JP2009048234A - System and method for face recognition - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a face recognition system and a face recognition method having a high recognition rate, by generating an image constituted of only data representing the features of a face after deleting an image of a background other than the face from an image obtained by imaging the face, and by performing a variety of recognition process using the image constituted of only the data representing the features of the face, in the face recognition system for recognizing and discriminating the face. <P>SOLUTION: The face recognition system includes: a face imaging section for imaging the face; a face part cutout section for discriminating and cutting out major parts of the face from the image obtained by the face imaging section; a face part rearranged image generation section for rearranging the face parts cut out by the face part cutout section, on a blank work area and generating the image; and a recognition section for recognizing the image in which the face parts are disposed by the face part rearranged image generation section. Thus, by properly deleting the background from the image, the face recognition system shows efficient recognition process with high accuracy. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a face recognition system and method for recognizing a face.

  In recent years, in a building entrance / exit management system, biometrics authentication systems such as personal authentication based on fingerprints and personal identification based on passwords have been widely used to determine whether or not a person should be permitted to enter a building. If this biometrics authentication system is further developed, the entrance / exit management system will be more convenient if the entrance / exit management system can authenticate and discriminate people by face, as in the case where humans authenticate and discriminate others. improves. In addition, if a technique for authenticating and discriminating an individual by face is applied to a robot, the function of the robot is dramatically improved, and it is possible to allow the robot to perform reception work or to realize an advanced healing robot.

  In a face authentication system for authenticating and discriminating human faces, facial features are extracted from the human face that serves as the basis for discrimination, and the extracted features are individually stored in a multidimensional feature space having various features as elements. As a characteristic region unique to the face, it is determined whether or not it is a specific face by comparing the characteristic regions.

  In such a face authentication system, the features of a person's face are grasped as the shape of the facial parts themselves, such as the eyes, nose, mouth, and facial contours, and the arrangement information between these parts as the features unique to the main face. To do.

  However, when a person's face is imaged in order to grasp the features of the face, a background other than the face is always included in the image. The background is not data unique to a person, and changes variously depending on the imaging conditions when the face is imaged. This is an obstacle to extracting features unique to the face and performing recognition processing. This will be described with reference to the drawings.

FIG. 3 is an image of a face taken with a CCD camera.
In FIG. 3, 300 is an entire image, 301 is an image of a background curtain, 303 is an image of a background window glass, and 302 is hair.
In the case of this image, a window and a curtain are shown as a background. In the image showing the face in this way, the boundary between the outline of the face and the background can be accurately grasped due to the presence of a part of the face that does not maintain a certain shape, such as the background or hair. This is a challenge in the face recognition technology field.

In order to solve this problem, in the past, the boundary between the background and the face was grasped as much as possible, and the background was simply made non-patterned or the face was masked to cut out only the central part of the face. .
JP-A-4-277873

  However, the background changes unexpectedly depending on the imaging conditions, and the face outline changes variously depending on the hairstyle etc., so there is a limit to accurately capturing the boundary between the face and the background. The method did not completely eliminate the background.

  Therefore, the present invention is a face recognition system for recognizing and discriminating a face, by accurately excluding the background from the image obtained by capturing the face, and accurately acquiring only a characteristic image related to the human face. An object of the present invention is to provide an efficient and high recognition rate face recognition system and face recognition method that are not adversely affected.

The invention according to claim 1 is a face recognition system for recognizing a face,
A face imaging unit for imaging a face;
A background deletion unit that deletes a background from the image captured by the face imaging unit;
A face part cutout unit that discriminates and cuts out major parts of the face (hereinafter also referred to as “face parts”) from an image (hereinafter also referred to as “face image”) from which the background has been deleted by the background deletion unit;
A face part information acquisition unit for acquiring information on the shape and position of each face part (hereinafter also referred to as “face part information”);
Each face part cut out by the face part cutout unit is rearranged in a plain work area based on the face part information, and the face part is rearranged in the plain work area (hereinafter referred to as “re-arrangement”). A face part rearrangement image creation unit for creating a layout face image),
A recognition processing unit that performs recognition processing on the rearranged face image created by the face part rearranged image creation unit,
A recognition process is performed by deleting an image related to the background from the captured image.

