JP2007199766A - Face image recording device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a face image recording device for easily recording the face image of a human body passing a specific position. <P>SOLUTION: A stereo-camera 14 photographs a region E1 where a person passes from the upper part, and a front camera 16 photographs a region E2 as an arbitrary portion of the region E1 from the direction of movement of the person. Therefore, a three-dimensional image looking down at the whole region E1 and an image catching a portion of the region E1 from the front are stored in an RAM 20. A CPU 26 detects a vertex Ht of an object based on the three-dimensional image, and calculates the moving speed of a detection result, and judges whether or not height conditions and moving speed conditions are respectively satisfied by the detection result and the calculation result. When it is judged that the both height conditions and moving speed conditions are satisfied, the face position Hc of the human body is estimated from the vertex Ht, and the attitude of the front camera 16 is controlled so that the region E2 including the face position Hc can be photographed. Then, the image photographed by the front camera 16 after control processing is recorded in a hard disk 24d. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a face image recording apparatus, and more particularly to a face image recording apparatus that photographs a human body passing through a gate and records the face image.

  In general, this type of face image recording apparatus repeatedly captures the vicinity of a gate, detects an area where a face is likely to exist from the obtained image, and records the image in the detected area as a face image. . However, since it is uncertain whether or not there is a person near the gate, an area where a face is likely to exist must be detected efficiently and accurately from the image.

As a conventional technique for detecting a region where a face exists from an image obtained by photographing an object scene in which the existence of a human body is indeterminate, a technique disclosed in Patent Document 1 is known. This conventional technique divides an image into a plurality of blocks, compresses them, calculates a feature vector indicating the characteristics of the human body for each block, and detects a region where a face is likely to exist based on the calculation result It is.
JP-A-2005-190400

    However, even if this conventional technology is applied to a face image recording apparatus, it is necessary to perform feature calculation (pattern matching) over the entire image, but every time it is taken, even though it is compressed, so a huge amount of calculation is required. Is required.

  Therefore, a main object of the present invention is to provide a face image recording apparatus capable of easily recording a face image of a human body passing through a specific position such as a gate.

  The face image recording apparatus according to the first aspect of the present invention is a face image recording apparatus for recording a face image of a human body passing through a specific area (E1) in a specific direction (y), wherein the specific area (E1) is the specific area. A first photographing means (14) for photographing three-dimensionally from above, a second photographing means (16) for photographing an arbitrary partial area (E2) which is a part of the specific area from a specific direction, and a first photographing means. Vertex detection means (S47) for detecting the vertex (Ht) of the subject in the specific area based on the photographed three-dimensional image, and first to determine whether or not the detection result of the vertex detection means satisfies the height condition When the discrimination means (S9) and the first discrimination means determine that the height condition is satisfied, the estimation means (S11) for estimating the face position (Hc) of the human body from the detection result of the vertex detection means is estimated by the estimation means. Control means (S13) for controlling the posture of the second photographing means so as to photograph the partial region including the face position; and Recording means for recording images captured by the second imaging means after the control process of the control means comprises a (S29).

  In the first aspect of the invention, the specific area is imaged three-dimensionally from above the specific area by the first imaging means. An arbitrary partial area that is a part of the specific area is imaged from the specific direction by the second imaging means. The vertex detecting means detects the vertex of the subject in the specific area based on the three-dimensional image photographed by the first photographing means. Whether the detection result of the vertex detection means satisfies the height condition is determined by the first determination means. When it is determined by the first determining means that the height condition is satisfied, the estimating means estimates the face position of the human body from the detection result of the vertex detecting means, and the control means determines a partial region including the face position estimated by the estimating means. The posture of the second photographing means is controlled so as to photograph. Then, after such control processing, an image photographed by the second photographing means is recorded by the recording means.

  According to the first aspect of the present invention, the specific region through which the human body passes is photographed three-dimensionally from above with the first photographing means, while a part thereof, that is, an arbitrary partial region is photographed in the specific direction, that is, the human body. Shoot from the front. Accordingly, a three-dimensional image overlooking the entire specific area is obtained from the first photographing means, and an image in which a part of the specific area is captured from the front is obtained from the second photographing means.

  The vertex of the subject is detected based on the three-dimensional image by the first photographing means, and it is determined whether or not the detection result satisfies the height condition. Accordingly, by setting a height condition suitable for the human body, it is possible to distinguish the human body from other subjects.

