JP2552727B2 - Face area detection method for photographic originals - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial field of application] The present invention is directed to face area detection for detecting a face portion of a person from an image recorded in a photographic film, a magnetic floppy of an electronic still camera, a memory card or the like. It is about the method.

[Conventional technology]

A background image is recorded in addition to the main image on the photographic original image. However, even if the background image is not properly finished if the density and color balance of the main image are properly finished when the hard copy is made. , Print is often good.

Since most of the main images are human faces, it is known to extract the face part from the image recorded in the original picture and reproduce the hard copy so that the face part is finished properly. . For example, regarding a photographic film, there is JP-A-52-156624, and in the exposure control method described therein, a negative color (hereinafter, referred to as a skin color) corresponding to a skin color in a positive image is defined in advance. ing. Then, compare the color density of each point obtained by scanning the frame,
It is determined whether the pixel at each point has a skin color. Next, calculate the average color density of the area composed of the extracted skin color pixels,
This average color density is used to determine the printing exposure amount. This method is excellent as an exposure control method, but since it is difficult to extract skin color pixels, there is a problem that, for example, if a wall similar to the skin color is included, this portion is mistaken for skin. is there.

Therefore, in order to prevent misidentification of objects and human skin,
By displaying the frame on the monitor and observing it and specifying a part of the face part with a light pen etc., pixels with the same or similar color as this specified pixel are automatically extracted, and A main image detecting method for determining a face area is known (Japanese Patent Laid-Open No. 62-189456).

[Problems to be Solved by the Invention]

The above-described main image detection method is effective in preventing misidentification, but has a drawback that the operation is troublesome because the face portion of a person must be designated with a light pen.

SUMMARY OF THE INVENTION It is an object of the present invention to provide a face area detection method for a photographic original image, which enables the face area to be accurately detected without performing a special operation.

[Means for solving the problem]

In order to achieve the above-mentioned object, the present invention records the focal length data and the object distance data of a taking lens at the time of taking a photograph, reads these data at the time of making a hard copy, and determines the size of the face. On the other hand, by comparing the sizes of the regions estimated to be faces on the original photograph image, it is determined whether or not this region is a face region. In the invention according to claim 2, the face size on the photographic original image is determined from the average size of the face based on the photographing magnification of the photographic subject on the photographic original image. By comparing the sizes of areas estimated to be faces, it is determined whether or not this area is a face area.

[Action]

According to the first aspect of the present invention, the size of the face of the person is determined from the focal length of the photographing lens and the subject distance recorded at the time of photographing, and the size of the region estimated to be the face is compared with the face size. Determine if it is a region.

Further, in the invention according to claim 2, the size of the face on the photographic original image is determined from the average size of the face based on the photographing magnification of the photographic original image of the subject, and the size of the area estimated to be the face and the face size. And is compared to determine whether the area is a face area.

〔Example〕

Examples in which the present invention is applied to a photographic camera will be described below.

FIG. 1 shows a camera for recording subject distance data and focal length data. The shooting lens 10 is an optical system.
It is composed of 10a and a lens barrel 10b that holds it, and is moved along an optical axis 12 by a lens moving mechanism 11. The lens moving mechanism 11 is operated by, for example, an operation member 13 which is linked to a range finder. Further, the lens moving mechanism 11 activates a sensor 14 such as a potentiometer to activate the photographing lens.
The movement amount (linear displacement or rotation amount) of 10 is converted into an electric signal, and this is sent to the bar code conversion circuit 15. The sensor 14 may detect the position of the operation member 13 or the taking lens 10. Furthermore, as the sensor 14, the taking lens 10
In addition to the linear conversion of the movement amount of
It is also possible to use one that divides the movement range of the above into a plurality of parts and generates a zone signal that changes for each of these ranges.

The focal length signal generation circuit 16 generates a signal representing the focal length of the taking lens 10 and sends it to the bar code conversion circuit 15. The barcode conversion circuit 15 sends the barcode data to the driver 17. The driver 17 drives the bar code recording device 18 to record the object distance data and the focal length data represented by the bar code on the photographic film 19. In this embodiment, the bar code recording device 18 is composed of a liquid crystal display 20 for displaying a bar code and a flash discharge tube 21 for illuminating the bar code, and at the same time as or before or after photographing a subject, a screen 19a and a film edge. Subject distance data and focal length data are imprinted between and.

