Information processing apparatus, information processing method, and computer program product
According to an embodiment, an information processing apparatus includes an extraction unit and a generation unit. The extraction unit is configured to use a learning model that extracts a combining region from a subject image to extract the combining region from a first subject image of a first subject. The generation unit is configured to generate a combined image obtained by combining the extracted combining region with a target region of a corrected image in which transparency of the target region in a second subject image of a second subject has been corrected to a threshold or more.
1. An information processing apparatus comprising:
one or more hardware processors configured to:
extract, by using a learning model for extracting a combining region from a subject image, the combining region from a first subject image of a first subject; and
generate a combined image obtained by combining the combining region with a target region of a corrected image in which transparency of the target region in a second subject image of a second subject has been corrected to a threshold or more, wherein
generation of the combined image comprises combining the combining region with the target region of the corrected image obtained by adjusting an alpha value indicating the transparency of a boundary region between an inside and an outside of the target region in the second subject image so as to be stepwise or continuously reduced from the inside of the target region to the outside of the target region, and
the one or more hardware processors are further configured to:
correct a color tone of the combining region such that a difference between a color tone of a feature region characteristically representing a skin color and including a nose tip region, a nostril region, and a skin region adjacent the nose tip region in the first subject image and a color tone of the feature region in the second subject image is a predetermined value or less, yielding a corrected combining region; and
generate the combined image by combining the corrected combining region with the target region of the corrected image.
2. The apparatus according to claim 1 , wherein the second subject image, in which the transparency is corrected, is an image including a face and clothing of the second subject.
3. The apparatus according to claim 1 , wherein the second subject image, in which the transparency is corrected, is an image including a trial target site of the first subject.
4. The apparatus according to claim 3 , wherein the trial target site is at least one of a hair site, a skin site, or an item worn by the second subject.
5. The apparatus according to claim 3 , wherein the combining region is a region adjacent to the trial target site in the first subject image.
6. The apparatus according to claim 5 , wherein the trial target site is a hair site, and the combining region is a face region in the first subject image.
7. The apparatus according to claim 1 , wherein the combining region and the target region of the corrected image are rectangular regions.
8. The apparatus according to claim 1 , wherein the one or more hardware processors are further configured to:
reshape the corrected image so that the target region of the corrected image and the combining region match with each other to yield a reshaped corrected image; and
generate the combined image by combining the combining region with the target region of the reshaped corrected image.
9. The apparatus according to claim 1 , wherein the one or more hardware processors are configured to generate the combined image for which at least one of a region between the first subject's nose and lips or a cheek region is the skin region.
10. The apparatus according to claim 1 , wherein the one or more hardware processors are further configured to output the combined image.
11. The apparatus according to claim 10 , wherein
the one or more hardware processors are further configured to output the combined image and related information related to the second subject image corresponding to the corrected image used for combining the combined image.
12. The apparatus according to claim 11 , wherein
the related information is at least one of: supplementary information related to the second subject image; affinity between the feature of the first subject and a trial target site of the second subject image; a user name other than the first subject similar to the combined image; characteristic information indicating a characteristic of the combined image; or difference information indicating a difference between the first subject and the second subject.
13. The apparatus according to claim 11 , wherein
the one or more hardware processors are further configured to output the combined image selected based on the related information among the plurality of generated combined images.
14. An information processing method implemented by a computer, the method comprising:
using a learning model to extract a combining region from a first subject image of a first subject; and
generating a combined image obtained by combining the combining region with a target region of a corrected image in which transparency of the target region in a second subject image of a second subject has been corrected to a threshold or more, wherein
the generating comprises generating the combined image obtained by combining the combining region with the target region of the corrected image obtained by adjusting an alpha value indicating the transparency of a boundary region between an inside and an outside of the target region in the second subject image so as to be stepwise or continuously reduced from the inside of the target region to the outside of the target region, and
the method further comprises:
correcting a color tone of the combining region such that a difference between a color tone of a feature region characteristically representing a skin color and including a nose tip region, a nostril region, and a skin region adjacent the nose tip region in the first subject image and a color tone of the feature region in the second subject image is a predetermined value or less, wherein the correcting yields a corrected combining region; and
generating the combined image by combining the corrected combining region with the target region of the corrected image.
15. A computer program product comprising a non-transitory computer-readable recording medium including an information processing program, the program causing a computer to execute:
using a learning model to extract a combining region from a first subject image of a first subject; and
generating a combined image obtained by combining the combining region with a target region of a corrected image in which transparency of the target region in a second subject image of a second subject has been corrected to a threshold or more, wherein
the generating generates the combined image obtained by combining the combining region with the target region of the corrected image obtained by adjusting an alpha value indicating the transparency of a boundary region between an inside and an outside of the target region in the second subject image so as to be stepwise or continuously reduced from the inside of the target region to the outside of the target region, and
the program causes the computer to further execute:
correcting a color tone of the combining region such that a difference between a color tone of a feature region characteristically representing a skin color and including a nose tip region, a nostril region, and a skin region adjacent the nose tip region in the first subject image and a color tone of the feature region in the second subject image is a predetermined value or less, wherein the correcting yields a corrected combining region; and
generating the combined image by combining the corrected combining region with the target region of the corrected image.