JP3168546B2 - Portrait processing device and automatic cash transaction device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION In an automatic machine or the like for handling cash or issuing a certificate equivalent to cash, a portrait processing apparatus for processing a portrait of an operator taken as a moving image and a cash provided with the portrait processing apparatus The present invention relates to an automatic transaction device.

[0002]

2. Description of the Related Art In Japan, a seal is used to identify a person in the processing of depositing and depositing cash, issuing securities, issuing certificates, and the like. Conventionally, these processes have been performed manually, so that the identification of the person using a seal has been established. recent years,
The case where the above processing is replaced with an automatic machine is remarkable in terms of processing efficiency and accuracy. For example, in a cash dispenser (CD, ATM, etc.), a card in which a password is registered in advance is used to obtain online information from a password recorded on the card or information such as an account number recorded on the card. Reads out the personal identification number registered in the center, compares the personal identification number with the personal identification number input by the operator to the automatic machine, and identifies the personal identification when they match. In the above system, if another person knows the password, the security is reduced. In an automatic machine that issues securities, certificates, and the like that do not use a card, a system that uses a fingerprint to identify an individual is being studied. However, it is generally difficult to spread the system to make criminals aware.

In the above-described automatic machine, a portrait of an operator may be photographed as a moving image or a still image for personal identification or operator history management. In this case, the face of the operator is distributed over a wide range with respect to the portrait imaging camera, such as a person who is tall, a person who is short, a person who leans to the right side of the apparatus, or a person who leans to the left side due to the dominant hand. In order to cover these ranges, a camera generally adopts a wide-angle shooting method capable of shooting a wide range. The portrait captured at the wide angle is small, and it is difficult to identify the operator. Further, in the still image capturing method, it is very difficult to capture an image when the operator faces the camera. Japanese Patent Application Laid-Open Nos. 60-147894, 60-153578, and 3-201871 are examples of applications related to the above conventional example.

[0004]

SUMMARY OF THE INVENTION The present invention eliminates the above-mentioned disadvantages of the prior art and allows the operator to move in a predetermined direction (for example,
A portrait recording device for recognizing a state when the camera is facing the camera in a right-angled manner, and further cutting out a required portrait in a predetermined direction from information captured by a wide-angle camera, and an automatic cash transaction including the portrait processing device It is to provide a device.

[0005]

A portrait memory for storing portrait image information of a person having no background information, and a contour extraction for binarizing the portrait image information and generating portrait contour information based on the binarized information. A portrait dictionary storing portrait data, which is a set of feature data including a person's eyes, nose, mouth, etc., and cutout area information of an image corresponding to the portrait data for various persons; A recognition processor for sequentially comparing portrait outline information with each portrait data in the portrait dictionary and outputting, when a match is obtained, cropping area information of the corresponding image; An image clipper is provided for inputting image information and extracting and outputting portrait part information from portrait image information based on clipped area information of the image. Further, an imaging device, means for detecting that a person is in front of the imaging device, and a background memory for storing the output of the background screen from the imaging device at predetermined intervals when there is no detection output of the means, An image memory for storing the output of the screen from the imaging device at predetermined intervals when there is a detection output of the means, and image information in the image memory and image information in the background memory when there is a detection output of the means , A changing pixel extractor that generates portrait image information of a person from which background information has been removed and outputs the portrait image information to the portrait memory, and a storage device that stores the extracted and output portrait portion information.

[0006]

According to the present invention, portrait outline information is generated based on portrait image information of a person having no background information, and the portrait outline information is sequentially compared with portrait data stored in a portrait dictionary. The cut-out area information of the paired image can be extracted, and the portrait portion information can be cut out from the portrait image information based on the cut-out area information of the image. Further, the imaging device sends a background screen to a background memory to store the person when the person is not in front of the imaging device, and sends and stores the background screen and the person screen to the image memory when the person is present. Generates portrait image information of a person from whom background information has been removed based on the contents stored in the image memory and the background memory, and stores the portrait image information in the portrait memory. Then, the extracted and output portrait part information is stored in the storage device. Thereby, only a portrait when facing a predetermined direction (for example, facing directly) can be cut out, and only a portrait existing as a part thereof can be cut out from a wide range, and further, it can be cut out in a background. Only the portrait when the moving person faces a predetermined direction (for example, facing directly) can be cut out by removing the background.

[0007]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described in detail with reference to FIGS.
FIG. 1 is a block diagram showing a configuration of a portrait processing apparatus according to the present invention, and shows an example in which the present invention is applied to an automatic machine that requires an operation history of an operator. A portrait processing section of an automatic machine using the portrait processing apparatus of the present invention is a camera (imaging apparatus) for capturing a portrait of an operator.
1, a human sensor 2 for detecting that an operator is in front of an automatic machine, a control circuit 3, an image memory 4 for storing a captured image of a camera, a background memory 5 for storing a background image, an image processing unit 6, image compression Device 7 and an image file 8.