  If it is determined that the height condition is satisfied, the face position of the human body is estimated from the detection result, that is, the apex, the posture of the second photographing means is controlled so as to photograph the partial region including the estimation result, and after the control process An image photographed by the second photographing means is recorded. Therefore, by setting the partial area, that is, the field of view of the second recording means to a size suitable for the human face, a human face image passing through the specific area is recorded.

  According to the first aspect of the present invention, the face image of the human body passing through the specific position can be recorded without performing complicated calculation such as pattern matching.

  In a preferred embodiment, the first photographing means includes a plurality of cameras (14a, 14b,...) That simultaneously photograph the first object scene from different directions, and the three-dimensional images are respectively photographed by the plurality of cameras. The vertex detection means is configured to perform detection based on parallax between the plurality of images.

  That is, the specific area is simultaneously photographed from different directions by a plurality of cameras. Therefore, the three-dimensional image is composed of a plurality of images respectively captured by a plurality of cameras, and the vertex detection means performs detection based on the parallax between the plurality of images. Thereby, the vertex of the subject in the specific region, that is, the top of the human body can be easily detected.

  The estimation means includes three-dimensional coordinate calculation means (S63) for calculating three-dimensional coordinates corresponding to the face position by performing a predetermined calculation on the detection result of the vertex detection means. That is, the three-dimensional coordinates corresponding to the face position are calculated by performing a predetermined calculation on the detection result of the vertex detection means. As a simple calculation, for example, by adding a constant (α, β, γ) calculated from the average face shape and size to the vertex coordinates (Xt, Yt, Zt), the center coordinates of the face There is an operation for obtaining (Xc, Yc, Zc). Then, such three-dimensional coordinates are used as an estimation result.

  The face image recording apparatus according to the invention of claim 2 is dependent on claim 1, wherein the first photographing means repeats photographing, and the vertex detecting means performs detection every time photographing is performed.

  According to the second aspect of the present invention, by repeating vertex detection, accurate vertex detection can be performed without a special sensor for detecting the movement of the subject.

  The face image recording apparatus according to the invention of claim 3 is dependent on claim 2, and further comprises a movement speed calculation means for calculating a movement speed of the vertex based on a plurality of detection results of the vertex detection means, wherein the first determination means is It is further determined whether or not the calculation result of the moving speed calculating means satisfies the moving speed condition, and the estimating means performs estimation when it is determined by the first determining means that both the height condition and the moving speed condition are satisfied.

  In the invention of claim 3, the moving speed calculating means calculates the moving speed of the vertex based on a plurality of detection results of the vertex detecting means. The vertex moving speed can be obtained simply by obtaining the distance between the vertex detected last time and the vertex detected this time and dividing this distance by the imaging period. Whether or not the calculation result satisfies the moving speed condition is further determined by the first determining means, and the estimation process of the estimating means is executed when it is determined that both the height condition and the moving speed condition are satisfied.

  According to the invention of claim 3, since the moving speed is taken into consideration in addition to the height of the apex, the discrimination accuracy between the human body and the other subjects is increased.

  A face image recording apparatus according to a fourth aspect of the present invention is dependent on any one of the first to third aspects, and includes a contour line detecting means (S17) for detecting a contour line from an image photographed by the second photographing means, and a contour line. And a second discriminating unit (S21) for discriminating whether or not the contour line detected by the detecting unit and the image in the region surrounded by the contour line satisfy the size condition and the color condition, respectively, and the recording unit When the second determining means determines that both the size condition and the color condition are satisfied, the recording process is executed.

  In the invention according to claim 4, the contour line is detected by the contour line detection unit from the image photographed by the second photographing unit. The second discriminating unit discriminates whether or not the detected contour line and the image in the region surrounded by the contour line satisfy the size condition and the color condition, respectively. Therefore, by setting a size condition and a color condition suitable for a human face, it is possible to discriminate the human face from other subjects.

  The recording process is executed when it is determined that both the size condition and the color condition are satisfied. Therefore, a human face image can be recorded with higher accuracy.

  A face image recording apparatus according to the invention of claim 5 is dependent on claim 4 and third determination means for determining whether or not an image in an area surrounded by the contour detected by the contour detection means has been recorded. (S25) is further provided, and the recording means executes the recording process when it is determined by the third determining means that recording has not been completed.

  According to the fifth aspect of the present invention, whether or not the image within the contour line has been recorded is determined by the third determination means determination. The recording process is executed when it is determined that the image within the contour line has not been recorded.

