Apparatus, method, and program for discriminating subjects
First Claim
1. A subject discriminating apparatus comprising:
- an image input means for receiving input of a target image, which is a target of discrimination;
a first characteristic amount calculating means for calculating first characteristic amounts including directions of gradient vectors or color data, which do not require normalization and are employed to discriminate a predetermined subject, from the target image which is being currently received;
a first discriminating means for discriminating whether a candidate of the predetermined subject is included in the target image based on the first characteristic amounts, which are calculated from the target image only, by referring to first reference data, in which learned first characteristic amounts and first discrimination conditions corresponding to the learned first characteristic amounts are defined in advance from sample images of various subjects;
a second characteristic amount calculating means for calculating second characteristic amounts, which are normalized and employed to discriminate the predetermined subject, from the candidate of the predetermined subject only when the first discriminating means judges that the candidate is included in the target image;
a second discriminating means for discriminating whether the candidate is the predetermined subject based on the normalized second characteristic amounts, by referring to second reference data, in which learned second characteristic amounts and second discrimination conditions corresponding to the learned second characteristic amounts are defined in advance from the sample images; and
a control means, for comparing a number of times N1 that the first discriminating means judges that a candidate is included in the target image to a number of times N2 that the second discriminating means judges that the candidate is a face, judging whether a ratio of the number of times N2 to the number of times N1 is equal to or greater than a predetermined percentage when a number of target images for which discriminating has been performed is equal to a predetermined number, and if the judgment is affirmative, controlling the first characteristic amount calculating means, the first discriminating means, the second characteristic amount calculating means, and the second discriminating means so that the candidate discriminated by the first discriminating means is judged to be the predetermined subject by calculating only the first characteristic amounts from the target images which are subsequently discriminated.
2 Assignments
0 Petitions
Accused Products
Abstract
A characteristic amount calculating means calculates first characteristic amounts, which do not require normalization, and normalized second characteristic amounts. A first discriminating portion discriminates whether a candidate for a face is included in the target image, by referring to first reference data with the first characteristic amounts, calculated from the target image. The first reference data is obtained by learning the first characteristic amounts of a plurality of images, which are known either to be of faces or to not be of faces. In the case that the candidate is included, a second discriminating portion discriminates whether the candidate is a face, by referring to second reference data, obtained by learning the second characteristic amounts of a plurality of images, which known either to be of faces or to not to be of faces.
41 Citations
29 Claims
-
1. A subject discriminating apparatus comprising:
-
an image input means for receiving input of a target image, which is a target of discrimination; a first characteristic amount calculating means for calculating first characteristic amounts including directions of gradient vectors or color data, which do not require normalization and are employed to discriminate a predetermined subject, from the target image which is being currently received; a first discriminating means for discriminating whether a candidate of the predetermined subject is included in the target image based on the first characteristic amounts, which are calculated from the target image only, by referring to first reference data, in which learned first characteristic amounts and first discrimination conditions corresponding to the learned first characteristic amounts are defined in advance from sample images of various subjects; a second characteristic amount calculating means for calculating second characteristic amounts, which are normalized and employed to discriminate the predetermined subject, from the candidate of the predetermined subject only when the first discriminating means judges that the candidate is included in the target image; a second discriminating means for discriminating whether the candidate is the predetermined subject based on the normalized second characteristic amounts, by referring to second reference data, in which learned second characteristic amounts and second discrimination conditions corresponding to the learned second characteristic amounts are defined in advance from the sample images; and a control means, for comparing a number of times N1 that the first discriminating means judges that a candidate is included in the target image to a number of times N2 that the second discriminating means judges that the candidate is a face, judging whether a ratio of the number of times N2 to the number of times N1 is equal to or greater than a predetermined percentage when a number of target images for which discriminating has been performed is equal to a predetermined number, and if the judgment is affirmative, controlling the first characteristic amount calculating means, the first discriminating means, the second characteristic amount calculating means, and the second discriminating means so that the candidate discriminated by the first discriminating means is judged to be the predetermined subject by calculating only the first characteristic amounts from the target images which are subsequently discriminated. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26)
-
-
27. A subject discriminating method comprising:
-
receiving input of a target image, which is a target of discrimination; calculating first characteristic amounts including directions of gradient vectors or color data, which do not require normalization and are employed to discriminate a predetermined subject, from the target image which is being currently received; first discriminating whether a candidate of the predetermined subject is included in the target image based on the first characteristic amounts, which are calculated from the target image only, by referring to first reference data, in which learned first characteristic amounts and discrimination conditions corresponding to the learned first characteristic amounts are defined in advance from sample images of various subjects; calculating second characteristic amounts, which are normalized and employed to discriminate the predetermined subject, from the candidate of the predetermined subject if the first discriminating means judges that the candidate is included in the target image; second discriminating whether the candidate is the predetermined subject based on the normalized second characteristic amounts, by referring to second reference data, in which learned second characteristic amounts and discrimination conditions corresponding to the learned second characteristic amounts are defined in advance from the sample images; comparing a number of times N1 when it is judged in the first discriminating that a candidate is included in the target image to a number of times N2 when it is judged in the second discriminating that the candidate is a face and judging whether a ratio of the number of times N2 to the number of times N1 is equal to or greater than a predetermined percentage when a number of target images for which the first discriminating and the second discriminating have been performed is equal to a predetermined number; and if the judgment is affirmative, performing controlling of the calculating the first characteristic amounts, the first discriminating, the calculating the second characteristic amounts, and the second discriminating so that the candidate discriminated in the first discriminating is judged to be the predetermined subject by calculating only the first characteristic amounts from the target images which are subsequently discriminated. - View Dependent Claims (28)
-
-
29. A non-transitory computer-readable memory storing a program that, when executed by a computer, causes the computer to execute a subject discriminating method, the method comprising:
-
receiving input of a target image, which is a target of discrimination; calculating first characteristic amounts including directions of gradient vectors or color data, which do not require normalization and are employed to discriminate a predetermined subject, from the target image which is being currently received; first discriminating whether a candidate of the predetermined subject is included in the target image based on the first characteristic amounts, which are calculated from the target image only, by referring to first reference data, in which learned first characteristic amounts and discrimination conditions corresponding to the learned first characteristic amounts are defined in advance from sample images of various subjects; calculating second characteristic amounts, which are normalized and employed to discriminate the predetermined subject, from the candidate of the predetermined subject if the first discriminating means judges that the candidate is included in the target image; second discriminating whether the candidate is the predetermined subject based on the normalized second characteristic amounts, by referring to second reference data, in which learned second characteristic amounts and discrimination conditions corresponding to the learned second characteristic amounts are defined in advance from the sample images; comparing a number of times N1 that it is judged in the first discriminating that a candidate is included in the target image to a number of times N2 that is judged in the second discriminating that the candidate is a face and judging whether a ratio of the number of times N2 to the number of times N1 is equal to or greater than a predetermined percentage when a number of target images for which the first discriminating and the second discriminating have been performed is equal to a predetermined number; and if the judgment is affirmative, performing controlling of the calculating the first characteristic amounts, the first discriminating, the calculating second characteristic amounts, and the second discriminating so that the candidate discriminated in the first discriminating is judged to be the predetermined subject by calculating only the first characteristic amounts from the target images which are subsequently discriminated.
-
Specification