If no operator is present in front of the automatic machine and there is no output from the person sensor 2, the control circuit 3
Is activated, and the output of the background screen from the camera 1 is recorded in the background memory 5 at predetermined intervals. When the operator approaches the automatic machine, the person sensor 2 detects the approach and sends a sensed output to the control circuit 3. The control circuit 3 deactivates the background memory 5, and the image memory 4, the image processing unit 6, The image compressor 7 and the image file 8 are activated, and the screen output from the camera 1 is recorded in the image memory 4 at predetermined intervals. The image information recorded in the image memory 4 and the image information recorded in the background memory 5 are processed by the image processing unit 6. The details of the image processing unit 6 will be described with reference to the block diagram of FIG. FIG. 3 shows an example of a processed image showing the progress of processing.
Shown in The image processing unit 6 includes a changing pixel extracting unit 61, an outline extracting unit 62, a portrait dictionary 63, a recognition processing unit 64, an image extracting unit 65, and a portrait memory 66.

[0009] The person who operates the automatic machine is a tall person, a short person, and various people, and has various habits such as a person standing on the right side and a person standing on the left side of the automatic machine. Accordingly, since the portrait portion of the operator is distributed over a considerably wide range with respect to the camera installed in the automatic machine, the image pickup range of the camera is set sufficiently wide so that the portrait face is not cut off.
FIG. 3A shows an example of an image of the operator taken by the camera 1, and the background where the apparatus is installed is also taken at the same time.

[0010] The changing pixel extractor 61 is provided with image information of the background memory 5 where the image immediately before the operator approaches is captured,
A circuit for comparing image information of the image memory 4 in which a portrait of an operator is captured and extracting a portion having an arbitrary difference between pixel signals of both image information.
The output of 1 is image information including only the portrait portion of the operator, in which background image information is removed from all images captured by the camera. This example is shown in FIG. The portrait image information of the operator obtained by the changed pixel extracting device 61 is stored in the portrait memory 6.
6 is stored. The data in the portrait memory 66 is processed by the contour extractor 62, and portrait contour information, which is the contour information of the portrait image, is output to the recognition processor 64. The processing in the contour extractor 62 is obtained, for example, by binarizing a portrait at an arbitrary threshold level of an image, and setting data having the same value in both the upper and lower / left and right directions to "0" for an arbitrary pixel and "1" for the other pixels. .

In the portrait dictionary 63, various persons assumed as operators, such as a standard male, a standard female, a spectacle user, an eye patch user, etc., are directed in a predetermined direction (for example, directly facing) to the automatic machine. A set of feature data including eyes, a nose, a mouth, and the like when facing, for example, reference information based on portrait outline information is registered in a portrait dictionary as portrait data of various assumed persons. . Each of the dictionary data registered in the portrait dictionary is registered portrait data and cutout area information of an image of the portrait data.
In the recognition processor 64, portrait data on each type of person registered in the portrait dictionary 63 based on face width, eyes, mouth, nose, and other information in the portrait outline information input from the outline extractor 62. And if there is portrait data for which a match equal to or greater than the predetermined threshold value is obtained, the cut-out area information of the image corresponding to the portrait data is read from the portrait dictionary 63, and the portrait memory 66 is read out using the cut-out area information of this image. From the portrait image information read from the operator, that is, image information of the face portion is cut out and output. This example is shown in FIG.

Next, the cutout area information of the image and the cutout of the image information will be described in detail. FIG. 4 shows an image serving as a basis of the portrait dictionary, in which the x and y directions are binarized with m and n dot sizes, respectively, and the outline is extracted. In FIG. 4, the image center X _C in the x direction is a value at which the number of pixels having a signal (portrait image) existing to the left and right of X _C is equal. Further, the image center Y _C in the y direction is a value in which the number of pixels having a signal existing above and below Y _C is equal to each other as described above. Assuming that the total number of signals (portrait pixels) present in the entire image in FIG. 4 is T, when portrait pixels in each column are sequentially integrated from the first column in the x direction, the product value is T /
The value immediately before (after) equal to or exceeding 2 is X
_C. If Y is determined in the same manner, Y _C is obtained. The center of gravity of the image is defined as a point (X _C , Y _C ). The cut-out range of the image is X _C −
X ₁ , X _C + X ₂ , and Y _C −Y ₁ , Y _C + Y _{2 in} the y direction. Here, X ₁ , X ₂ , Y ₁ , and Y ₂ are image cutout area information relating to portrait data. Next, clipping of image information will be described with reference to FIG. The image center of gravity is obtained by the same method as described above with respect to the portrait outline information representing the outline extraction image shown in FIG. 3B, and is image cutout area information related to portrait data in the portrait dictionary that matches the portrait outline information. With X ₁ , X ₂ , Y ₁ , and Y ₂ , a portion indicated by a broken line in FIG. Here, if the portion to be imaged in FIG. 3B becomes wide, a slight error occurs in the cutout range, but X ₁ , X ₂ ,
By properly selecting Y ₁ and Y ₂ , the portrait can be prevented from being cut off.

