Face recognition-matching system effective to images obtained in different imaging conditions

Claim 1

A system which enables pattern matching between a first face image obtained in a first imaging condition and a second face image obtained in a second imaging condition noisier than the first imaging condition, the system comprising: first means for obtaining, for a plurality of sample faces, respective first sample images in a first sample imaging condition similar to or the same as said first imaging condition; second means for obtaining, for said plurality of said sample faces, respective second sample images in a second sample imaging condition similar to or the same as said second imaging condition; means for calculating a weighted sum of a first covariance matrix calculated from said first sample images and a second covariance matrix calculated from differences between said first sample images and said second sample images; means for applying a principal component analysis to said weighted sum to find a result with which principal components are calculated from a given face image; and means for enabling said result to be used for a later pattern matching process executed within the system or by an external system, wherein said means for applying a principal component analysis to said weighted sum comprises means operative on the basis of the spectrum resolution of said weighted sum for finding a characterizing matrix which, when applied to a given face image vector, yields a principal component vector for said given face image vector.

Claim 2

A system which enables pattern matching between a first face image obtained in a first imaging condition and a second face image obtained in a second imaging condition noisier than the first imaging condition, the system comprising: first means for obtaining, for a plurality of sample faces, respective first sample images in a first sample imaging condition similar to or the same as said first imaging condition; second means for obtaining, for said plurality of said sample faces, respective second sample images in a second sample imaging condition similar to or the same as said second imaging condition; means for calculating a weighted sum of a first covariance matrix calculated from said first sample images and a second covariance matrix calculated from differences between said first sample images and said second sample images; means for applying a principal component analysis to said weighted sum to find a result with which principal components are calculated from a given face image; and means for enabling said result to be used for a later pattern matching process executed within the system or by an external system, wherein said means for applying a principal component analysis to said weighted sum comprises means operative on the basis of the spectrum resolution of said weighted sum for finding a characterizing matrix which, when applied to a given face image vector, yields a principal component vector for said given face image vector.

Claim 4

A system which enables pattern matching between a first face image obtained in a first imaging condition and a second face image obtained in a second imaging condition noisier than the first imaging condition, the system comprising: first means for obtaining, for a plurality of sample faces, respective first sample images in a first sample imaging condition similar to or the same as said first imaging condition; second means for obtaining, for said plurality of said sample faces, respective second sample images in a second sample imaging condition similar to or the same as said second imaging condition; means for calculating a weighted sum of a first covariance matrix calculated from said first sample images and a second covariance matrix calculated from differences between said first sample images and said second sample images; means for applying a principal component analysis to said weighted sum to find a result with which principal components are calculated from a given face image; and means for enabling said result to be used for a later pattern matching process executed within the system or by an external system, wherein: letting said weighted sum, said first covariance matrix calculated from said first sample images, and said second covariance matrix calculated from said differences be Cs, Cm and Cp, respectively, said means for calculating a weighted sum comprises means for obtaining Cs=a Cm+(1−α)Cp, where α is a weight coefficient and 0≦α<1.

Claim 5

A system which enables pattern matching between a first face image obtained in a first imaging condition and a second face image obtained in a second imaging condition noisier than the first imaging condition, the system comprising: first means for obtaining, for a plurality of sample faces, respective first sample images in a first sample imaging condition similar to or the same as said first imaging condition; second means for obtaining, for said plurality of said sample faces, respective second sample images in a second sample imaging condition similar to or the same as said second imaging condition; means for calculating a weighted sum of a first covariance matrix calculated from said first sample images and a second covariance matrix calculated from differences between said first sample images and said second sample images; means for applying a principal component analysis to said weighted sum to find a result with which principal components are calculated from a given face image; and means for enabling said result to be used for a later pattern matching process executed within the system or by an external system, wherein: letting said weighted sum, said first covariance matrix calculated from said first sample images, and said second covariance matrix calculated from said differences be Cs, Cm and Cp, respectively, said means for calculating a weighted sum comprises means for obtaining Cs=a Cm+(1−α)Cp, where α is a weight coefficient and 0≦α<1.

