JP4709812B2 - Image generation method and image processing apparatus - Google Patents

Description

  The present invention relates to an image processing apparatus and an image generation method for taking an image of a face and generating an image for an ID photo to be printed with a predetermined size on, for example, an ID card.

  2. Description of the Related Art Conventionally, a photographing machine that captures a face photograph used for an ID card or the like generates an image having a predetermined output size based on the contour of a photographed person's face. That is, the contour of the face is detected from the captured image. Then, a center position that is the center of the face is calculated from the detected face contour, and an output image is generated by cutting out the image as a predetermined output size so that the center position is the center.

  However, as described above, when an image having the center position of the face outline is cut out from the photographed image, the part including the hairstyle is detected as the face outline. When a large hair ornament is attached to the top, the center of the image cut out as the output image is shifted from the center position of the face with the shape.

  As described above, in a camera that automatically cuts out an output image from a photographed image, the cutout position of the output image is shifted due to a hairstyle or the like, and an image in which the center of the face is not the center of the image is generated. In a camera that automatically cuts out an output image from an image that has been removed, a stable image in which the center of the face is the center of the image is generated without the output image being cut out by a hairstyle or the like. An object of the present invention is to provide an image processing apparatus and an image generation method capable of performing the above.

According to the image generation method of the present invention, a face contour is detected from an input image, at least two eye candidates are detected within the detected face contour, and the detected eye candidates are stored in a storage means in advance. The degree of similarity with the left eye sample data is calculated, the degree of similarity between the detected eye candidate and the right eye sample data stored in the storage means in advance is calculated, and the left eye sample data The eye candidate with the highest similarity is extracted as the left eye candidate, the eye candidate with the highest similarity with the right eye sample data is extracted as the right eye candidate, and the similarity of the extracted left eye candidate is The left eye candidate is determined to be the left eye if it is greater than or equal to a predetermined threshold, the right eye candidate is determined to be the right eye if the extracted similarity of the right eye candidate is greater than or equal to a predetermined threshold, and the determined left eye The position and the position of the determined right eye They based determining the center position of the face, and generates an image based on the center position of the determination by the face.

The image processing apparatus according to the present invention includes a first detection unit that detects a face outline from an input image, and a second detection unit that detects at least two eye candidates within the face outline detected by the first detection unit. A similarity between the detection means and each eye candidate detected by the second detection means and the sample data of the left eye stored in advance in the storage means is calculated, and the eye detected by the second detection means A candidate for the eye having the highest degree of similarity between the similarity calculation means for calculating the similarity between the right eye sample data stored in the storage means in advance and the left eye sample data calculated by the similarity calculation means Are extracted as left eye candidates, candidate extraction means for extracting the eye candidate with the highest similarity with the right eye sample data calculated by the similarity calculation means, and the left eye extracted by the candidate extraction means Candidate If the similarity is greater than or equal to a predetermined threshold, the left eye candidate is determined as the left eye, and if the similarity of the right eye candidate extracted by the candidate extraction unit is greater than or equal to a predetermined threshold, the right eye candidate is determined as the right eye. The determination means, a determination means for determining the center position of the face based on the position of the left eye determined by the determination means and the position of the right eye determined by the determination means, and the center position of the face determined by the determination means Image generating means for generating the image

  As described above in detail, according to the present invention, in a camera that automatically cuts out an output image from a taken image, the center of the face is the center of the image without the output image being cut out by a hairstyle or the like. An image processing apparatus and an image generation method capable of generating a stable image can be provided.

  Embodiments of the present invention will be described below with reference to the drawings.

  FIG. 1 shows the appearance of a photographing system including a photographing machine according to an embodiment of the present invention.

  This photographing system includes an illuminator 1 that irradiates a subject, a CCD camera 2 that photographs the subject's face and converts it into image data as a photographed image, a display that displays photographing conditions, photographer operation information, and the like. When the camera 3 processes the image data (captured image) from the CCD camera 2 or controls the display of the display device 3 in accordance with an operation instruction from the photographer, the subject 3 is photographed when the face of the subject is photographed. A chair 6 on which a person sits is provided, and a background plate 7 is provided as a background of the person to be photographed at the time of photographing.

  FIG. 2 is a block diagram showing an internal configuration of the control stand 4 of the photographing machine.

  As shown in FIG. 2, the control table 4 of the photographing machine performs a real-time A / D conversion of analog signal image data as a photographed image like the CCD camera 2 and transfers it to a video memory, a main memory or the like. 9. A video accelerator 10 for performing display processing on the display device 3 by performing drawing processing of image data from the video capture unit 9 and other various image data, image data from the video capture unit 9 and other A CPU for performing various arithmetic processes on data, a main memory for temporarily storing arithmetic results by the CPU, a CPU unit 11 having an interface with various peripheral devices, and an external bus for connecting external peripheral devices such as a printing press (for example, External bus control unit 12 for controlling SCSI), OS, application, various data for control, And a large-capacity storage medium 13 such as a magnetic disk for storing specimen data for a portion extracted from the captured image (for example, a predetermined part constituting a face such as a pupil), image data of a predetermined output size generated from the captured image ( A portable storage medium (for example, a magneto-optical disk such as MO) 14 for storing output data such as an output image), a read / write device (for example, an MO drive) 15 for the storage medium, and an input device operated by an operator It is comprised from the keyboard 16 and the mouse | mouth 17 which are.

