Face authentication system

Patent metadata

Field	Value
Publication number	US-9922237-B2
Application number	US-201415117795-A
Country	US
Kind code	B2
Filing date	Feb 26, 2014
Priority date	Feb 26, 2014
Publication date	Mar 20, 2018
Grant date	Mar 20, 2018

Abstract

A face authentication system of an example of the present invention includes: a first face area detection unit configured to detect a first face image from a first captured image captured by a camera; a first deformation unit configured to deform the first face image non-linearly using a template face image; a face image recording unit configured to record the deformed first face image as a registered face image; a second face area detection unit configured to detect a second face image from a second captured image captured by a camera; a second deformation unit configured to deform the second face image non-linearly using the template face image; and a face image matching unit configured to compare the deformed second face image with the registered face image.

First claim

What is claimed is:

1. A face authentication system comprising:
a first face area detector for detecting a first face image from a first captured image;
a first deformation processor for deforming the first face image non-linearly using a template face image;
a face image recorder for recording the deformed first face image as a registered face image, wherein the face image recording unit is configured to record a plurality of deformed first face images obtained from a plurality of first captured images of a same person as registered face images;
a second face area detector for detecting a second face image from a second captured image;
a second deformation processor for deforming the second face image non-linearly using the template face image; and
a face image matching processor for comparing the deformed second face image with at least one of the registered face images.

2. The face authentication system according to claim 1 , wherein the first deformation processor is configured to:
extract sampling points from the template face image;
extract sampling points corresponding to the sampling points in the template face image from the deformed first face image;
calculate similarity between the template face image and the deformed first face image using the sampling points extracted from the template face image and the sampling points extracted from the first face image; and
re-deform the deformed first face image based on the similarity.

3. The face authentication system according to claim 2 , wherein the first deformation processor is configured to:
set control points to the first face image in order to deform the first face image; and
deform the first face image by shifting the control points in the first face image.

4. The face authentication system according to claim 1 , wherein the second deformation processor is configured to:
extract sampling points corresponding to sampling points in the template face image from the deformed second face image;
calculate similarity between the template face image and the deformed second face image using the sampling points extracted from the template face image and the sampling points extracted from the second face image; and
re-deform the deformed second face image based on the similarity.

5. The face authentication system according to claim 4 , wherein the first deformation processor and the second deformation processor are configured to:
separate a face area of the template face image into a plurality of sub-areas; and
determine a number of sampling points to be extracted from each of the plurality of sub-areas in accordance with a weight correspondingly assigned to each of the plurality of sub-areas.

6. The face authentication system according to claim 1 , wherein the template face image is a face image created by synthesizing a plurality of different face images.

7. The face authentication system according to claim 1 ,
wherein the second deformation processor is configured to deform the second face image deformed using the template face image non-linearly using the registered face image, and
wherein the face image matching processor is configured to compare the second face image deformed using the registered face image with the registered face image.

8. A face authentication method comprising:
detecting a first face image from each first captured image of a plurality of first captured images;
deforming, for each of a plurality of first face images, the first face image non-linearly using a template face image;
recording, for each of a plurality of deformed first face images obtained from each of a plurality of first captured images of a same person as registered face images, the deformed first face image as a registered face image;
detecting a second face image from a second captured image;
deforming the second face image non-linearly using the template face image; and
comparing the deformed second face image with the registered face image.

9. The face authentication method according to claim 8 , wherein the deforming the first face image comprises:
extracting sampling points from the template face image;
extracting sampling points corresponding to the sampling points in the template face image from the deformed first face image;
calculating similarity between the template face image and the deformed first face image using the sampling points extracted from the template face image and the sampling points extracted from the first face image; and
re-deforming the deformed first face image based on the similarity, and
wherein the deforming the second face image comprises:
extracting sampling points from the template face image;
extracting sampling points corresponding to sampling points in the template face image from the deformed second face image;
calculating similarity between the template face image and the deformed second face image using the sampling points extracted from the template face image and the sampling points extracted from the second face image; and
re-deforming the deformed second face image based on the similarity.

10. The face authentication method according to claim 9 , wherein the deforming the first face image comprises:
setting control points to the first face image in order to deform the first face image; and
deforming the first face image by shifting the control points in the first face image, and
wherein the deforming the second face image comprises:
setting control points to the second face image in order to deform the second face image; and
deforming the second face image by shifting the control points in the second face image.

11. The face authentication method according to claim 9 , wherein each of deforming the first face image and deforming the second face image comprises:
separating a face area of the template face image into a plurality of sub-areas; and
determining a number of sampling points to be extracted from each of the sub-areas in accordance with a weight correspondingly assigned to each of the plurality of sub-areas.

12. The face authentication method according to claim 8 ,
wherein the step of deforming the second face image includes deforming the second face image deformed using the template face image non-linearly using the registered face image, and
wherein the step of comparing includes comparing the second face image deformed using the registered face image with the registered face image.

Assignees

• Hitachi Ltd

Inventors

• Li Zisheng
• Kurihara Tsuneya
• Kiyomizu Harumi
• Kagehiro Tatsuhiko

Classifications

• G06V40/161

  Detection; Localisation; Normalisation · CPC title

• G06V40/172 Primary

  Classification, e.g. identification · CPC title

• G07C9/37

  using biometric data, e.g. fingerprints, iris scans or voice recognition · CPC title

• G07C9/00158

  Physics · mapped topic

• G06K9/00288

  Physics · mapped topic