Claim 6

A system which enables pattern matching between a first face image obtained in a first imaging condition and a second face image obtained in a second imaging condition noisier than the first imaging condition, the system comprising: first means for obtaining, for a plurality of sample faces, respective first sample images in a first sample imaging condition similar to or the same as said first imaging condition; second means for obtaining, for said plurality of said sample faces, respective second sample images in a second sample imaging condition similar to or the same as said second imaging condition; means for calculating a weighted sum of a first covariance matrix calculated from said first sample images and a second covariance matrix calculated from differences between said first sample images and said second sample images; means for applying a principal component analysis to said weighted sum to find a result with which principal components are calculated from a given face image, and means for enabling said result to be used for a later pattern matching process executed within the system or by an external system, wherein assuming that said first sample images are V 1 , V 2 , . . . , V N (N is the number of sample faces), said second sample images are U 1 , U 2 , . . . , U N , said differences are Δ 1 , Δ 2 , . . . , Δ N , and said weighted sum is Cs, then said means for calculating a weighted sum comprises means for calculating Cs = α   ∑ i = 1 N  ( V i - V _ )  ( V i - V _ ) T + ( 1 - α )   ∑ i = 1 N  Δ i  Δ i T where V _ = 1 N  ∑ i = 1 N  V i , Δ i = V i - U i and α is a weight coefficient in a range 0≦α<1.

Claim 7

A system which enables pattern matching between a first face image obtained in a first imaging condition and a second face image obtained in a second imaging condition noisier than the first imaging condition, the system comprising: first means for obtaining, for a plurality of sample faces, respective first sample images in a first sample imaging condition similar to or the same as said first imaging condition; second means for obtaining, for said plurality of said sample faces, respective second sample images in a second sample imaging condition similar to or the same as said second imaging condition; means for calculating a weighted sum of a first covariance matrix calculated from said first sample images and a second covariance matrix calculated from differences between said first sample images and said second sample images; means for applying a principal component analysis to said weighted sum to find a result with which principal components are calculated from a given face image, and means for enabling said result to be used for a later pattern matching process executed within the system or by an external system, wherein assuming that said first sample images are V 1 , V 2 , . . . , V N (N is the number of sample faces), said second sample images are U 1 , U 2 , . . . , U N , said differences are Δ 1 , Δ 2 , . . . , Δ N , and said weighted sum is Cs, then said means for calculating a weighted sum comprises means for calculating Cs = α   ∑ i = 1 N  ( V i - V _ )  ( V i - V _ ) T + ( 1 - α )   ∑ i = 1 N  Δ i  Δ i T where V _ = 1 N  ∑ i = 1 N  V i , Δ i = V i - U i and α is a weight coefficient in a range 0≦α<1.

Claim 10

A system which enables pattern matching between a first face image having been obtained in a first imaging condition and a second face image having been obtained in a second imaging condition noisier than the first imaging condition, the system comprising: first means for obtaining, for a plurality (N) of first sample faces, respective first sample images in a first sample imaging condition similar to or the same as said first imaging condition; second means for obtaining, for said N sample faces, respective second sample images in a second sample imaging condition similar to or the same as said second imaging condition, said first means obtaining, for a plurality (L) of second sample faces, respective third sample images in said sample imaging condition; means for calculating a weighted sum of a covariance matrix calculated from difference between said first sample images and said second sample images and a covariance matrix calculated from said third sample images; means for applying a principal component analysis to said weighted sum to find a result with which principal components are calculated from a given face image; and means for enabling said result to be used for a later pattern matching process executed within the system or by an external system, wherein assuming that said first sample images are V 1 , V 2 , . . . , V N (N is the number of sample faces), said second sample images are U 1 , U 2 , . . . , U N , said differences are Δ 1 , Δ 2 , . . . , Δ N , said third sample images are V 1 ′, V 2 ′ V L ′ (L is the number of second sample faces) and said weighted sum is Cs, then said means for calculating a weighted sum comprises means for calculating Cs = α   ∑ i = 1 L  ( V i ′ - V ′ _ )  ( V i ′ - V ′ _ ) T + ( 1 - α )   ∑ i = 1 N  Δ i  Δ i T where V ′ _ = 1 L  ∑ i = 1 L  V i ′ , Δ i = V i - U i and α is a weight coefficient in a range 0≦α<1.