The invention according to claim 2 is a face recognition method for recognizing a face,
A face imaging process for imaging a face;
A background deletion step of deleting a background from the image of the face imaged by the face imaging step;
From the face image from which the background has been deleted by the background deletion step, the face part of the face is determined and cut out, a face part cutout step,
A facial part information acquisition step of acquiring the facial part information relating to the shape and position of each facial part;
The rearranged face image obtained by rearranging the face parts cut out by the face part cutout process in a plain work area based on the face part information and rearranging the face parts in the plain work area. A face part rearrangement image creation process to create,
A recognition processing step of performing recognition processing on the rearranged face image created by the face part rearrangement image creation step,
A recognition process is performed by deleting an image related to the background from the captured image.

  According to the face recognition system of the present invention, it is possible to create an image in which the background other than the face is accurately excluded from the image captured by the imaging unit. As a result, in subsequent image processing, various recognition processes can be performed on an image that does not include a background, and an efficient face recognition system with a high recognition rate can be realized.

  Hereinafter, the best mode for carrying out the invention will be described with reference to the drawings.

(Embodiment 1)
FIG. 1 is a block diagram showing a configuration of a face recognition system according to the present invention.
In FIG. 1, 101 is an imaging unit, 102 is a background exclusion unit, 103 is a face part cutout unit, 104 is a face part information acquisition unit, 105 is a face part rearranged image creation unit, Reference numeral 106 denotes a recognition processing unit. Reference numeral 107 denotes a personal computer that constitutes the background exclusion unit 102, the facial part cutout unit 103, the facial part information acquisition unit 104, the facial part rearranged image creation unit 105, and the recognition processing unit 106.

  Next, each part will be described.

  The imaging unit 101 is a device that captures a face, and is configured by, for example, a CCD camera. However, you may comprise with apparatuses other than a CCD camera.

  The background exclusion unit 102 excludes the background from the captured image and leaves only the face image. This will be described with reference to the drawings.

FIG. 3 is a face image captured by the image capturing unit 101.
In FIG. 3, 300 is an entire image, 301 is an image of a background curtain, 303 is an image of a background window glass, and 302 is hair.
FIG. 4 is an image of the face after removing the background.
The background exclusion unit 102 deletes the background such as a window or curtain that is reflected behind the face. In this case, it is difficult to accurately grasp the boundary between the face and the background, and in order to avoid the remaining of the background, the background remains by cutting out the periphery of the face as in the image shown in FIG. There are many cases to avoid.

  The face part cutout unit 103 searches for a face part that is a main part of the face from the image from which the background is excluded, and cuts out the face part.

Next, the face part information acquisition unit 104 acquires the shape and position of each face part cut out by the face part cutout unit 103.
FIG. 5 is an explanatory diagram for explaining information on the shape and position of a face part.
In FIG. 5, reference numerals 501 and 502 are coordinate axes (X axis, Y axis) for indicating information on the shape and position of the face part, 502 is the X axis, and 501 is the Y axis. In FIG. 5, the intersection of a horizontal line connecting both eyes horizontally and a vertical line that cuts the center of the face (nose) is the origin (0, 0) of the XY coordinates. A is the right edge of the lips and B is the left edge of the lips. a1 is the absolute value of the Y coordinate of position A, a2 is the absolute value of the X coordinate of position A, b1 is the absolute value of the Y coordinate of position B, and b2 is the absolute value of the X coordinate of position B is there.

The face part information acquisition unit 104 grasps the shape and position of the face part, digitizes the shape and position of each face part using an appropriate coordinate axis, and stores it as face part information.
For example, the right end position A of the lips is (a2, -a1) in XY coordinates, and the left end position B of the lips is (-b2, -b1) in XY coordinates. Similarly, with respect to other face parts, the position of each face part is grasped using XY coordinates, and face part information is acquired. Further, if information on the position of each part of the face part is acquired, information on the shape of the face part is also acquired.

The face part rearranged image creation unit 105 finds information about the position of each face part acquired by the face part information acquisition unit 104 in a blank work area that has no face parts searched out by the face parts cutout unit 103. Relocate based on
FIG. 6 is an explanatory diagram for explaining an image in which face parts are rearranged.
In FIG. 6, reference numeral 601 denotes a rearranged face image in which face parts are rearranged in an empty work area. 602 is a face part related to the nose, which is one of the face parts, 603 is a face part related to the eyes, which is one of the face parts, and 604 is a face part related to the mouth, which is one of the face parts. . Note that reference numerals 602 to 604 are examples of face parts, and are not limited to these.
The rearranged face image 601 includes information on the shape of each face part and information on the shape and position between the face parts. This information is not noise data such as the background but data unique to the face. It is data composed only of

  The recognition processing unit 106 performs various image processes for recognizing the face on the rearranged face image 601. Since the rearranged face image 601 is obtained by rearranging the face parts in the empty work area, information regarding the background is not included. For this reason, the recognition processing unit 106 processes the rearranged face image 601, thereby realizing high-accuracy image processing without being disturbed by data such as a background.