A shutter mechanism 25 is arranged behind the taking lens 10. The shutter mechanism 25 is composed of, for example, two shutter blades 26 and 27, and notches 26a and 27a are formed in the shutter blades 26 and 27, respectively. These shutter blades 26 and 27 are moved by a drive mechanism 28 to a movable pin 29.
When moving toward the fixed pin 30, the fixed pin 30 is moved in the direction away from the fixed pin 30 so that the notches 26a and 27a are overlapped. The overlapping portions of the notches 26a and 27a are opened, and the light passing through the notches 26a and 27a is incident on the photographic film 19 and the subject is photographed therein.

FIG. 2 shows an example of a photographic film on which a bar code is recorded. Perforations 19b are perforated at regular intervals in the photographic film 19, and the perforations 19b are detected by a sensor of a winding stop device to feed one frame. Perforation 19b
On the opposite side of the, subject distance data 3 represented by a bar code
2. The focal length data 33 is imprinted and converted into a visible image by photographic development.

FIG. 3 shows a photographic printer. The white light emitted from the light source 35 enters the mixing box 39 after passing through the cyan filter 36, the magenta filter 37, and the yellow filter 38. For these color correction filters 36 to 38, the amount of insertion into the optical path 41 is adjusted by the filter adjustment unit 40,
Thereby, the three-color light components of the printing light and the intensity thereof are adjusted. In the mixing box 39, diffusion plates are attached to both ends of a rectangular tube whose inner surface is a mirror surface.

The film carrier 44 is arranged at the print position, and the developed photographic film 19 is set and illuminated by the light transmitted through the mixing box 39. In order to ensure the flatness of the photographic film 19, a film mask 45 is provided above the print position. This film mask
As is well known, the opening 45 is formed corresponding to the size of the frame, and when the photo film 19 is transferred, it is lifted by a solenoid (not shown) and presses the photo film 19 during printing.

A bar code reader 46 is arranged in front of the print position, and when sent to the print position, the subject distance data 32 and the focal length data 33 recorded in each frame are read. The read data is sent to the decoder 47 and decoded, and then sent to the calculation unit 48.

Further, a scanner 52 including a lens 50 and an image area sensor 51 is arranged diagonally above the print position, and the transmitted light at each point of the frame set at the print position is measured. The signal of this scanner 52 is A /
After being converted to a digital signal by the D converter 53, it is logarithmically converted by the logarithmic converter 54. The logarithmic converter 54 converts the density signal and sends the density signal to the calculation unit 48. The extraction of the face area, the selection of the printing exposure amount calculation formula based on the presence or absence of the face region, and the calculation of the printing exposure amount by the selected calculation formula are performed. The printing exposure amount of each color obtained is sent to the controller 55.

A printing lens 58 is placed above the print position, and the color paper placed behind the paper mask 59 is
The image of the set frame is enlarged and projected on 60. A shutter 62 whose opening and closing is controlled by a shutter driving unit 61 is arranged between the printing lens 58 and the color paper 60.

FIG. 4 shows the function of the arithmetic unit. The memory 66 stores the three-color density (red density (red density R, green density G, blue density B)) of each pixel obtained by photometry of each shop on the screen. It is read out at the time of calculation and sent to the flesh color pixel extraction unit 67, where the flesh color pixels are extracted. The flesh color extraction is performed by referring to the flesh color definition stored in the memory 68. For example, as shown in FIG. 5, an ellipse is defined in an orthogonal coordinate system having density differences (R−G) and (G−B) as axes, and a pixel density difference included in this ellipse is a skin color. In addition to this ellipse, the skin color range can be defined by a rectangle, a rhombus, or the like.

The position of the pixel determined to be the skin color pixel is sent to the skin color region determination unit 69, where the closed region composed of the skin color pixel is determined to be the skin color region, and then the size of the skin color region is calculated. As the size, even if the length is the longest of the straight lines connecting two points on the contour, the number of skin color pixels may be used instead of the area in order to simplify the calculation.

On the other hand, the memory 72 stores subject distance data, and the memory 73 stores focal length data of the taking lens 10. The arithmetic unit 74 uses these data to obtain the size S of the face of the person recorded in the photo film 19 from the following equation.

S = S ₀ × f / (L−f) Here, each symbol is as follows.

S ₀ : Average face size (length) L: Subject distance f: Focal length f / (L−f): Shooting magnification, which actually represents the reduction ratio of the subject to the original picture.