Claim 11

A system which enables pattern matching between a first face image having been obtained in a first imaging condition and a second face image having been obtained in a second imaging condition noisier than the first imaging condition, the system comprising: first means for obtaining, for a plurality (N) of first sample faces, respective first sample images in a first sample imaging condition similar to or the same as said first imaging condition; second means for obtaining, for said N sample faces, respective second sample images in a second sample imaging condition similar to or the same as said second imaging condition, said first means obtaining, for a plurality (L) of second sample faces, respective third sample images in said sample imaging condition; means for calculating a weighted sum of a covariance matrix calculated from difference between said first sample images and said second sample images and a covariance matrix calculated from said third sample images; means for applying a principal component analysis to said weighted sum to find a result with which principal components are calculated from a given face image; and means for enabling said result to be used for a later pattern matching process executed within the system or by an external system, wherein assuming that said first sample images are V 1 , V 2 , . . . , V N (N is the number of sample faces), said second sample images are U 1 , U 2 , . . . , U N , said differences are Δ 1 , Δ 2 , . . . , Δ N , said third sample images are V 1 ′, V 2 ′ V L ′ (L is the number of second sample faces) and said weighted sum is Cs, then said means for calculating a weighted sum comprises means for calculating Cs = α   ∑ i = 1 L  ( V i ′ - V ′ _ )  ( V i ′ - V ′ _ ) T + ( 1 - α )   ∑ i = 1 N  Δ i  Δ i T where V ′ _ = 1 L  ∑ i = 1 L  V i ′ , Δ i = V i - U i and α is a weight coefficient in a range 0≦α<1.

Claim 12

A system capable of pattern matching between a first face image having been obtained in a first imaging condition and a second face image having been obtained in a second imaging condition noisier than the first imaging condition, the system comprising: data means stored in a storage for use in extracting principal components from a given face image, said data means having been obtained by: for a plurality of sample faces, taking in respective first sample images in a first sample imaging condition similar to or the same as said first imaging condition; taking in, for said plurality of said sample faces, respective second sample images in a second sample imaging condition similar to or the same as said second imaging condition; calculating a weighted sum of a first covariance matrix calculated from said first sample images and a second covariance matrix calculated from differences between said first sample images and said second sample images; and applying a principal component analysis to said weighted sum to find a result with which principal components are calculated from a given face image; first means for obtaining said first face image; second means for obtaining said second face image; means operative on the basis of a standard formula for calculating a degree of similarity between said first and second face images by using said data means; and means operative on the basis of a comparison of said degree of similarity with a predetermined threshold value for determining whether said first face image matches said second image, wherein said data means comprises eigenvectors and corresponding eigenvalues resulted from solving an eigenvalue problem with respect to said weighted sum, wherein said first and second images (of m×n pixels) are given as vectors x and y, said first face images are given as V 1 , V 2 , . . . , V N (N is the number of sample faces), and said eigenvectors and said corresponding eigenvalues are given as e 1 , e 2 , . . . , e M , and λ 1 , λ 2 , . . . , λ M (M<<m×n), respectively, and wherein said means for calculating the degree of similarity comprises means for calculating, as said degree of similarity, the Maharanobis distance ∑ j = 1 M   ( ( x - V _ , e j ) - ( y - V _ , e j ) 2 ) λ j where V _ = 1 N · ∑ i = 1 N  V i .

Claim 13

A system capable of pattern matching between a first face image having been obtained in a first imaging condition and a second face image having been obtained in a second imaging condition noisier than the first imaging condition, the system comprising: data means stored in a storage for use in extracting principal components from a given face image, said data means having been obtained by: for a plurality of sample faces, taking in respective first sample images in a first sample imaging condition similar to or the same as said first imaging condition; taking in, for said plurality of said sample faces, respective second sample images in a second sample imaging condition similar to or the same as said second imaging condition; calculating a weighted sum of a first covariance matrix calculated from said first sample images and a second covariance matrix calculated from differences between said first sample images and said second sample images; and applying a principal component analysis to said weighted sum to find a result with which principal components are calculated from a given face image; first means for obtaining said first face image; second means for obtaining said second face image; means operative on the basis of a standard formula for calculating a degree of similarity between said first and second face images by using said data means; and means operative on the basis of a comparison of said degree of similarity with a predetermined threshold value for determining whether said first face image matches said second image, wherein said data means comprises eigenvectors and corresponding eigenvalues resulted from solving an eigenvalue problem with respect to said weighted sum, wherein said first and second images (of m×n pixels) are given as vectors x and y, said first face images are given as V 1 , V 2 , . . . , V N (N is the number of sample faces), and said eigenvectors and said corresponding eigenvalues are given as e 1 , e 2 , . . . , e M , and λ 1 , λ 2 , . . . , λ M (M<<m×n), respectively, and wherein said means for calculating the degree of similarity comprises means for calculating, as said degree of similarity, the Maharanobis distance ∑ j = 1 M   ( ( x - V _ , e j ) - ( y - V _ , e j ) 2 ) λ j where V _ = 1 N · ∑ i = 1 N  V i .

