Specifying position of characteristic portion of face image

1. An image processing apparatus that specifies a position of a predetermined characteristic portion of a target face image, the image processing apparatus comprising:

an initial disposition unit that determines an initial disposition of a characteristic point of the target face image based on a disposition of the characteristic point of a reference face image that is set based on a statistical analysis on a plurality of sample face images of which dispositions of the characteristic points representing the positions of the characteristic portions are known;

an image transforming unit that performs a first transformation for at least one of the reference face image and the target face image, so that disposition patterns of the characteristic points of the reference face image and the target face image are identical to each other;

a normalization unit that performs a first normalization process, in which predetermined first index values that represent distributions of pixel values of the reference face image and the target face image approach each other, for at least one of the reference face image and the target face image after the first transformation;

a determination unit that compares the reference face image with the target face image after the first normalization process and determines whether to update the disposition of the characteristic point of the target face image based on a result of the comparison; and

an update unit that updates the disposition of the characteristic point of the target face image based on the result of the comparison between the reference face image and the target face image in a case where the update is determined to be performed.

2. The image processing apparatus according to claim 1 , wherein the determination unit calculates a predetermined second index value that represents the degree of difference between the reference face image and the target face image based on the result of comparing the reference face image with the target face image and determines whether to perform the update based on a result of comparing the second index value with a threshold value.

3. The image processing apparatus according to claim 2 , wherein the normalization unit performs a second normalization process, in which the first index values of the reference face image and the target face image approach each other, for at least one of the reference face image and the target face image in a reference correspondence relationship, which is a correspondence relationship between the reference face image and the target face image, that becomes a reference and transformed correspondence relationships of N types that are correspondence relationships between the reference face image and the target face image at a time when the second transformations of N (here, N is an integer equal to or greater than one) types are performed for at least one of the reference face image and the target face image with reference to the reference correspondence relationship, and

wherein the initial disposition unit selects the correspondence relationship, for which the degree of difference between the reference face image and the target face image after the second normalization process is the smallest, from among the reference correspondence relationship and the transformed correspondence relationships and determines the initial disposition of the characteristic point of the target face image based on the disposition of the characteristic point of the reference face image in the selected correspondence relationship.

4. The image processing apparatus according to claim 3 , wherein the second transformations of N types are transformations in which at least one of parallel movement, a change in tilt, and enlargement or reduction of the entire characteristic point is performed.

5. The image processing apparatus according to claim 4 , wherein the first index value includes at least one of an average value and a variance value of pixel values.

6. The image processing apparatus according to claim 5 , wherein the reference face image is an average image of the plurality of sample face images that has been transformed such that the disposition of the characteristic point is identical to an average shape that represents an average position of the characteristic points of the plurality of sample face images.

7. The image processing apparatus according to claim 6 , further comprising: a face area detecting unit that detects an image area, which includes at least a part of a face image, from the target face image as a face area,

wherein the initial disposition unit that determines the initial disposition of the characteristic point in the target face image based on at least one of the position, the size, and the tilt of the face area of the target face image.

8. The image processing apparatus according to claim 7 , wherein the first transformation is an affine transformation for each polygonal area having the characteristic points set in the reference face image and the target face image as apexes thereof.

9. The image processing apparatus according to claim 8 , further comprising: a memory unit storing model information that is used for specifying the disposition model of the characteristic point that is a model of the characteristic point set based on the statistical analysis and is acquired by a sum of an average shape that represents an average position of the characteristic points of the plurality of sample face images and a linear combination of shape vectors representing the characteristics of the disposition of the characteristic points of the plurality of sample face images therein,

wherein the update unit updates the disposition of the characteristic point of the target face image by changing a coefficient of at least one of the shape vectors in the disposition model that represents the disposition of the characteristic point of the target face image based on the result of comparing the reference face image with the target face image.

10. An image processing method of specifying a position of a predetermined characteristic portion of a target face image, using a computer comprising:

determining an initial disposition of a characteristic point of the target face image based on a disposition of the characteristic point of a reference face image that is set based on a statistical analysis on a plurality of sample face images of which dispositions of the characteristic points representing the positions of the characteristic portions are known;

performing a first transformation for at least one of the reference face image and the target face image, so that disposition patterns of the characteristic points of the reference face image and the target face image are identical to each other;

performing a first normalization process, in which predetermined first index values that represent distributions of pixel values of the reference face image and the target face image approach each other, for at least one of the reference face image and the target face image after the first transformation;

comparing the reference face image with the target face image after the first normalization process and determining whether to update the disposition of the characteristic point of the target face image based on a result of the comparison; and

updating the disposition of the characteristic point of the target face image based on the result of the comparison between the reference face image and the target face image in a case where the update is determined to be performed.

11. A non-transitory computer readable medium storing a computer program for specifying a position of a predetermined characteristic portion of a target face image, the computer program implements, in a computer, functions comprising:

a function for determining an initial disposition of a characteristic point of the target face image based on a disposition of the characteristic point of a reference face image that is set based on a statistical analysis on a plurality of sample face images of which dispositions of the characteristic points representing the positions of the characteristic portions are known;

a function for performing a first transformation for at least one of the reference face image and the target face image, so that disposition patterns of the characteristic points of the reference face image and the target face image are identical to each other;

a function for performing a first normalization process, in which predetermined first index values that represent distributions of pixel values of the reference face image and the target face image approach each other, for at least one of the reference face image and the target face image after the first transformation;

a function for comparing the reference face image with the target face image after the first normalization process and determining whether to update the disposition of the characteristic point of the target face image based on a result of the comparison; and

a function for updating the disposition of the characteristic point of the target face image based on the result of the comparison between the reference face image and the target face image in a case where the update is determined to be performed.