It should be noted that the above equation is for the case where length is used as the face size S, but when using the area as the face size S, as is well known, the above-mentioned shooting is performed for an average face size (area). The face size S (area) is obtained by multiplying the square of the magnification. The determination unit 75 compares the calculated face size S with the size of the extracted skin color area, and when the size S is the same as or similar to the face size S, determines this skin color area to be a face area. The position coordinates of the face area are sent to the characteristic value extracting unit 76, and the characteristic value of the arithmetic expression for determining the exposure amount for printing is extracted with an emphasis on the finish of the face. That is, the feature value extraction unit 76
Reads the color densities of pixels in the face area from the memory 66 and calculates the average value from the color densities of these pixels. Furthermore,
The feature value extraction unit 76 also calculates the maximum value and the minimum value in the screen, the average density of a specific area such as the central portion and the peripheral portion of the screen, and the like. The obtained feature value is sent to the calculation unit 77, and is substituted into the skin-color-oriented printing exposure amount calculation formula 77a. On the other hand, the characteristic value extraction unit 78 extracts the characteristic value used in the general printing exposure calculation formula 77b. Specific examples of these baking exposure amount calculation formulas are described in detail, for example, in JP-A-62-189456.

The arithmetic expression selection unit 79 selects the printing exposure amount arithmetic expression 77a when the face area exists in the frame 19a, and selects the printing exposure amount arithmetic expression 77b when it does not exist. The printing exposure amount obtained from this selected calculation formula is the controller 55
Sent to

Next, the operation of the above embodiment will be described. When taking a picture, the camera is aimed at the main subject, the composition is determined, and then the operation member 13 is operated. The lens driving mechanism 11 interlocks with the operation member 13 to move the taking lens 10 along the optical axis 12 to adjust the focus. During this focus adjustment, the sensor 14 detects the position of the taking lens 10 and sends a signal to the barcode conversion circuit 15. In addition, the focal length signal generation circuit 16
Is a barcode conversion circuit for the focal length signal of the shooting lens 10.
Send 15

After the focus adjustment, if the release button is pressed, the drive mechanism 28 causes the drive pin 29 to reciprocate, and the shutter mechanism 28 is opened / closed to perform photographing.

Bar code conversion circuit at the same time as before or after taking a picture
15 converts the subject distance and the focal length into a barcode,
This is sent to drive 17. The drive 17 drives the barcode recording device 38, and first displays a barcode indicating the subject distance and the focal length on the liquid crystal display 39. When this display becomes stable, the flash discharge tube 40 is caused to emit light, and the subject distance data 32 and the focal length data 33 are imprinted on the periphery of the screen 19a as shown in FIG.

When the filming of one photo film 19 is completed, the photo film 19 is taken out from the camera and submitted to the photo lab. At the photo developing station, the photo film 19 is developed to convert each frame recorded as a latent image and the data recorded for each frame into a visible image.

The developed photographic film 19 is loaded into the film carrier 44 of the photographic printer shown in FIG. When the photo film 19 is transferred on the film carrier 44, it is read by the bar code reader 46 as shown in FIG.
The subject distance data 32 and the focal length data 33 recorded in the peripheral portion of are read. FIG. 6 shows an embodiment of face area detection. The read data 32 to 33 are decoded by the decoder 47 and then sent to the calculation unit 48. The calculation unit 48 calculates the face size S on the photographic film 19 using these data.

When the top 19a is set at the print position, the scanner 52 measures red, green, and blue transmitted light at each point on the top 19a. The photometric values of the three colors are sent to the calculation unit 48 after undergoing A / D conversion processing and logarithmic conversion processing.

After photometry of each pixel, the calculation unit 48 extracts the flesh color pixel,
Next, a skin color area constituted by skin color pixels is determined. The skin color area and the face size S are compared, and if the same or similar to the face size S, it is determined to be a face area. When the determination of all skin color areas is completed, scene classification is performed.
In this scene classification, if a face area exists, it is determined to be a person shooting scene, and if not, it is determined to be another shooting scene. For a person shooting scene, the printing exposure amount is calculated by substituting the feature value extracted from the feature value extraction unit 76 into the printing exposure amount calculation formula 77a. On the other hand, for other scenes, the printing exposure amount calculation formula 77b
Substituting the feature value extracted from the feature value extraction unit 78 into the, the printing exposure amount is calculated.