  According to the invention of claim 5, only the unrecorded face image can be recorded efficiently.

  In the preferred embodiment, the recording means records only the image in the area surrounded by the contour detected by the contour detecting means. That is, since the face image from which the background has been removed is recorded, the data amount of the recorded face image can be suppressed, and it is easy to determine whether or not it has been recorded.

  According to the present invention, it is possible to easily record a face image of a human body passing through a specific position.

  The above object, other objects, features and advantages of the present invention will become more apparent from the following detailed description of embodiments with reference to the drawings.

  Referring to FIG. 1, an entrance management system 10 of this embodiment includes a stereo camera 14 that is fixedly installed vertically above a gate G, and a front camera 16 that is rotatably installed in front of the gate G. And a server 12 that controls both cameras and manages room entry.

  The stereo camera 14 is composed of a plurality of cameras, here two cameras 14a and 14b. The two cameras 14a and 14b capture the same scene, and two corresponding images of the same scene (hereinafter “2”). Two images "). Accordingly, there is a parallax between the two images output from the stereo camera 14.

  The server 12 three-dimensionally identifies the position of the face H of the person entering the room from the gate G based on the two images, and the posture of the front camera 16 so that the center of the object scene overlaps the identified position. To control.

  The server 12 stores face images of persons who have entered and registered in the past as registered face images, and each time a face image is captured by the front camera 16, the captured face image is registered as a registered face image. Check the image and register it if it is not registered. On the other hand, if the photographed face image has been registered, the name (or identifier) of the person corresponding to the face image is recorded together with the entry time. In addition, when registering a face image of a room occupant or recording a name or the like, information to be notified to the person may be output, or a warning message for an attendant may be output.

  Specifically, the server 12 includes a hard disk drive (HDD) 24. The registered face image is stored in the hard disk 24d in the HDD 24 (see FIG. 3B). Referring also to FIG. 2, the stereo camera 14 uses an area E1 allocated in the vicinity of the gate G as an imaging area. Here, the region E1 is a rectangular parallelepiped region obtained by translating the surface of the gate G by a certain distance in the y direction, and has a size of about 1 m wide × 1 m deep × 2 m high, for example.

  In this embodiment, as shown in FIGS. 1 and 2, the stereo camera 14 is the origin, the vertical downward direction is the z axis, the direction perpendicular to the plane of the gate G and going from the outdoor to the indoor is the y axis, and The direction perpendicular to the z-axis and the y-axis is taken as the x-axis.

  Further, the front camera 16 sets an arbitrary area E2 in the area E1 as an imaging area. The region E2 has a size of about 0.4 m × 0.3 m × 0.4 m, for example. The server 12 controls the posture of the front camera 16 so that the center point of the area E2 matches the center point Hc of the human face H. Accordingly, the area E1 is a fixed area, while the area E2 is a moving area that moves according to a person.

  The two images output from the stereo camera 14 are temporarily stored separately in the stereo camera image storage area 20a of the RAM 20 through the memory controller 22 (see FIG. 3A). On the other hand, an image photographed by the front camera 16 is temporarily stored in the front camera image storage area 20b of the RAM 20 through the memory controller 22 (see FIG. 3A).

  The CPU 26 reads two images from the stereo camera image storage area 20a through the memory controller 22, and estimates the face position of the person based on the parallax between the two read images.

  That is, the CPU 26 reads two images from the stereo camera image storage area 20a, and first determines whether or not a subject other than the background is included in the scene based on at least one of the two read images. . If it is determined that there is a subject other than the background, the vertex of the subject is detected based on the parallax between the two read images.

  Specifically, vertex detection is performed according to the following procedure. First, the area E1 is divided into a predetermined number (for example, 50 × 100) of areas. Next, the parallax between the two images is obtained for each region, and the distance to the subject is calculated for each region based on the parallax. Thereby, distance distribution information indicating the distance to the subject for each region is obtained. Then, the vertex of the subject is detected from this distance distribution information. The detected vertex coordinates (Xt, Yt, Zt) are stored in the vertex coordinate storage area 20c of the RAM 20 (see FIG. 3A).

  Subsequently, the CPU 26 calculates the moving speed of the vertex from the plurality of coordinates in the vertex coordinate storage area 20c. And it is discriminate | determined whether the height and moving speed of a vertex are in the predetermined range, respectively. If both the height and the moving speed are within a predetermined range, the subject is regarded as a human body.

