Moving object recognition apparatus for tracking a moving object based on photographed image

1. A moving object recognition apparatus for recognizing and tracking a person based on moving images photographed by a camera, a first frame of the moving images being a preceding frame of a second frame of the moving images, the moving object recognition apparatus comprising:

a feature image extraction unit that extracts facial feature image portions for enabling the person to be determined by face recognition from the moving images;

a position detection unit that detects position information of the person in time-series from the moving images, each time-series position information indicating a position of the person in a respective moving image;

a determination unit, using a processor, that makes a comparison between the facial feature image portion in the first frame and the facial feature image portion in the second frame, and determines correspondence therebetween, the facial feature image portion in the first frame being associated with an identity of the person;

a search section, using a processor, that, when the determination unit determines that the facial feature image portions in the first frame and the second frame do not have the correspondence therebetween, searches a database for finding another facial feature image portion stored in the database, said another facial feature image portion having correspondence with the facial feature image portion in the second frame; and

a track information retention unit, using a processor, that retains the time-series position information in the form of coordinate data in the moving images linked with an identifier that identifies the person having the facial feature image portions, which are determined to have the identity, wherein the identifier is a face image portion of the identified person associated with each time-series position information to retain a position information history associated with the identified person,

wherein the track information retention unit includes an identifier applying section and the search section, and

when the searching by the search section fails, the determination unit concludes that the extracted facial feature image portion in the second frame is for a new person, and the identifier applying section applies to the new person a new identifier with which the detected time-series position information is to be associated.

2. The moving object recognition apparatus as claimed in claim 1 ,

wherein the track information retention unit includes a search section that searches information retained in the track information retention unit for the identifier of the person with which the detected time-series position information is to be associated.

3. The moving object recognition apparatus as claimed in claim 1 , wherein the track information retention unit manages the extracted facial feature image portions in association with the identifier of the person from which the facial feature image portions are extracted.

4. The moving object recognition apparatus as claimed in claim 1 , wherein the feature image extraction unit tracks an image portion where motion exists between frames as a person image portion from the moving images and extracts the facial feature image portion from the person image portion.

5. The moving object recognition apparatus as claimed in claim 1 ,

wherein the feature image extraction unit extracts facial feature image portions for enabling the person to be determined from a plurality of moving images photographed by a plurality of cameras, and

the moving object recognition apparatus includes an integration unit that integrates position information of the feature image portions detected by the position detection unit in the same coordinate system.

6. The moving object recognition apparatus as claimed in claim 1 , wherein the facial feature image portions are image portions of a skin color in the moving image.

7. A computer-readable medium having recorded thereon a program for causing a computer system to execute procedures for retaining a person photographed in moving images, a first frame of the moving images being a preceding frame of a second frame of the moving images, and time-series position information of the person in association with each other, the procedures comprising:

extracting facial feature image portions for enabling the person to be determined from the moving images;

detecting position information of the person in time series from the moving images, each time-series position information indicating a position of the person to a respective moving image;

making a comparison between the facial feature image portion in the first frame and the facial feature image portion in the second frame and determining a correspondence therebetween, the facial feature image portion in the first frame being associated with an identity of the person;

searching a database for finding another facial feature image portion stored in the database, when the facial feature image portions in the first frame and the second frame do not have correspondence therebetween, said another facial feature image portion having correspondence with the facial feature image portion in the second frame; and

retaining the time-series position information in the form of coordinate data in the moving images linked with an identifier that identifies the person having the facial feature image portions, which are determined to have the identity, wherein the identifier is a face image portion of the identified person associated with each time-series position information to retain a position information history associated with the identified person,

wherein when the searching fails, determining that the extracted facial feature image portion in the second frame is for a new person, and a new identifier is applied to the new person with which the detected time-series position information is to be associated.

8. A computer-implemented moving object recognition method for retaining a person photographed in moving images, a first frame of the moving images being a preceding frame of a second frame of the moving images, and time-series position information of the person in association with each other, the method comprising:

extracting facial feature image portions for enabling the person to be determined from the moving images;

detecting, using a processor, position information of the person in time series from the moving images, each time-series position information indicating a position of the person in a respective moving image;

making a comparison, using the processor, between the facial feature image portion in the first frame and the facial feature image portion in the second frame and determining a correspondence therebetween, the facial feature image portion in the first frame being associated with an identity of the person;

searching a database, using the processor, for finding another facial feature image portion stored in the database, when the facial feature image portions in the first frame and the second frame do not have correspondence therebetween, said another facial feature image portion having correspondence with the facial feature image portion in the second frame;

retaining the time-series position information in a computer-readable storage medium in association with an identifier that identifies the person having the facial feature image portions, which are determined to have the identity, wherein the identifier is a face image portion of the identified person associated with each time-series position information to retain a position information history associated with the identified person,

wherein when the searching fails, determining that the extracted facial feature image portion in the second frame is for a new person, and a new identifier is applied to the new person with which the detected time-series position information is to be associated.