Image cropping
Briefly, embodiments disclosed herein relate to image cropping, such as for digital images, for example.
1. A method, comprising:
obtaining, for a region of a single image, a plurality of pre-composite saliency maps, each pre-composite saliency map of the plurality of pre-composite saliency maps corresponding to at least one of a plurality of saliency map types, at least one of the plurality of pre-composite saliency maps electronically modeling human eye-attention for the region of the single image;
generating, for the region of the single image, a composite saliency map from the plurality of pre-composite saliency maps;
determining, for the region of the single image, a cropping window based, at least in part, on the composite saliency map,
wherein the determining comprises:
responsive to determining that a potential cropping window has a first width exceeding a threshold, generating the cropping window using a first region of the single image and a second region of the single image;
obtaining, for a second region of a second single image, a second plurality of pre-composite saliency maps, each pre-composite saliency map of the second plurality of pre-composite saliency maps corresponding to at least one of a second plurality of saliency map types, at least one of the second plurality of pre-composite saliency maps electronically modeling human eye-attention for the second region of the second single image;
generating, for the second region of the second single image, a second composite saliency map from the second plurality of pre-composite saliency maps; and
determining, for the second region of the second single image, a second cropping window based, at least in part, on the second composite saliency map,
wherein the determining comprises:
responsive to determining that a second potential cropping window has a second width not exceeding the threshold, performing a region selection operation to select a third region of the second single image but not a fourth region of the second single image and generating the second cropping window using the third region of the second single image but not the fourth region of the second single image.
2. The method of claim 1 , wherein an entirety of the single image consists of the first region of the single image and the second region of the single image.
3. The method of claim 1 , wherein generating the composite saliency map comprises:
applying, for each pre-composite saliency map of the plurality of pre-composite saliency maps, a different one of a plurality of weights.
4. The method of claim 1 , wherein at least one pre-composite saliency map of the plurality of pre-composite saliency maps:
represents a first area, of the single image, estimated to attract human eye-attention with pixel values comprising a first value; and
represents a second area, of the single image, estimated to not attract human eye-attention with pixel values comprising a second value different than the first value.
5. The method of claim 1 , comprising:
detecting one or more separating boundaries in the single image; and
determining the region based, at least in part, on the one or more separating boundaries.
6. The method of claim 1 , the region of the single image comprising a contiguous portion of the single image.
7. The method of claim 1 , the region of the single image comprising non-contiguous portions of the single image.
8. An apparatus, comprising:
a processor; and
a memory, at least one of the processor or the memory being configured to:
obtain, for a region of a single image, a plurality of pre-composite saliency maps, each pre-composite saliency map of the plurality of pre-composite saliency maps corresponding to at least one of a plurality of saliency map types, at least one of the plurality of pre-composite saliency maps electronically modeling human eye-attention for the region of the single image;
generate, for the region of the single image, a composite saliency map from the plurality of pre-composite saliency maps;
determine, for the region of the single image, a cropping window based, at least in part, on the composite saliency map,
wherein the determining comprises:
responsive to determining that a potential cropping window has a first width exceeding a threshold, generating the cropping window using a first region of the single image and a second region of the single image;
obtain, for a second region of a second single image, a second plurality of pre-composite saliency maps, each pre-composite saliency map of the second plurality of pre-composite saliency maps corresponding to at least one of a second plurality of saliency map types, at least one of the second plurality of pre-composite saliency maps electronically modeling human eye-attention for the second region of the second single image;
generate, for the second region of the second single image, a second composite saliency map from the second plurality of pre-composite saliency maps; and
determine, for the second region of the second single image, a second cropping window based, at least in part, on the second composite saliency map,
wherein the determining comprises:
responsive to determining that a second potential cropping window has a second width not exceeding the threshold, performing a region selection operation to select a third region of the second single image but not a fourth region of the second single image and generating the second cropping window using the third region of the second single image but not the fourth region of the second single image.
9. The apparatus of claim 8 , wherein an entirety of the single image consists of the first region of the single image and the second region of the single image.
10. The apparatus of claim 8 , wherein generating the composite saliency map comprises:
applying, for each pre-composite saliency map of the plurality of pre-composite saliency maps, a different one of a plurality of weights.
11. The apparatus of claim 8 , wherein at least one pre-composite saliency map of the plurality of pre-composite saliency maps:
represents a first area, of the single image, estimated to attract human eye-attention with pixel values comprising a first value; and
represents a second area, of the single image, estimated to not attract human eye-attention with pixel values comprising a second value different than the first value.
12. The apparatus of claim 8 , at least one of the processor or the memory being configured to:
detect one or more separating boundaries in the single image; and
determine the region based, at least in part, on the one or more separating boundaries.
13. The apparatus of claim 8 , the region of the single image comprising a contiguous portion of the single image.
14. The apparatus of claim 8 , the region of the single image comprising non-contiguous portions of the single image.
15. A computer program product comprising at least one non-transitory computer readable storage medium having computer program instructions stored thereon, the computer program instructions being configured such that, when executed by one or more processors, the computer program instructions cause the one or more processors to:
obtain, for a region of a single image, a plurality of pre-composite saliency maps, each pre-composite saliency map of the plurality of pre-composite saliency maps corresponding to at least one of a plurality of saliency map types, at least one of the plurality of pre-composite saliency maps electronically modeling human eye-attention for the region of the single image;
generate, for the region of the single image, a composite saliency map from the plurality of pre-composite saliency maps;
determine, for the region of the single image, a cropping window based, at least in part, on the composite saliency map,
wherein the determining comprises:
responsive to determining that a potential cropping window has a first width exceeding a threshold, generating the cropping window using a first region of the single image and a second region of the single image;
obtain, for a second region of a second single image, a second plurality of pre-composite saliency maps, each pre-composite saliency map of the second plurality of pre-composite saliency maps corresponding to at least one of a second plurality of saliency map types, at least one of the second plurality of pre-composite saliency maps electronically modeling human eye-attention for the second region of the second single image;
generate, for the second region of the second single image, a second composite saliency map from the second plurality of pre-composite saliency maps; and
determine, for the second region of the second single image, a second cropping window based, at least in part, on the second composite saliency map,
wherein the determining comprises:
responsive to determining that a second potential cropping window has a second width not exceeding the threshold, performing a region selection operation to select a third region of the second single image but not a fourth region of the second single image and generating the second cropping window using the third region of the second single image but not the fourth region of the second single image.
16. The computer program product of claim 15 , wherein an entirety of the single image consists of the first region of the single image and the second region of the single image.
17. The computer program product of claim 15 , wherein generating the composite saliency map comprises:
applying, for each pre-composite saliency map of the plurality of pre-composite saliency maps, a different one of a plurality of weights.
18. The computer program product of claim 15 , wherein at least one pre-composite saliency map of the plurality of pre-composite saliency maps:
represents a first area, of the single image, estimated to attract human eye-attention with pixel values comprising a first value; and
represents a second area, of the single image, estimated to not attract human eye-attention with pixel values comprising a second value different than the first value.
19. The computer program product of claim 15 , the computer program instructions being configured to:
detect one or more separating boundaries in the single image; and
determine the region based, at least in part, on the one or more separating boundaries.
20. The computer program product of claim 15 , the region of the single image comprising a contiguous portion of the single image.