  FIG. 3 is a diagram for explaining a relationship with an output image having a predetermined output size cut out from an input image (captured image) captured by the camera.

  FIG. 3 shows a cut-out image 19 cut out at a predetermined output size so that the center position of the face contour detected from the input image 18 taken by the CCD camera 2 of the photographing machine is the center. The detection area 20 that is the target area for detecting the position of the pupil, the face center position 21 determined from the left and right pupils detected from the detection area 20, and the face center position 21 so as to be centered. A cut image 22 cut out with the output size is shown.

  For example, when photographing the face of the subject whose hairstyle is not equal left and right, the CPU unit 11 detects the outline of the face including the hairstyle by determining the color of the background plate 7 from the input image 18. When an image having a predetermined output size is cut out so that the center position of the detected face outline is centered, as shown in FIG. 3, the cut-out image 19 is an image in which the face portion other than the hairstyle is shifted from the center. Become.

  On the other hand, if the center position of the face is obtained from the pupil position of the subject and the image is cut out with a predetermined output size so that this center position is the center, the center of the face is the output size regardless of the hairstyle. An image located at the center of the image is obtained.

  That is, the CPU 11 detects a pupil in the detection area 20 where left and right pupils will exist in the face contour detected from the input image 18. This is because if detection of the pupil is performed from the entire input image or the entire contour of the face, it takes extra time for the detection processing. Therefore, a detection region where the pupil will exist is set in advance for the contour of the face. This is because the processing can be shortened by using only this detection region as a pupil detection target.

  When the left and right pupils are detected from the detection region 20, the CPU unit 11 determines the center position from the detected positions of the left and right pupils. For example, the center position of the left and right pupils is determined as the center position of the face. Further, a cut-out image 22 is generated by cutting out from the input image 18 so that the center position of the face determined from the positions of the left and right pupils is the center.

  As described above, by extracting an image having the center position of the face determined from the position of the pupil as the center, the face position is centered in the image of a predetermined output size without being affected by the hairstyle or the like. An image can be generated.

  Next, processing for cutting out an output image having a predetermined size will be described with reference to the flowchart shown in FIG.

  First, the person to be photographed sits down on the chair 6 installed in front of the background board 7. The photographer (operator) instructs the start of photographing by adjusting the direction of the CCD camera 3 of the photographing device 1 with the handle while viewing the display device 3 on which the image shown in the CCD camera 2 is displayed.

  When the photographer instructs to start photographing, the CPU 11 captures an input image by the CCD camera 2 (step 1). The CPU 11 detects an area other than the color of the background plate 7 from the input image as the position of the face outline (step 2). The CPU 11 determines whether or not the detected position of the face contour is within a specified range (step 3). When it is determined that the position of the face contour is not within the specified range, the CPU 11 performs error processing (step 4). In this error processing, when the photographer looks at the input image displayed on the display device 3 and the face outline in the input image is large, the input image is reduced or the face outline is small. The magnification of the input image is adjusted by enlarging the input image. In addition, when the face outline in the input image is extremely shifted, the input image is too bright, or too dark, the input image is discarded and the photographing process is performed again.

  When it is determined in step 3 that the detected position of the face contour is within the specified range, the CPU unit 11 in the preset pupil detection area 20 has a round part having the same shape as the pupil. Is extracted (step 5). As a result, the CPU unit 11 determines whether there are two or more extracted parts (step 6).

  When it is determined by this determination that the number of extractions is two or more, each extracted part is compared with the sample data of the left and right pupils stored in the HDD 13, and among them, the left and right pupils of similarity are obtained. The highest extraction site is detected (step 7).

  Further, the CPU unit 11 determines whether or not the degree of similarity between the parts detected as the left and right pupils is equal to or greater than a predetermined threshold value (step 8). When it is determined that the similarity between the parts detected as the left and right pupils is equal to or greater than a predetermined threshold, the CPU unit 11 detects the center position of the face from the positions of the left and right pupils, and the center position. The cut-out position of the photograph centering on is calculated (step 9).

  Further, when it is determined in step 6 that the number of extractions is not two or more, or when it is determined in step 8 that the similarity between the parts detected as the left and right pupils is not equal to or greater than a predetermined threshold value, 2, the center position of the face is detected based on the detected face contour position, and the cutout position of the photograph centered on the center position is calculated (step 10).