  Therefore, the CPU 26 estimates the center position Hc of the face H from the coordinates (Xt, Yt, Zt) of the vertex, that is, the top of the head. The estimation result is stored in the face center coordinate storage area 20d. The center coordinates (Xc, Yc, Zc) of the face H are simply constants (α, Yt, Zt) calculated from the average face shape and size to the coordinates (Xt, Yt, Zt) of the top of the head. β, γ) can be obtained by adding them respectively.

  Then, the CPU 26 refers to the face center coordinates and controls the posture of the front camera 16 so that the center of the imaging area E2 of the front camera 16 matches the center of the face H. Then, after the posture control is completed, a contour line is detected from the front camera image in the front camera image storage area 20b.

  The contour line detection is performed by the following procedure, for example. First, the area E2 is divided into a predetermined number of areas (for example, 400 × 400). Next, with reference to the luminance information included in the front camera image, the luminance difference between two adjacent areas is calculated. Then, the calculation result is compared with a threshold value, and if the threshold value is exceeded, it is determined that a boundary line exists at the position of the two regions. By repeating this process, a contour line is detected.

  Subsequently, the CPU 26 determines whether or not the image surrounded by the contour line is a human face, based on the detected size of the contour line and the color information in the region surrounded by the contour line. If both the size and the color are within a predetermined range, it is determined that the image surrounded by the contour line is a human face, and the image surrounded by the contour line is extracted from the front camera image. The extracted image is stored in the face image storage area 20e (see FIG. 3A) as a face image of the person who is about to enter the room.

  Next, the CPU 26 compares the face image in the face image storage area 20e with the registered face image in the hard disk 24d, and performs new registration if it is not registered. In the new registration process, a face image is read from the face image storage area 20e, and the read face image is recorded on the hard disk 24d via the HDD 24. In addition, an operation for inputting gender, age and other personal information via a keyboard or the like (not shown) is accepted, and the accepted personal information is added to a database (not shown). On the other hand, if the face image has been registered, the personal information of the person corresponding to the face image is acquired from the database, and the name of the person, that is, the person entering the room, the room entry time, etc. are recorded on the hard disk 24d.

  More specifically, the CPU 26 of the server 12 operates according to the flowcharts of FIGS. Note that programs corresponding to these flowcharts are stored in the ROM 28.

  Referring to FIG. 4, when the power is turned on, first, in step S1, the memory controller 22 is instructed to capture images from the stereo camera 14 and the front camera 16. In response to this, the memory controller 22 acquires images from the stereo camera 14 and the front camera 16, respectively, and writes the acquired images in the stereo camera image storage area 20a and the front camera image storage area 20b in the RAM 20, respectively.

  Subsequently, in step S3, the image in the stereo camera image storage area 20a is referred to, and in step S5, it is determined whether or not only the background is present. If the image in the stereo camera image storage area 20a and the background image stored in advance satisfy the matching condition, “YES” is determined in the step S5, and the process returns to the step S3. Note that the determination may be based on a three-dimensional image composed of two images in the stereo camera image storage area 20a or based on one of the two images, that is, a planar image.

  If “NO” in the step S5, the process shifts to a step S7 to detect the vertex of the subject and calculate the moving speed of the vertex based on the two images in the stereo camera image storage area 20a (described later). The coordinates (Xt, Yt, Zt) of the detected vertex are stored in the vertex coordinate storage area 20c (see FIG. 3A).

  In a succeeding step S9, it is determined whether or not the detected vertex height and the calculated moving speed are appropriate as a human body value. If “NO” here, the process returns to the step S3, and if “YES”, the process proceeds to the step S11. More specifically, it is determined whether or not the height and the moving speed satisfy the height condition and the moving speed condition. If both the height condition and the moving speed condition are satisfied, the human body is determined. If either one is not satisfied, it is determined that it is not a human body.

  In step S11, the center position of the face H is estimated based on the coordinates (Xt, Yt, Zt) of the vertex, that is, the top of the head (described later). In step S13, the posture of the front camera 16 is controlled so that the center of the region E2 coincides with the center Hc of the face H (see FIGS. 1 and 2). When the posture control is completed, the process proceeds to step S15.