Claim 14

A system capable of pattern matching between a first face image having been obtained in a first imaging condition and a second face image having been obtained in a second imaging condition noisier than the first imaging condition, the system comprising: data means stored in a storage for use in extracting principal components from a given face image, said data means having been obtained by: for a plurality of sample faces, taking in respective first sample images in a first sample imaging condition similar to or the same as said first imaging condition; taking in, for said plurality of said sample faces, respective second sample images in a second sample imaging condition similar to or the same as said second imaging condition; calculating a weighted sum of a first covariance matrix calculated from said first sample images and a second covariance matrix calculated from differences between said first sample images and said second sample images; and applying a principal component analysis to said weighted sum to find a result with which principal components are calculated from a given face image; first means for obtaining said first face image; second means for obtaining said second face image; means operative on the basis of a standard formula for calculating a degree of similarity between said first and second face images by using said data means; and means operative on the basis of a comparison of said degree of similarity with a predetermined threshold value for determining whether said first face image matches said second image, wherein said data means comprises eigenvectors and corresponding eigenvalues resulted from solving an eigenvalue problem with respect to said weighted sum, wherein said first and second images (of m×n pixels) are given as vectors x and y, said first face images are given as V 1 , V 2 , V N (N is the number of sample faces), and said eigenvectors and said corresponding eigenvalues are given as e 1 , e 2 , . . . , e M and λ 1 , λ 2 , . . . , λ M (M<<m×n), respectively, and wherein said means for calculating the degree of similarity comprises means for calculating, as said degree of similarity, the correlation between x and y ∑ j = 1 M   ( x - V _ , e j )  ( y - V _ , e j ) / λ j ∑ j = 1 M   ( x - V _ , e j ) 2 / λ j   ∑ j = 1 M   ( y - V _ , e j ) 2 / λ j where V _ = 1 N · ∑ i = 1 N  V i .

Claim 15

A system capable of pattern matching between a first face image having been obtained in a first imaging condition and a second face image having been obtained in a second imaging condition noisier than the first imaging condition, the system comprising: data means stored in a storage for use in extracting principal components from a given face image, said data means having been obtained by: for a plurality of sample faces, taking in respective first sample images in a first sample imaging condition similar to or the same as said first imaging condition; taking in, for said plurality of said sample faces, respective second sample images in a second sample imaging condition similar to or the same as said second imaging condition; calculating a weighted sum of a first covariance matrix calculated from said first sample images and a second covariance matrix calculated from differences between said first sample images and said second sample images; and applying a principal component analysis to said weighted sum to find a result with which principal components are calculated from a given face image; first means for obtaining said first face image; second means for obtaining said second face image; means operative on the basis of a standard formula for calculating a degree of similarity between said first and second face images by using said data means; and means operative on the basis of a comparison of said degree of similarity with a predetermined threshold value for determining whether said first face image matches said second image, wherein said data means comprises eigenvectors and corresponding eigenvalues resulted from solving an eigenvalue problem with respect to said weighted sum, wherein said first and second images (of m×n pixels) are given as vectors x and y, said first face images are given as V 1 , V 2 , V N (N is the number of sample faces), and said eigenvectors and said corresponding eigenvalues are given as e 1 , e 2 , . . . , e M and λ 1 , λ 2 , . . . , λ M (M<<m×n), respectively, and wherein said means for calculating the degree of similarity comprises means for calculating, as said degree of similarity, the correlation between x and y ∑ j = 1 M   ( x - V _ , e j )  ( y - V _ , e j ) / λ j ∑ j = 1 M   ( x - V _ , e j ) 2 / λ j   ∑ j = 1 M   ( y - V _ , e j ) 2 / λ j where V _ = 1 N · ∑ i = 1 N  V i .

