JPH04346333A - Data extracting method for human face and exposure deciding method - Google Patents

Abstract

PURPOSE:To extract data of a human face from an image in which the human face and ground, woods, and so on resembling to a hue of the human face mixedly exist. CONSTITUTION:An original image is divided into a large number of pieces to measure luminus intensity, the measured data are changed to hue value H and saturation value S, and a two dimensional histogram of HS is made 100-104. The two dimensional histogram is divided into every single peak of mountain 106. To which of the divided mountains the respective picture elements of the original image belong is judged to divide the picture elements into groups corresponding to the divided mountains, an image is divided to every each group, and a proposed area is extracted 108. Whether it is a human face or not is judged from a contour of the proposed area and an inner structure, and photometry data of the area judged to be the human face are output 110.

Description

[Detailed description of the invention]

0001

[Field of Industrial Application] The present invention relates to a method for extracting data on a person's face and a method for determining exposure amount, and more specifically, the present invention relates to a method for extracting data on a person's face and a method for determining the exposure amount. The present invention relates to a method for extracting facial density data and a method for determining exposure amount using this method.

0002

[Prior Art and Problems to be Solved by the Invention] When viewing a photograph of a person, the part that attracts the most attention is the person's face, and in order to create a high-quality photograph, it is necessary to adjust the color of the person's face appropriately. It needs to be printed in a certain color.

Conventionally, the density data of a person's face is extracted by specifying the face area in the original image of a color film with a light pen, and the color of the face is printed appropriately based on this extracted density data. The amount of exposure is determined. Such techniques include Japanese Patent Application Laid-Open Nos. 62-115430, 1987-115431, and 62-1154.
32, JP 62-189456, JP 62-189457, JP 63-138340
There are those described in Japanese Patent Application Laid-open No. 178222/1983.

However, the conventional technique described above has a problem in that the printing process takes time because the operator must specify the facial area with a light pen for each image. Furthermore, since the operator must visually specify the facial area, unmanned operation is difficult.

[0005] Also, Japanese Patent Application Laid-Open No. 52-156624,
JP-A-52-156625, JP-A-53-123
30, JP 53-145620, JP 53-145621, JP 53-145622
The publication describes the following method for extracting data on a person's face by extracting skin color data. In other words, the color original image is divided into a large number of photometric points, each photometric point is separated into three colors of R (red), G (green), and B (blue), and each photometric point is calculated from the photometric data. Determine whether the color is within the skin color range. Then, a cluster (group) of photometric points determined to be within the skin color range is used as face density data. However, this method assumes that colors within the skin color range are the facial density data, so parts other than the face that have skin colors or colors similar to the skin color, such as the ground, tree trunks, and clothes, can also be used as facial density data. It will be extracted. Further, even when the same subject is photographed under the same conditions, the color tone of the photographed image differs depending on the film type, so it may not be possible to automatically extract facial density data if the film type is different. Furthermore, if the color of the light source illuminating the subject differs, the color tone of the photographed image will differ (
For example, an image taken using fluorescent light as a light source will have a greenish tinge), so if the light source color is different, facial density data may not be automatically extracted.

[0006] In order to solve the above-mentioned problems caused by different light source colors, it is sufficient to perform light source color correction and then extract photometric data in the skin color range. As a light source,
Light can be broadly classified into sunlight, fluorescent light, and tungsten light, but the color of sunlight varies depending on the season and time of day, and even within the same season and time of day, the color varies depending on whether it is direct or indirect light. Furthermore, artificial light such as fluorescent lamps comes in a variety of colors as products become more diverse. Therefore, it is difficult to specify the type of light source for each light source and perform light source correction. Furthermore, even if the light source correction could be performed perfectly, it would not be possible to prevent the extraction of skin tones or areas similar to skin tones, such as the ground or tree trunks, and furthermore, it would be difficult to deal with the situation when the film type is different. Can not.

The present invention has been made to solve the above-mentioned problems, and is a method for automatically extracting only human face data with high accuracy from a color original image such as a negative film. The purpose of this invention is to provide a data extraction method and an exposure amount determination method using this method.

[0008]

[Means for Solving the Problems] In order to achieve the above object, the invention of claim 1 divides an original color image into a large number of pixels and separates each pixel into three colors of red light, green light, and blue light. Measure the light, calculate a histogram of hue values based on the data obtained by photometry, divide the obtained histogram into peaks, judge which of the divided peaks each pixel of the original color image belongs to, and At the same time, the color original image is divided into groups corresponding to the divided mountains, and the color original image is divided into each group, and at least one of the outline and internal structure of each divided area is judged to determine whether it is a human face or not. Then, the data of the area determined to be a person's face is extracted.