  Next, in step S15, an image in the front camera image storage area 20b is referred to, and in step S17, a contour line is detected from the front camera image (described later). In step S19, the size of the contour line and the color in the region surrounded by the contour line are each evaluated. In step S21, it is determined whether the evaluation result is appropriate as a human face value. If the size and the color are within the predetermined ranges, YES is determined in step S21, and the process proceeds to step S23. If “NO” here, the process returns to the step S3.

  In step S23, an image in the region surrounded by the contour line is extracted from the front camera image. The extracted image is stored in the face image storage area 20e (see FIG. 3A) as a face image. In a succeeding step S25, it is determined whether or not the face image in the face image storage area 20e has been registered by collating with the registered face image in the hard disk 24d. If there is a registered face image that satisfies the matching condition with this face image, YES is determined in the step S25, and the process proceeds to a step S27. If there is no registered face image satisfying the matching condition, the process proceeds to step S29.

  In step S27, a room entry recording process is executed. In the room entry recording process, personal information of a person corresponding to the face image is acquired from a database (not shown), and the name and room entry time are recorded on the hard disk 24d. Then, it returns to step S3.

  On the other hand, in step S29, a new face image is registered. In the new registration process, this face image is recorded in the hard disk 24d, and personal information corresponding to this face image is added to the database. Then, it returns to step S3. In steps S27 to S29, a notice to the person, a warning to a staff member, or the like may be performed as necessary.

  Next, the vertex detection and movement speed calculation processing in step S7 follows the subroutine of FIG. Referring to FIG. 5, first, in step S41, the object scene of stereo camera 14, that is, area E1, is divided into a predetermined number (for example, 50 × 100) of areas. Subsequently, in step S43, the parallax between the two images stored in the stereo camera image storage area 20a is calculated for each area, and in step S45, the distance to the subject is calculated for each area based on the parallax. Thereby, distance distribution information indicating the distance to the subject for each region is obtained.

  In step S47, the vertex of the subject is detected based on the distance distribution information. In step S49, the three-dimensional coordinates of the detected vertex are stored in the vertex coordinate storage area 20c. In step S51, the vertex moving speed is calculated with reference to the vertex coordinate storage area 20c. After the calculation, the process returns to the upper layer routine.

  On the other hand, the face center estimation process in step S11 follows the subroutine of FIG. Referring to FIG. 6, the coordinates (Xt, Yt, Zt) of the vertex coordinate storage area 20c are referred to in step S61, and the face center coordinates (Xc, Yc, Zc) are calculated by the following approximate expression in step S63. .

Xc = Xt + α, Yc = Yt + β, Zc = Zt + γ
Here, (α, β, γ) are constants calculated from the average face shape and size.

  In step S65, the calculation result is stored in the face center coordinate storage area 20d. After storing, return to the upper layer routine.

  Further, the contour detection processing in step S17 follows the subroutine of FIG. Referring to FIG. 7, first, in step S71, the object scene (area E2) of front camera 16 is divided into a predetermined number (for example, 400 × 400). Subsequently, in step S73, the luminance information included in the front camera image is referred to, and in step S75, a contour line is detected based on the luminance difference between adjacent regions.

  As specific processing of this contour line detection, for example, a luminance difference between two adjacent regions is calculated, and if the calculation result exceeds a threshold value, the boundary line between the two regions is part of the contour line. Is determined. By executing such determination processing over the entire front camera image, it is possible to detect the contour line of the face image included in the front camera image. Alternatively, it is possible to detect the outline efficiently by performing sequential determination until a part of the boundary line is detected for the first time, and determining only the region near the outline while proceeding along the direction of the outline after detection. can do. After detection, return to the upper layer routine.

  As is apparent from the above, the face image recording apparatus 10 of this embodiment photographs the area E1 through which a person passes from above with the stereo camera 14, while the front camera 16 captures an area E2 that is an arbitrary part of the area E1. Shoot with. Therefore, the RAM 20 stores a three-dimensional image overlooking the entire area E1 and an image capturing a part of the area E1 from the front.

  The CPU 26 detects the vertex Ht of the subject based on the three-dimensional image stored in the RAM 20 (S47), calculates the movement speed of the detection result (S51), the detection result and the calculation result are the height condition and the movement speed condition. Are determined (S9). Therefore, by setting the height condition and the moving speed condition suitable for the human body, it is possible to discriminate the human body from other subjects.