  Then, the CPU unit 11 cuts an input image at the cutout position based on the positions of the left and right pupils at the step 9 or at the cutout position based on the face contour position at the step 10 to obtain an image of a predetermined output size. Generate (step 11).

In the above embodiment, the center position of the face is detected by detecting the pupil, but the center position of the face is detected by detecting a part other than the pupil constituting the face, such as the nose, mouth, ear, and eyebrows. You may make it judge. Further, for example, a plurality of parts constituting the face such as the ear and pupil, the pupil nose, the nose and mouth, and the mouth and ear may be detected, and the center position of the face may be determined by the combination thereof. A predetermined part constituting the face is detected from the photographed input image to determine the center position of the face, and an image having an output size centered on the center position is generated.

  Thereby, even when a face photograph in which the contour of the face is not equal to the left and right is photographed by a hairstyle or the like, an image having an output size centered on the center position of the face can be generated.

  Further, the position of the pupil is detected as a predetermined part constituting the face from the photographed input image to determine the center position of the face, and an image having an output size centered on the center position is generated.

  This makes it possible to generate an image that is the center position of the face and the center of the output image even when a facial photograph with a non-equal facial contour is taken with a hairstyle, etc. By detecting this, the center position of the face can be reliably detected.

The figure which shows the external appearance structure of the imaging | photography system containing the imaging device for describing embodiment of this invention. The block diagram which shows the internal structure of an imaging device. The figure for demonstrating the output image cut out from an input image. The flowchart for demonstrating the process which produces | generates an output image.

Explanation of symbols

2 ... CCD camera 3 ... Display 11 ... CPU unit 13 ... Recording medium

Claims (10)

Detect the face outline from the input image,
Detect at least two eye candidates within the detected face contour,
The degree of similarity between the detected eye candidates and the left eye sample data stored in advance in the storage means is calculated, and the detected eye candidates and the right eye sample data stored in the storage means in advance The similarity of
Extracting the candidate for the eye with the highest degree of similarity with the sample data for the left eye as the candidate for the left eye, extracting the candidate for the eye with the highest degree of similarity with the sample data for the right eye,
If the extracted left-eye candidate has a similarity greater than or equal to a predetermined threshold, the left-eye candidate is determined as the left eye, and if the extracted right-eye candidate similarity is greater than or equal to a predetermined threshold, the right-eye candidate is determined as the right eye. And
Determining the center position of the face based on the determined left eye position and the determined right eye position ;
An image based on the determined center position of the face is generated.
An image generation method characterized by the above. In the eye detection, at least two eye candidates are detected from a predetermined region set in advance with respect to the face outline.
The image generation method according to claim 1 , wherein: In the detection of the eyes, a part that is likely to be an eye shape of at least two persons from the predetermined region is detected as an eye candidate.
Image generating method according to claim 2, characterized in that. Furthermore, in the generation of the image, an image having a predetermined size based on the determined center position of the face is generated.
Image generating method according to any one of claims 1 to 3, characterized in that. Furthermore, if the position of the face contour detected from the input image is not within a specified range, an error is assumed.
Image generating method according to any one of claims 1 to 4, characterized in that. First detection means for detecting a contour of a face from an input image;
Second detection means for detecting at least two eye candidates within the face contour detected by the first detection means;
The degree of similarity between each eye candidate detected by the second detection means and the left eye sample data stored in the storage means is calculated, and the eye candidates detected by the second detection means are calculated in advance. Similarity calculation means for calculating the similarity with the sample data of the right eye stored in the storage means;
The eye candidate having the highest similarity with the left eye sample data calculated by the similarity calculation means is extracted as a left eye candidate, and the eye having the highest similarity with the right eye sample data calculated by the similarity calculation means Candidate extraction means for extracting the candidate as a right eye candidate;
If the similarity of the left eye candidate extracted by the candidate extraction unit is equal to or greater than a predetermined threshold, the left eye candidate is determined as the left eye, and the similarity of the right eye candidate extracted by the candidate extraction unit is equal to or greater than a predetermined threshold. If there is a determination means for determining the right eye candidate as the right eye,
Determining means for determining the center position of the face based on the position of the left eye determined by the determining means and the position of the right eye determined by the determining means;
Image generating means for generating an image based on the center position of the face determined by the determining means;
An image processing apparatus comprising: The second detection means detects at least two eye candidates from a predetermined region set in advance with respect to the contour of the face;
The image processing apparatus according to claim 6, characterized in that. The second detection means detects, as eye candidates, a part that is likely to be the shape of at least two human eyes from the predetermined region.
The image processing apparatus according to claim 7 , wherein: Further, the image generation means generates an image of a predetermined size with reference to the center position of the face determined by the determination means.
The image processing apparatus according to claim 6 , wherein the image processing apparatus is an image processing apparatus. In addition, there is an error processing means for making an error when the position of the face contour detected from the input image is not within a prescribed range.
The image processing apparatus according to claim 6 , wherein the image processing apparatus is an image processing apparatus.

Images