Claim 18

A system capable of pattern matching between a first face image having been obtained in a first imaging condition and a second face image having been obtained in a second imaging condition noisier than the first imaging condition, the system comprising: data means stored in a storage for use in extracting principal components from a given face image, said data means having been obtained by: for a plurality of sample faces, taking in respective first sample images in a first sample imaging condition similar to or the same as said first imaging condition; taking in, for said plurality of said sample faces, respective second sample images in a second sample imaging condition similar to or the same as said second imaging condition; calculating a weighted sum of a first covariance matrix calculated from said first sample images and a second covariance matrix calculated from differences between said first sample images and said second sample images; and applying a principal component analysis to said weighted sum to find a result with which principal components are calculated from a given face image; first means for obtaining said first face image; second means for obtaining said second face image; means operative on the basis of a standard formula for calculating a degree of similarity between said first and second face images by using said data means; and means operative on the basis of a comparison of said degree of similarity with a predetermined threshold value for determining whether said first face image matches said second image, wherein said data means comprises a characterizing matrix H which, when applied to a first face image vector x, and a second face image vector y, yields principal component vectors x′ and y′ thereof, respectively, as follows: x′=Hx; and y′=Hy.

Claim 19

A system capable of pattern matching between a first face image having been obtained in a first imaging condition and a second face image having been obtained in a second imaging condition noisier than the first imaging condition, the system comprising: data means stored in a storage for use in extracting principal components from a given face image, said data means having been obtained by: for a plurality of sample faces, taking in respective first sample images in a first sample imaging condition similar to or the same as said first imaging condition; taking in, for said plurality of said sample faces, respective second sample images in a second sample imaging condition similar to or the same as said second imaging condition; calculating a weighted sum of a first covariance matrix calculated from said first sample images and a second covariance matrix calculated from differences between said first sample images and said second sample images; and applying a principal component analysis to said weighted sum to find a result with which principal components are calculated from a given face image; first means for obtaining said first face image; second means for obtaining said second face image; means operative on the basis of a standard formula for calculating a degree of similarity between said first and second face images by using said data means; and means operative on the basis of a comparison of said degree of similarity with a predetermined threshold value for determining whether said first face image matches said second image, wherein said data means comprises a characterizing matrix H which, when applied to a first face image vector x, and a second face image vector y, yields principal component vectors x′ and y′ thereof, respectively, as follows: x′=Hx; and y′=Hy.

Claim 24

A method of determining whether a first face image, having been obtained in a first imaging condition, matches a second face image, having been obtained in a second imaging condition noisier than the first imaging condition, the method comprising: a preparation stage for preparing a data set for use in the determination; and a matching or test stage for carrying out the determination, said preparation stage comprising the steps of: for a plurality of sample faces, obtaining respective first sample images in a first sample imaging condition similar to or the same as said first imaging condition; for said plurality of said sample faces, obtaining respective second sample images in a second sample imaging condition similar to or the same as said second imaging condition; calculating a weighted sum of a first covariance matrix calculated from said first sample images and a second covariance matrix calculated from differences between said first sample images and said second sample images; applying a principal component analysis to said weighted sum to find a result with which principal components are calculated from a given face image; and storing said result as said data set in a storage, and said matching stage comprising the steps of: obtaining said first face image in said first imaging condition; obtaining said second face image in said second imaging condition; calculating a degree of similarity between said first and second face images on the basis of a standard formula by using said data set; and carrying out said determining on the basis of a comparison of said degree of similarity with a predetermined threshold value, wherein one of said step of obtaining said first face image and said step of obtaining said second face image comprises the step of obtaining and adding said face image to a set of reference face images prior to all other steps of said matching stage, and wherein said step of calculating a degree of similarity comprises the step of calculating a degree of similarity between a given face image and each face image of said set.