In addition, the invention of claim 2 divides the color original image into a large number of pixels, separates each pixel into three colors of red light, green light, and blue light, performs photometry, and performs photometry based on the data obtained by photometry. A two-dimensional histogram of hue and saturation values is obtained using the method, the obtained two-dimensional histogram is divided into peaks, and pixels are divided by determining which of the divided peaks each pixel of the color original image belongs to. In addition to dividing into groups corresponding to the mountains,
The color original image is divided into each group, and at least one of the outline and internal structure of each divided area is judged to determine whether it is a human face or not, and the data of the area determined to be a human face is collected. Extract.

According to the third aspect of the present invention, the amount of exposure to the copying material is determined based on the density data of the person's face extracted as described above.

[0011]

[Operation] In the invention as claimed in claim 1, a color original image is divided into a large number of pixels, each pixel is separated into three colors of red light, green light, and blue light, and photometry is performed, and the hue is determined based on the data obtained by photometry. Find a histogram of values. Next, the obtained histogram is divided into peaks using the valleys or bottoms of the histogram as boundaries. This determines the hue value range of each mountain. Next, by determining which hue value range the hue value of each pixel belongs to, it is determined which of the divided mountains each pixel belongs to, and a large number of pixels are divided into groups ( cluster). Subsequently, the original color image is divided into regions corresponding to the divided groups. At this time, pixels included in the same group may be divided into different areas, but pixels included in different groups are never included in the same area. As a result, the color original image is divided into regions containing pixels having hue values within the hue value range divided by the histogram. Therefore, one region on the color original image includes pixels whose hue values are within a predetermined range, and the outline of the person's face and the outlines of other parts, the internal structure of the person's face, and other Since it is clearly different from the internal structure of the body part, it is possible to determine whether or not it is a human face by determining at least one of the outline and internal structure of each region, and the data of the region determined to be a human face can be determined by determining at least one of the outline and internal structure of each region. By extracting , it is possible to extract data on a person's face.

Due to changes in the type of film or light source, changes over time, differences in film development, etc., the color tone of the original color image changes uniformly across the entire screen, but even if the color tone changes in this way, the histogram does not change. The group formed by each pixel of the image is preserved by simply changing the position of the image, so the divided regions of the original color image do not change even if the color tone changes. Therefore, in the present invention,
Density data of a person's face can be extracted even if the color tone or color range of the original color image changes due to changes in film type or light source type, changes over time, differences in film development, etc.

[0013] If the hue of a person's face, which is a characteristic part of the image, is the same or similar to the hue of other parts, dividing the original color image based on the histogram of only the hue values will separate the person's face from the other parts. It may be difficult to distinguish between the two parts. Therefore, in the invention of claim 2, a chroma value is further introduced in addition to the hue value, a two-dimensional histogram of the hue value and the chroma value is obtained, this two-dimensional histogram is divided into peaks, and coloring is performed in the same manner as above. The original image is divided, and at least one of the outline and internal structure of the divided regions is determined to extract human face data.

In the present invention, since the hue value and the saturation value are used, even if there are parts (for example, the ground, trees, etc.) that have the same or similar hue as the face of the person, data on the face of the person can be obtained. can be extracted. In other words, the hue of a person's face is similar to the skin-colored parts of the ground, trees, etc., but in most cases the saturation is different, so the hue of a person's face can be calculated based on a two-dimensional histogram of hue and saturation values. By extracting data, it is possible to extract data on a person's face even from an image containing a mixture of faces, the ground, trees, etc.

[0015]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. In this embodiment, the present invention is applied to an auto printer. As shown in FIG. 1, the automatic printer of this embodiment includes a conveyance roller 12 that conveys a color negative film 10. As shown in FIG. A light source 14 is located below the color negative film 10 being transported by the transport roller 12.
, a color correction filter 16 such as a dimmer filter, and a diffusion box 18 are arranged in this order. Also, negative film 1
A distribution prism 20 that distributes the light beam transmitted through the negative film 10 into two directions is arranged above the negative film 10 . On one optical path distributed by the distribution prism 20, a projection optical system 22, a black shutter 23, and color paper (photographic paper) 24 are arranged in order, and on the other optical path, a projection optical system 26 and a CCD image sensor are arranged. 28 are arranged in order. This CCD image sensor 28 is
The entire screen (one frame) of the negative film 10 is divided into a large number of pixels (for example, 256 x 256 pixels), and each pixel is divided into R (
The light is measured by separating it into three colors: red), G (green), and B (blue). The CCD image sensor 28 is connected to an amplifier 30 that amplifies the CCD image sensor output and an analog-to-digital (
It is connected via an A/D converter 32 to a 3×3 matrix circuit 34 for sensitivity correction of the CCD image sensor. The 3×3 matrix circuit 34 is connected to an appropriate exposure calculation circuit 40 via a face extraction circuit 36 configured with a microcomputer that stores a routine program to be described below, and calculates the average density of the entire screen. It is connected to an appropriate exposure amount calculation circuit 40 via an average density calculation circuit 38 that calculates it. Then, the appropriate exposure amount calculation circuit 4
0 is connected to the color correction filter 16 via a driver 42 that drives the color correction filter.