  If it is determined that both the height condition and the moving speed condition are satisfied, the face center position Hc is estimated from the vertex Ht (S11), and the front camera 16 so as to photograph the region E2 including the face center position Hc as the center. Is controlled (S13). Then, a contour line is detected from the image photographed by the front camera 16 after the control process (S17), and the detected contour line and the image in the region surrounded by the contour line have the size condition and the color condition, respectively. It is determined whether or not satisfied (S21). Therefore, by setting a size condition and a color condition suitable for a human face, it is possible to discriminate the human face from other subjects.

  If it is determined that both the size condition and the color condition are satisfied, an image in the area surrounded by the outline is extracted from the front camera image (S23), and whether or not the extracted image has been recorded is determined. If it is not recorded, it is recorded on the hard disk 24d (S29). Therefore, by setting the area E2, that is, the field of view of the front camera 16, to a size suitable for the human face, an unregistered human face image passing through the area E1 is recorded on the hard disk 24d.

  Thereby, an unregistered human face image entering from the gate G can be recorded without performing complicated calculation such as pattern matching.

  In this embodiment, when it is determined that the human body is included in the image of the stereo camera 14, the position of the face is estimated to control the posture of the front camera 16, and then the face is included in the image of the front camera 16. The image is recorded when it is determined that the face is included, but the latter determination is omitted and the image of the front camera 16 is recorded immediately after the posture control. May be. However, the recording accuracy is improved by performing the latter determination.

  Further, although only the image in the region surrounded by the contour line is recorded, the front camera image may be recorded as it is. However, by recording only the image within the contour line, the data amount of the image to be recorded can be reduced, and it becomes easy to determine whether or not the image has been recorded.

It is a block diagram which shows the structure of one Example of this invention. It is an illustration figure which shows the operation example of FIG. 1 Example. (A) is a block diagram showing a part of a memory map applied to the embodiment of FIG. 1, and (B) is a block diagram showing another part of the memory map applied to the embodiment of FIG. It is a flowchart which shows a part of CPU operation | movement applied to FIG. 1 Example. It is a flowchart which shows a part of other CPU operation | movement applied to FIG. 1 Example. It is a flowchart which shows a part of other CPU operation | movement applied to FIG. 1 Example. FIG. 12 is a flowchart showing yet another portion of the CPU operation applied to the embodiment in FIG. 1.

Explanation of symbols

10 ... Room management system 12 ... Server 14 ... Stereo camera 16 ... Front camera 20 ... RAM
24d ... Hard disk 26 ... CPU
G ... Gate H ... Face (head)
Hc ... Center of face Ht ... Top of head

Claims (5)

A face image recording apparatus for recording a face image of a human body passing through a specific area in a specific direction,
First imaging means for imaging the specific area three-dimensionally from above the specific area;
A second imaging means for imaging an arbitrary partial area that is a part of the specific area from the specific direction;
Vertex detecting means for detecting the vertex of the subject in the specific area based on the three-dimensional image photographed by the first photographing means;
First discrimination means for discriminating whether or not the detection result of the vertex detection means satisfies a height condition;
Estimating means for estimating the face position of the human body from the detection result of the vertex detecting means when the first determining means determines that the height condition is satisfied;
Control means for controlling the posture of the second photographing means so as to photograph a partial region including the face position estimated by the estimating means, and an image photographed by the second photographing means after the control processing of the control means A face image recording apparatus comprising recording means for recording the image. The first photographing means repeatedly performs photographing,
The face image recording apparatus according to claim 1, wherein the vertex detection unit performs detection every time photographing is performed. A moving speed calculating means for calculating a moving speed of the vertex based on a plurality of detection results of the vertex detecting means;
The first determining means further determines whether or not the calculation result of the moving speed calculating means satisfies a moving speed condition;
The face image recording apparatus according to claim 2, wherein the estimation unit performs estimation when the first determination unit determines that both a height condition and a movement speed condition are satisfied. A contour detection unit that detects a contour line from an image photographed by the second photographing unit; and a contour condition detected by the contour detection unit and an image in an area surrounded by the contour line A second discriminating unit for discriminating whether or not each of the color conditions is satisfied;
4. The face image recording apparatus according to claim 1, wherein the recording unit performs a recording process when it is determined by the second determination unit that both the size condition and the color condition are satisfied. 5. A third determination unit for determining whether or not an image in the region surrounded by the contour detected by the contour detection unit has been recorded;
The face image recording apparatus according to claim 4, wherein the recording unit executes a recording process when it is determined by the third determination unit that recording has not been completed.

Images