Claim 25

A method of determining whether a first face image, having been obtained in a first imaging condition, matches a second face image, having been obtained in a second imaging condition noisier than the first imaging condition, the method comprising: a preparation stage for preparing a data set for use in the determination; and a matching or test stage for carrying out the determination, said preparation stage comprising the steps of: for a plurality of sample faces, obtaining respective first sample images in a first sample imaging condition similar to or the same as said first imaging condition; for said plurality of said sample faces, obtaining respective second sample images in a second sample imaging condition similar to or the same as said second imaging condition; calculating a weighted sum of a first covariance matrix calculated from said first sample images and a second covariance matrix calculated from differences between said first sample images and said second sample images; applying a principal component analysis to said weighted sum to find a result with which principal components are calculated from a given face image; and storing said result as said data set in a storage, and said matching stage comprising the steps of: obtaining said first face image in said first imaging condition; obtaining said second face image in said second imaging condition; calculating a degree of similarity between said first and second face images on the basis of a standard formula by using said data set; and carrying out said determining on the basis of a comparison of said degree of similarity with a predetermined threshold value, wherein one of said step of obtaining said first face image and said step of obtaining said second face image comprises the step of obtaining and adding said face image to a set of reference face images prior to all other steps of said matching stage, and wherein said step of calculating a degree of similarity comprises the step of calculating a degree of similarity between a given face image and each face image of said set.

Claim 26

A method of determining whether a first face image, having been obtained in a first imaging condition, matches a second face image having been obtained in a second imaging condition noisier than the first imaging condition, the method comprising: a preparation stage for preparing a data set for use in the determination; and a matching or test stage for carrying out the determination, wherein said preparation stage comprises the steps of: for a plurality (N) of first sample faces, obtaining respective first sample images in a sample imaging condition similar to or the same as said first imaging condition; for said N sample faces, obtaining respective second sample images in a second sample imaging condition similar to or the same as said second imaging condition; for a plurality (L) of second sample faces, obtaining respective third sample images in said sample imaging condition; calculating a weighted sum of a first covariance matrix calculated from said third sample images and a second covariance matrix calculated from differences between said first sample images and said second sample images; applying a principal component analysis to said weighted sum to find a result with which principal components are calculated from a given face image; and storing said result as said data set in a storage, and said matching stage comprising the steps of: obtaining said first face image in said first imaging condition; and obtaining said second face image in said second imaging condition; calculating a degree of similarity between said first and second face images on the basis of a standard formula by using said data set; and carrying out said determining on the basis of a comparison of said degree of similarity with a predetermined threshold value, wherein one of said step of obtaining said first face image and said step of obtaining said second face image comprises the step of obtaining and adding said face image to a set of reference face images prior to all other steps of said matching stage, and wherein said step of calculating a degree of similarity comprises the step of calculating a degree of similarity between a given face image and each face image of said set.

Claim 27

A method of determining whether a first face image, having been obtained in a first imaging condition, matches a second face image having been obtained in a second imaging condition noisier than the first imaging condition, the method comprising: a preparation stage for preparing a data set for use in the determination; and a matching or test stage for carrying out the determination, wherein said preparation stage comprises the steps of: for a plurality (N) of first sample faces, obtaining respective first sample images in a sample imaging condition similar to or the same as said first imaging condition; for said N sample faces, obtaining respective second sample images in a second sample imaging condition similar to or the same as said second imaging condition; for a plurality (L) of second sample faces, obtaining respective third sample images in said sample imaging condition; calculating a weighted sum of a first covariance matrix calculated from said third sample images and a second covariance matrix calculated from differences between said first sample images and said second sample images; applying a principal component analysis to said weighted sum to find a result with which principal components are calculated from a given face image; and storing said result as said data set in a storage, and said matching stage comprising the steps of: obtaining said first face image in said first imaging condition; and obtaining said second face image in said second imaging condition; calculating a degree of similarity between said first and second face images on the basis of a standard formula by using said data set; and carrying out said determining on the basis of a comparison of said degree of similarity with a predetermined threshold value, wherein one of said step of obtaining said first face image and said step of obtaining said second face image comprises the step of obtaining and adding said face image to a set of reference face images prior to all other steps of said matching stage, and wherein said step of calculating a degree of similarity comprises the step of calculating a degree of similarity between a given face image and each face image of said set.