Next, the operation of this embodiment will be explained. light source 14
The light rays emitted from the CCD image sensor 28 pass through the color correction filter 16, the diffusion box 18, and the color negative film 10, are distributed by the distribution prism 20, and are received by the CCD image sensor 28 via the projection optical system 26. Note that at this time, the black shirt door 23 is closed. Upon receiving this light, the CCD image sensor 28 divides the entire screen into a large number of pixels, separates each pixel into three colors of R, G, and B, performs photometry, and outputs a photometry data signal. The photometric data signal is amplified by an amplifier 30 and then converted to a digital signal by an A/D converter 32. Sensitivity correction of the image sensor is performed by a 3×3 matrix circuit 34, and a face extraction circuit 36 and an average density calculation circuit 38 is input. This average concentration calculation circuit 3
In step 8, the average density of the entire one screen is calculated. The face extraction circuit 36 estimates the part of a person's face in one screen as described below, and outputs R, G, and B three-color photometric data of the part estimated to be the face. The exposure amount calculation circuit 40 calculates the exposure amount using the three-color photometric data output from the face extraction circuit 36 and the average density determined by the average density calculation circuit 38, and applies the exposure amount to the color correction filter 16 via the driver 42. At the same time, the black shutter 23 is opened and closed to perform printing. Note that when the average density determined by the average density calculation circuit 38 is used, the exposure correction amount for the average density can be determined. When the exposure correction amount is not determined, the average density calculation circuit 38 is not necessarily required, and the exposure amount may be determined directly from the three-color photometric data output from the face extraction circuit 36.

FIG. 2 shows a face extraction routine by the face extraction circuit 36, in which noise removal, ie, smoothing, is performed on the input three-color photometric data in step 100. In the next step 102, R, G, and B three color photometric data are converted into H (hue value) and L using the following equations (1) to (3).
(brightness value) and S (chroma value).

L=(R+G+B)/3
...(1) S=1-
min(r', g', b')...(2)
H=H'/2Pi
...(3) However, R, G, and B are 3 in Figure 3.
Three-color photometric data, min( ), standardized so that the minimum value is 0 and the maximum value is 1, as shown in the dimensional color coordinates.
is the minimum value of the numbers in parentheses, r', g', b' are r'
=R/L, g'=G/L, b'=B/L. Also H
' is given by the following equation (4), where Pi (i is R, G, B
One of them) is P in FIG.

[0019]

[Math 1]

However,

[0021]

[Math 2]

In step 104, as shown in FIG. 4(1), a two-dimensional histogram of hue values and saturation values is created using a coordinate system consisting of a hue value axis, a saturation value axis, and a pixel number axis that are orthogonal to each other. is obtained, and in the next step 106, the obtained two-dimensional histogram is divided into peaks, that is, the two-dimensional histogram is clustered, as will be described later. In the next step 108, the clustered 2
A large number of pixels are clustered based on the peaks of the dimensional histogram, the screen is divided based on this clustering, and areas that are candidates for human faces are extracted from the divided areas. In the next step 110, a face area is estimated from the area extracted as a face candidate, and R, G, and B three-color photometric data of the area estimated as a face is output. Then, in step 112, it is determined whether or not the printing of all frames has been completed, and when it is determined that the printing has been completed, this routine is terminated.

