual sizes of the salient regions; and (2) a number of the salient regions.

11. The method of claim 10 , wherein the saliency map is generated using a single color space or multiple color spaces.

12. The method of claim 10 , wherein the saliency map is generated using connectivity of pixels within the image to a boundary of the image.

13. The method of claim 10 , wherein generating the saliency map based on the image information includes generating a distance map per a channel of one or more color spaces, the distance map characterized by a path cost function.

14. The method of claim 10 , wherein the comparison of the individual sizes of the salient regions includes a comparison of relative sizes of two biggest salient regions, and the number of the salient regions is compared with a threshold number.

15. The method of claim 10 , wherein the saliency metric is generated further based on histograms within the salient regions indicating that a threshold percentage of pixels inside the salient regions is characterized by a high intensity or a low intensity, the high intensity including top 4 percent of an intensity range of the image and the low intensity including bottom 4 percent of the intensity range of the image.

16. The method of claim 10 , wherein the salient regions are identified by bounding boxes, the bounding boxes extracted using a flood-fill algorithm.

17. The method of claim 16 , wherein the flood-fill algorithm is started at a given pixel of the saliency map based on the given pixel having an intensity value meeting a threshold value.

18. The method of claim 10 , wherein the saliency metric for the image is further based on a linear classification of the image.

19. A system that identifies salient images, the system comprising:
one of more physical processors configured by machine-readable instructions to:
access image information defining an image, the image including salient objects;
generate a saliency map based on the image information, the saliency map is generated using connectivity of pixels within the image to a boundary of the image, the saliency map including regions corresponding to the salient objects, the regions characterized by different levels of intensity than other regions of the saliency map;
identify salient regions around the salient objects based on the saliency map, the salient regions identified by bounding boxes individual bounding boxes including individual salient objects and at least two of the bounding boxes having different sizes; and
generate a saliency metric for the image based on (1) a comparison of individual sizes of the bounding boxes; and (2) a number of the bounding boxes.

20. The system of claim 19 , wherein the one or more physical processors are, to generate the saliency map based on the image information, further configured by the machine-readable instruction to generate a distance map per a channel of one or more color spaces, the distance map characterized by a path cost function.

Assignees

• Gopro Inc

Inventors

• Piovano Jerome

Classifications

• G06T7/11 Primary

  Region-based segmentation · CPC title

• G06V10/464

  using a plurality of salient features, e.g. bag-of-words [BoW] representations · CPC title

• G06F18/24

  Classification techniques · CPC title

• G06T7/187

  involving region growing; involving region merging; involving connected component labelling · CPC title

• G06T2207/10024

  Color image · CPC title