Next, the face recognition method according to the present invention will be described.
FIG. 2 is a flowchart showing an outline of the face recognition method according to the present invention.
In FIG. 2, S200 is an imaging process, S201 is a background removal process, S202 is a face part cutout process, S203 is a face part information acquisition process, and S204 is a face part rearrangement face image creation process. , S205 is a recognition processing step.

  First, the target face is imaged with a CCD camera or the like to obtain an image of the face (S200).

  Next, the background is excluded from the image showing the face (S201). For example, a mask is applied to the center of the face, and the background is removed by deleting portions other than the masked portion.

  Then, main parts such as eyes, nose and mouth constituting the face are extracted from the face image from which the background is excluded and cut out (S202).

  Next, the position of each extracted face part is grasped and acquired as face part information (S203). Specifically, for example, as shown in FIG. 5, information on the shape and position of each face part is grasped based on XY coordinates with the origin as the origin. However, the grasping method is not limited to the method using the XY coordinates, and any other method may be used as long as the shape and position of each face part can be grasped.

  Next, the face parts cut out in the blank blank work area are rearranged based on the face part information (S204). As a result, a rearranged face image 601 (see FIG. 6) that does not include information irrelevant to the face such as the background can be obtained.

Next, various image processing for recognizing the face is performed based on the rearranged face image 601 (S205).
Since the rearranged face image 601 is composed only of data unique to the face, it is possible to realize an efficient and highly accurate recognition process that is not disturbed by noise such as the background.

It is a block diagram which shows the structure of a face recognition system. It is a flowchart which shows the outline of a face recognition method. It is the image of the face imaged with the CCD camera. It is an image of the face after removing the background. It is explanatory drawing for demonstrating the facial part information regarding a facial part. It is explanatory drawing for demonstrating the image by which the face part was rearranged.

Explanation of symbols

DESCRIPTION OF SYMBOLS 101 Image pick-up part 102 Background exclusion part 103 Face part clipping part 104 Face part information acquisition 105 Face part rearrangement image creation part 106 Recognition processing part 107 Background exclusion part 102, Face part clipping part 103, Face part information acquisition part 104, Face parts Personal computer constituting rearranged image creation unit 105 and recognition processing unit 106

Claims (2)

A face recognition system for recognizing a face,
A face imaging unit for imaging a face;
A background deletion unit that deletes a background from the face image captured by the face imaging unit;
A face part cutout unit that discriminates and cuts out major parts of the face (hereinafter also referred to as “face parts”) from an image (hereinafter also referred to as “face image”) from which the background has been deleted by the background deletion unit;
A face part information acquisition unit for acquiring information on the shape and position of each face part (hereinafter also referred to as “face part information”);
The face parts cut out by the face part cutout unit are rearranged in a plain work area based on the face part information, and the face parts are rearranged in the plain work area (hereinafter, “ A face part rearrangement image creation unit for creating a rearrangement face image),
A recognition processing unit that performs recognition processing on the rearranged face image created by the face part rearranged image creation unit,
A face recognition system, wherein an image related to a background is deleted from a captured image and recognition processing is performed. A face recognition method for recognizing a face,
A face imaging process for imaging a face;
A background deletion step of deleting a background from the image of the face imaged by the face imaging step;
From the face image from which the background has been deleted by the background deletion step, the face part of the face is determined and cut out, a face part cutout step,
A facial part information acquisition step of acquiring the facial part information relating to the shape and position of each facial part;
The rearranged face image obtained by rearranging the face parts cut out by the face part cutout process in a plain work area based on the face part information and rearranging the face parts in the plain work area. A face part rearrangement image creation process to create,
A recognition processing step of performing recognition processing on the rearranged face image created by the face part rearrangement image creation step,
A face recognition method comprising performing recognition processing by deleting an image related to a background from a captured image.

Images