Next, the details of steps 106 to 110 will be explained. FIG. 5 shows the details of step 106. In step 120, a region to be evaluated is cut out from a two-dimensional histogram of hue values and saturation values. In FIG. 4, one frame is used as an evaluation area for the sake of simplicity. In step 122, it is determined whether there is an evaluation area. When an evaluation area cannot be cut out in step 120, that is, when evaluation of all areas has been completed, there is no evaluation area, so this routine is ended. If there is an evaluation region, the X and Y axes for creating a histogram for mountain extraction are determined in step 124. In other words, the evaluation region is rotated around an axis parallel to the pixel number axis, and when the peaks of the histogram are viewed from the side, the position where multimodality is prioritized and the peaks are the sharpest is determined, and this position is used as a reference. X,
Determine the Y axis. If it is necessary to shorten the processing time, the axes with the maximum histogram variance may be used as the X and Y axes, although the accuracy will be somewhat degraded. In the example of Figure 4 (1),
When the four peaks numbered 1 to 4 are viewed from the side, multimodality is given priority, and the position where the peaks are the sharpest is the position where the three peaks are visible, so the X-axis is aligned perpendicular to the viewing direction. determine,
A Y-axis is defined in a direction perpendicular to this X-axis.

In the next step 126, the two-dimensional histogram is projected onto the X and Y axes to create one-dimensional histograms. In the example of FIG. 4 (1), when viewed from the direction orthogonal to the X-axis, the peaks labeled 1 and 2 appear to overlap, so the one-dimensional histogram for the X-axis includes the peaks labeled 3 and the peaks labeled 2. Three peaks appear, one labeled 1, 2, and one labeled 4, and when viewed from the direction perpendicular to the Y-axis, the peaks labeled 1 to 4 appear to overlap, so it is one-dimensional about the Y-axis. One mountain appears in the histogram. In the next step 128, the histogram is converted into an evaluation function H(a) using the following equation (5), and a peak is extracted from the histogram on the X axis based on this evaluation function.

[0025]

[Math 3]

Here, f(a) is the number of pixels when the value (feature amount) in the X-axis direction is a, and x is the displacement from the feature amount a.

That is, the average value T of the evaluation function H(a) is determined, and the range below the average value T of the evaluation function H(a) (the range in which valleys and tails exist) is determined. Next, the position where the histogram is minimum within this range is defined as the valley or tail of the histogram. Then, the histogram is cut out at the determined valley or tail.

The extraction of the above-mentioned mountain will be explained with reference to FIG. 6. When the evaluation function H(a) is determined from the histogram represented by the solid line SI, it becomes as shown by the broken line in the figure. The range in which this evaluation function H(a) is less than or equal to the average value T regarding the negative portion is the range in which the feature amounts are v0 to v1 and v2 to v3. The position of the minimum frequency of the histogram within this range is the range v
av0=v0 for 0 to v1, av1 for range v2 to v3
, av0 is determined as the hem, and av2 is determined as the valley, and the histogram is cut out at these positions.

In step 130, the histogram peaks on the Y axis are extracted in the same manner as the histogram peaks on the X axis. Next step 132
Now, an area where the peaks of the one-dimensional histograms for the X-axis and Y-axis cut out as described above overlap on the two-dimensional histogram is determined, and the peaks are cut out from the two-dimensional histogram for the hue value and the saturation value. Area E in Figure 4 (1)
1 shows an example of a mountain cut out as described above.

In the next step 134, it is determined whether the peak extracted from the two-dimensional histogram is a single peak or not. If it is not a single peak, steps 124 to 134 are performed until the peak extracted from the two-dimensional histogram becomes a single peak. repeat. Region E2 in FIG. 4(3) shows an example of a single peak cut out as described above.

[0031] In the next step 136, a process (labeling) for identifying the extracted single peak is performed, and in step 138, the labeled peak is masked and the process returns to step 120. Then, by repeating the above steps, the entire area of the two-dimensional histogram for hue and saturation values is divided into single peaks.

FIG. 7 shows the details of step 108 in FIG. 2. In step 140, the range XR (FIG. 4 (3)) and Y The range YR in the axial direction (Fig. 4 (3)) is determined for each single peak, and the clustering of pixels is performed by determining whether the hue and saturation values of each pixel in the original image belong to these ranges. At the same time, the pixels belonging to the range surrounded by ranges XR and YR are collected, and the original image is divided so that the collected pixels form one area on the original image. Also, the divided areas are numbered. Figure 4 (2) shows an example of dividing the original image, and the pixels in each area labeled 1 to 4 are included in the single peak labeled 1 to 4 in Figure 4 (1). It corresponds to the pixels that are displayed. Pixels that belong to the same single peak in Figure 4 (1) are divided into different regions in Figure 4 (2), but this is due to the hue value range and the single peak peak in Figure 4 (1). This is because although the pixel has a saturation value range, the regions are divided in FIG. 4(2).