After the printing exposure amount is calculated, the controller 55 adjusts the insertion amount of the color correction filters 36 to 38 into the optical path 41 according to the printing exposure amount. After this filter adjustment, the shutter 62 is opened for a fixed time, and the frame 19a is printed and exposed on the color paper 60. Thereafter, similarly, each frame is sequentially printed on the color paper 60.

FIG. 7 shows another embodiment of face area detection. In this embodiment, a rounded closed contour is extracted from the frame by contour detection. Next, the size of the rounded area is compared with the face size S, and when the size is the same as or similar to the face size, this area is determined to be the face area.

In addition, a face pattern corresponding to the subject distance is stored in a memory, a corresponding reference face pattern is selected from the subject distance data, and a pattern matching process extracts a pattern matching or similar to the reference face pattern from the frame. It may be determined that this is a face area. Since the face pattern differs depending on the orientation, it is preferable to store some reference face patterns with different orientations in the memory.

FIG. 8 shows an autofocus device. The distance measuring sensor unit 82 includes a lens 83 and a line sensor 84.
And a light emitting section including a lens 85 and a light source 86. At the time of distance measurement when the release button is pressed halfway, spot-like near infrared light is projected from the light projecting portion toward the subject, and the light reflected here is incident on the line sensor 84. The output signal of the line sensor 84 is sent to the distance measuring circuit 87, and the position of the line sensor 84 at which the reflected light is incident is checked. The subject distance is detected from this incident position, and a signal of this subject distance is sent to the lens setting mechanism 88. This lens set mechanism 88
Sets the taking lens 10 at a position according to the subject distance when the release button is completely pressed.

The object distance signal output from the distance measuring circuit 87 is
It is sent to the bar code conversion circuit 15 and is photographed on the photographic film 19 together with the focal length data as subject distance data 32.

In the above embodiment, the object distance data is represented by a bar code, but an optical mark displayed by a light emitting diode or the like may be used instead. Further, instead of the subject distance data and the focal length data, the face size obtained in the camera may be recorded. Further, in the above embodiment, the length is used as the face size S, but it is well known that not only the length but also the area and the peripheral length are used, and the face size using these may be used.

In addition, instead of recording on the photo film, the IC on the camera
A card may be set and recorded on it, or may be recorded on a cartridge. In an electronic still camera using a fixed focal length photographing lens, subject distance data is written together with image data in a magnetic floppy or a memory card. In this case, since the focal length data is known, the photographing magnification is obtained from the subject distance, and the face size S is calculated based on this.

〔The invention's effect〕

In the present invention having the above configuration, the size of the face of the person is determined from the focal length of the photographing lens and the subject distance recorded at the time of photographing, or from the average size of the face based on the photographing magnification of the photographed original image of the subject. The size of the face on the original image of the photograph is determined, and the size of the area estimated to be the face is compared with the face size to determine whether the area is the face area. Can be accurately detected.

[Brief description of drawings]

FIG. 1 is a schematic diagram showing a photographic camera embodying the present invention. FIG. 2 is an explanatory diagram showing an example of data recording of a photo film. FIG. 3 is a schematic diagram of a photographic printer embodying the present invention. FIG. 4 is a functional block diagram of a calculation unit that calculates the printing exposure amount. FIG. 5 is a graph showing the range of skin color. 6 and 7 are flowcharts showing the procedure for detecting the face area. FIG. 8 is a block diagram showing the autofocus device. 10 …… Shooting lens 18 …… Bar code recorder 19 …… Photo film 20 …… LCD display 21 …… Flash discharge tube 32 …… Subject distance data 33 …… Focal distance data 46 …… Bar code reader 52 …… Scanner 60 …… Color paper 84 …… Line sensor 86 …… Light source.

Claims (2)

(57) [Claims] 1. A focal length data of a taking lens and a subject distance data are recorded at the time of taking a photograph, the face size is determined by reading these data at the time of printing a photograph, and the face size is determined on the original image of the photograph. A face area detection method for a photographic original image, which comprises determining whether or not this area is a face area by comparing sizes of areas estimated to be faces. 2. A face size on a photographic original image is determined from an average size of the face based on a photographing magnification of the photographic subject to the photographic original image, and an area estimated to be a face on the photographic original image is determined for this face size. A method for detecting a face area in a photographic original image, which comprises determining whether or not this area is a face area by comparing the sizes of the areas.