PERSON RECOGNITION METHOD AND APPARATUS

• US 20110026782A1
• Filed: 07/28/2010
• Published: 02/03/2011
• Est. Priority Date: 07/29/2009
• Status: Active Grant

• Current Assignee: Fujifilm Corporation (Fujifilm Holdings Corporation)
• Original Assignee: Fujifilm Corporation (Fujifilm Holdings Corporation)
• Inventors: EGO, Shunta
• Granted Patent: US 8,509,497 B2
• Time in Patent Office: Days
• US Class Current: 382/118
• CPC Class Codes: G06F 18/285 Selection of pattern recognition techniques, e.g. of classifiers in a multi-classifier system
• CPC Class Codes: G06V 10/87 using selection of the recognition techniques, e.g. of a classifier in a multiple classifier system
• CPC Class Codes: G06V 40/172 Classification, e.g. identification

Abstract

Subtracting the date of taking a registrant'"'"'s face image from the date of taking a photographic image of a person to be recognized, the number of elapsed years Y of the registrant'"'"'s face image is calculated. If Y<5, an unsophisticated image comparator is selected. If 5≦Y<10, a first face parts comparator is selected to calculate the degree of resemblance between the photographic image and the registrant'"'"'s face image more precisely than the unsophisticated image comparator. If 10≦Y, a second face parts comparator is selected. If the face image is a child'"'"'s, the second face parts comparator compares the photographic image not only with the child'"'"'s face image but also with his/her cognate'"'"'s face image. After the photographic image is compared with every registrant'"'"'s face image, the personal data of the top four registrants having the highest degree of resemblance is displayed on an LCD.

Claims

1. A person recognition method for comparing a photographic image showing a face of a person to be recognized with face images of a plurality of registrants registered in a database, the person recognition method comprising the steps of:

subtracting a date of taking the face image from a date of taking the photographic image, and calculating a number of elapsed years from the date of taking the face image;

selecting one of a plurality of image comparison methods for comparing the photographic image with the face image in accordance with the number of elapsed years;

comparing the photographic image with the face image by the selected image comparison method, and calculating a degree of resemblance between the photographic image and the face image; and

displaying personal data of the one or more registrants, the personal data with the higher degree of resemblance being given a high priority to be displayed.

2. The person recognition method according to claim 1, wherein the photographic image is displayed on a screen of a display together with the personal data of at least one of the registrants.

3. The person recognition method according to claim 1, wherein the registrant'"'"'s personal data includes the registrant'"'"'s face image, a registrant'"'"'s name, and a registrant'"'"'s age at the time of taking the face image.

4. The person recognition method according to claim 1, wherein if a registrant'"'"'s cognate is already registered, the photographic image is compared with a cognate'"'"'s face image, in addition to the registrant'"'"'s face image.

5. The person recognition method according to claim 1, wherein one of the plurality of image comparison methods is an unsophisticated image comparison method being selected in the selecting step if the number of elapsed years is less than a first threshold value, the unsophisticated image comparison method comprising the steps of:

dividing each of the photographic image and the face image into small areas;

calculating an average color of each of the individual small areas;

comparing the average colors of the corresponding small areas between the photographic image and the face image; and

calculating the degree of resemblance between the photographic image and the face image in accordance with a number of the small areas having the similar average colors.

6. The person recognition method according to claim 5, wherein another one of the plurality of image comparison methods is a first face parts comparison method being selected in the selecting step if the number of elapsed years is equal to or more than the first threshold value and less than a second threshold value, the first face parts comparison method comprising the steps of:

extracting face parts from each of the photographic image and the face image; and

comparing at least one of a shape, a color, and a relative position of the face parts between the photographic image and the face image.

7. The person recognition method according to claim 6, wherein further another one of the plurality of image comparison methods is a second face parts comparison method being selected in the selecting step if the number of elapsed years is equal to or more than the second threshold value, the second face parts comparison method comprising the steps of:

checking whether or not the personal data of a registrant'"'"'s cognate is accessorily registered; and

if the cognate'"'"'s personal data is accessorily registered, comparing the photographic image with a cognate'"'"'s face image, in addition to the registrant'"'"'s face image.

8. The person recognition method according to claim 1, further comprising the steps of:

entering an age of the person to be recognized;

calculating a registrant'"'"'s current age;

judging whether or not a difference between the age of the person to be recognized and the registrant'"'"'s current age is within a predetermined allowance value; and

if the difference is out of the predetermined allowance value, excluding the registrant'"'"'s personal data from comparison.

9. The person recognition method according to claim 8, wherein the registrant'"'"'s current age is calculated from a registrant'"'"'s birthday contained in the registrant'"'"'s personal data and the current date.

10. The person recognition method according to claim 4, wherein the degree of resemblance is weighted in accordance with the number of elapsed years, so that the fewer the number of elapsed years, the higher the degree of resemblance becomes.

11. The person recognition method according to claim 4, wherein the degree of resemblance is weighted in accordance with a degree of consanguinity of the cognate, so that the closer the degree of consanguinity, the higher the degree of resemblance becomes.

12. A person recognition apparatus for comparing a photographic image showing a face of a person to be recognized with face images of a plurality of registrants registered in a database, and displaying personal data of the registrant having the face image resembling the photographic image, the person recognition apparatus comprising:

an elapsed year calculator for subtracting a date of taking the face image from a date of taking the photographic image, and calculating a number of elapsed years from the date of taking the face image;

a comparison method selector for selecting one of a plurality of image comparison methods for comparing the photographic image with the face image in accordance with the number of elapsed years;

an image comparator for comparing the photographic image with the face image by the selected image comparison method, and calculating a degree of resemblance between the photographic image and the face image; and

a display for displaying the personal data of the one or more registrants in descending order of the degree of resemblance.

13. The person recognition apparatus according to claim 12, being provided in a digital camera.