Further, another example of the cutout area information of the image and the cutout of the image information will be described. FIG.
FIG. 5 is a diagram showing an image serving as a basis of a portrait dictionary, similar to FIG. 4, in which the x and y directions are binarized with m and n dot sizes, respectively, and contours are extracted. In FIG. 5, Y _T is a signal in the y direction (portrait pixel).
Shows the top row with presence. A row separated from Y _T by a predetermined number of rows (one pixel in the y direction is defined as one row) Y _K is defined as Y _T + Y _K, and Y _T + Y
_The column where the leftmost pixel in the x direction is located in _K rows (one pixel in the x direction is defined as one column) is defined as _XL .

Here, the uppermost row Y _T and the left reference column _XL are used as cut-out reference coordinates.
Y _T -Y ₃ in the direction, Y _T + Y _4, in the x-direction X _L -X _3, X _L +
Let X ₄ be X ₄ , X ₃ , X ₄ , Y ₃ , and Y _{4 be} the cutout area information of the image related to the portrait data. Extraction of image information is performed in the same manner as described above. In other words, the reference coordinates (X _L, Y _T) by the same method to portrait contour information representing a contour extracting image shown in (b) of FIG. 3 seeking, portrait in portrait dictionary that match該肖image contour information By using X ₃ , X ₄ , Y ₃ , and Y ₄ , which are the cut-out area information of the image relating to the data, the portion indicated by the broken line in FIG. In this case, the reference coordinates has been described in X _L in the x-direction left reference column, no problem even in the right reference. Further, it is also possible to set an image cut-out range using feature points of an image (for example, eyes, nose, mouth, ears, etc.) as reference points.

The pixel data, which is the portrait part information cut out above, is directly or image-compressed to a magnetic disk,
K and the like, and output to a printer or the like.
In the above description, the recognition processing has been described using outline information with a reduced data amount, but binarized image information or unprocessed portrait data can also be used if a correspondence is established with a dictionary. The switching of the camera information between the background memory 5 and the image memory 4 is performed based on the information of the person sensor. However, if the operation state of the automatic machine becomes clear based on the processing flow, the switching is triggered by the processing flow or the like. May be determined according to the following.

[0016]

According to the present invention, it is possible to cut out only a portrait when facing a predetermined direction (for example, facing directly). Further, it is possible to cut out only a portrait existing as a part of the wide range. Furthermore, it is possible to remove only the portrait when the person moving in the background faces in a predetermined direction (for example, facing directly) by removing the background. Then, the portrait of the operator of the automatic teller machine can be cut out and stored only in the portrait direction in which the background is removed in a predetermined direction (for example, facing directly).

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of a portrait processing apparatus according to the present invention.

FIG. 2 is a block diagram illustrating details of an image processing unit 6 in FIG. 1;

FIG. 3 is a diagram illustrating a specific processed image example when a captured image is processed by a portrait processing apparatus.

FIG. 4 is a diagram showing an image serving as a basis of a portrait dictionary for explaining cutout area information of an image.

FIG. 5 is a diagram showing an image serving as a basis of a portrait dictionary for explaining cut-out area information of another image.

[Explanation of symbols]

 Reference Signs List 1 camera 2 human sensor 3 control circuit 4 image memory 5 background memory 6 image processing unit 7 image compressor 8 image file 61 changed pixel extractor 62 contour extractor 63 portrait dictionary 64 recognition processor 65 image cutout 66 portrait memory

──────────────────────────────────────────────────続 き Continuation of front page (72) Inventor Takeshi Miura 3-10-22, Sakae, Naka-ku, Nagoya-shi, Aichi Hitachi Central Software Co., Ltd. (56) References JP-A-62-267868 (JP, A) JP-A-63-157593 (JP, A) JP-A-60-116076 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) G06F 17/60 G06T 1/00 340 G06T 7/00 150 G07D 9/00 461

Claims (3)

(57) [Claims] 1. A portrait memory for storing portrait image information of a person having no background information, binarizing portrait image information input from the portrait memory, and generating portrait outline information based on the binarized information. A portrait dictionary storing portrait data, which is a set of feature data including a person's eyes, nose, mouth, and the like, and cutout area information of an image corresponding to the portrait data for various persons; A recognition processor for sequentially comparing portrait outline information input from the outline extractor with each portrait data in the portrait dictionary, and outputting a cut-out area information of a corresponding image when a match is obtained; An image cutout device that inputs cutout region information of an image, which is an output of the device, and portrait image information of the portrait memory, and cuts out and outputs portrait portion information from the portrait image information based on the cutout region information of the image. A portrait processing device. 2. The portrait processing apparatus according to claim 1, wherein
An imaging device, means for detecting that a person is in front of the imaging device, a background memory for storing a background screen output from the imaging device at predetermined intervals when there is no detection output from the means, and the means An image memory for storing the output of the screen from the imaging device at predetermined intervals when there is a detection output, and inputting the image information in the image memory and the image information in the background memory when there is a detection output of the means. A changing pixel extractor for generating portrait image information of a person from which background information has been removed and outputting the portrait image information to the portrait memory; and a storage device for storing the extracted and output portrait portion information. Portrait processing device. 3. An automatic cash transaction apparatus comprising the portrait processing apparatus according to claim 2.