Claim 28

A system capable of pattern matching between a first face image having been obtained in a first imaging condition and a second face image having been obtained in a second imaging condition noisier than the first imaging condition, the system comprising: data means stored in a storage for use in extracting principal components from a given face image, said data means having been obtained by: for a plurality of sample faces, taking in respective first sample images in a first sample imaging condition similar to or the same as said first imaging condition: taking in, for said plurality of said sample faces, respective second sample images in a second sample imaging condition similar to or the same as said second imaging condition; calculating a weighted sum of a first covariance matrix calculated from said first sample images and a second covariance matrix calculated from differences between said first sample images and said second sample images; and applying a principal component analysis to said weighted sum to find a result with which principal components are calculated from a given face image; first means for obtaining said first face image; second means for obtaining said second face image; means operative on the basis of a standard formula for calculating a degree of similarity between said first and second face images by using said data means; and means operative on the basis of a comparison of said degree of similarity with a predetermined threshold value for determining whether said first face image matches said second image, wherein one of said first means and said second means includes storage means that stores a reference face image set database, and wherein said means for calculating a degree of similarity comprises means for calculating a degree of similarity between a given face image and each face image of said reference face image set.

Claim 29

A system capable of pattern matching between a first face image having been obtained in a first imaging condition and a second face image having been obtained in a second imaging condition noisier than the first imaging condition, the system comprising: data means stored in a storage for use in extracting principal components from a given face image, said data means having been obtained by: for a plurality of sample faces, taking in respective first sample images in a first sample imaging condition similar to or the same as said first imaging condition: taking in, for said plurality of said sample faces, respective second sample images in a second sample imaging condition similar to or the same as said second imaging condition; calculating a weighted sum of a first covariance matrix calculated from said first sample images and a second covariance matrix calculated from differences between said first sample images and said second sample images; and applying a principal component analysis to said weighted sum to find a result with which principal components are calculated from a given face image; first means for obtaining said first face image; second means for obtaining said second face image; means operative on the basis of a standard formula for calculating a degree of similarity between said first and second face images by using said data means; and means operative on the basis of a comparison of said degree of similarity with a predetermined threshold value for determining whether said first face image matches said second image, wherein one of said first means and said second means includes storage means that stores a reference face image set database, and wherein said means for calculating a degree of similarity comprises means for calculating a degree of similarity between a given face image and each face image of said reference face image set.

Claim 30

A system capable of pattern matching between a first face image obtained in a first imaging condition and a second face image obtained in a second imaging condition noisier than the first imaging condition, the system comprising: data means stored in a storage for use in extracting principal components from a given face image, said data means having been obtained by: for a plurality (N) of first sample faces, taking in respective first sample images in a sample imaging condition similar to or the same as said first imaging condition; taking in, for said N sample faces, respective second sample images in a second sample imaging condition similar to or the same as said second imaging condition; for a plurality (L) of second sample faces, taking in respective third sample images in said sample imaging condition; calculating a weighted sum of a first covariance matrix calculated from said third sample images and a second covariance matrix calculated from differences between said first sample images and said second sample images; and applying a principal component analysis to said weighted sum; first means for obtaining said first face image; second means for obtaining said second face image; means operative on the basis of a standard formula for calculating a degree of similarity between said first and second face images by using said data means; and means operative on the basis of a comparison of said degree of similarity with a predetermined threshold value for determining whether said first face image matches said second image, wherein one of said first means and said second means includes storage means that stores a set of reference face images, and wherein said means for calculating a degree of similarity comprises means for calculating a degree of similarity between a given face image and each face image of said set.

Claim 31

A system capable of pattern matching between a first face image obtained in a first imaging condition and a second face image obtained in a second imaging condition noisier than the first imaging condition, the system comprising: data means stored in a storage for use in extracting principal components from a given face image, said data means having been obtained by: for a plurality (N) of first sample faces, taking in respective first sample images in a sample imaging condition similar to or the same as said first imaging condition; taking in, for said N sample faces, respective second sample images in a second sample imaging condition similar to or the same as said second imaging condition; for a plurality (L) of second sample faces, taking in respective third sample images in said sample imaging condition; calculating a weighted sum of a first covariance matrix calculated from said third sample images and a second covariance matrix calculated from differences between said first sample images and said second sample images; and applying a principal component analysis to said weighted sum; first means for obtaining said first face image; second means for obtaining said second face image; means operative on the basis of a standard formula for calculating a degree of similarity between said first and second face images by using said data means; and means operative on the basis of a comparison of said degree of similarity with a predetermined threshold value for determining whether said first face image matches said second image, wherein one of said first means and said second means includes storage means that stores a set of reference face images, and wherein said means for calculating a degree of similarity comprises means for calculating a degree of similarity between a given face image and each face image of said set.