In the next step 142, small areas are removed by determining the area of the divided areas, and numbering is performed again. In the next step 144, a contraction process is performed to remove all boundary pixels of the area by one skin, and an expansion process is performed to increase the boundary pixels in the direction of background pixels and thicken the area by one skin, which is the opposite of the contraction process. Separate small areas that are overgrown from large areas. In the next step 146, small areas are removed and renumbered in the same way as in step 142, and in step 148, the same contraction and expansion processes as above are performed to separate weakly connected areas. In step 150, small areas are removed and renumbered in the same manner as described above.

FIG. 8 shows the details of step 110. In step 162, one region is selected as the region of interest from among the regions extracted in step 108, that is, the routine of FIG. 7, and the horizontal fillet diameter of the region of interest is determined. Then, the size of the attention area is standardized by enlarging/reducing the attention area so that the vertical fillet diameter becomes a predetermined value, and the density value or brightness value is also standardized according to the following equation (6).

[0035]

[Math 4]

[0036] However, dmax: maximum density value (or brightness value) within the area dmi
n: Minimum density value (or brightness value) within the area ds:
Full-scale density value (or brightness value) of the image sensor d: Density value (or brightness value) before normalization dr
: Normalized density value (or brightness value) In step 164, standard face images (face images seen from the front, face images seen from the side) of multiple types (10 types in this embodiment) stored in advance are The correlation coefficient r of the attention area for the left and right), downward face image, upward face image, etc.) is calculated using the following equation (7),
This correlation coefficient is used as a feature quantity. This standard facial image is
The data may be data of only the contour of the face, or data of the internal structure of the face (eyes, nose, mouth, etc.) added to the data of the facial contour.

[0037]

[Math 5]

However,

[0039]

[Math 6]

where T is the length of the horizontal and vertical fillet diameters of the image (here, the lengths of the fillet diameters are the same), f(
x, y) represents the region of interest, and g(x, y) represents a standard face image.

[0041] Then, in step 166, it is determined whether the region of interest is a person's face by linear discriminant analysis using the above feature quantities as variables, and the R of the region determined to be a face is determined.
, G, and B photometric data are output to the appropriate exposure calculation circuit 40. In step 168, it is determined whether or not the determination of whether or not the entire extracted area is a face has been completed, and if the determination has not been completed, steps 162 to 168 are repeated.

In the above, the correlation coefficient was used as the feature quantity used to determine whether it is a human face or not, but the invariant derived from the normalized central moment around the center of gravity, which will be explained below, Autocorrelation functions or geometric invariants may also be used.

The central moment μpq of the image f(x, y) around the (p+q)th center of gravity is

[0044]

[Math. 7]

However,

[0046]

[Math. 8]

Then, the normalized central moment about the center of gravity is:

[0048]

[Math. 9]

However, y=(p+q+2)/2 p+q=2,3,... From the above, the following seven invariants ψi, (
i=1, 2, ..., 7) are derived.

[0050]

[Math. 10]

Further, the autocorrelation function Rf is expressed as follows.

[0052]

[Math. 11]

The geometrically invariant feature quantity is expressed by the following equation.

[0054]

[Math. 12]

The appropriate exposure amount calculation circuit 40 is the face extraction circuit 3.
R, G, B photometric data of the face area extracted as described above in step 6 and the screen average density Di of one frame calculated by the average density calculation circuit 38 (i = any one of R, G, B) Calculate the appropriate exposure amount Ei according to the following formula using
Output to driver 42. The driver 42 sets the appropriate exposure amount E.
Calculate the exposure control value from i and apply it to the light control filter 1.
Control 6.

logEi =LMi ・CSi ・(DNi −
Di ) + PBi + LBi + MBi
+NBi +K1 +K2 (8) However, each symbol represents the following.

LM: Magnification slope coefficient, which is set in advance according to the enlargement magnification determined by the type of negative and the print size.

CS: A color slope coefficient prepared for each type of negative, with one for underexposure and one for overexposure, and determines whether the average density of the frames to be printed is under or over the standard negative density value. Either underexposure or overexposure is selected.

DN: Standard negative density value. D: Average density value of print frames.

PB: A correction balance value for standard color paper, which is determined depending on the type of color paper.

LB: Relative to standard baked lens. This is a correction lens balance value, which is determined depending on the type of lens with which it is baked.

MB: Correction value (master balance value) for variations in print light source and paper development performance.

NB: Negative balance (color balance) value determined by the characteristics of negative film.

K2: Color correction amount. K1: Density correction amount expressed by the following formula.

[0065]

[Math. 13]

Here, Ka and Kb are constants, and F
D is the face area average density.

Alternatively, the density correction amount K1 in the above equation (8) may be the correction value determined by the film verification device, and the color correction amount K2 may be expressed using the face area average density as follows.

[0068]

[Math. 14]

However, Kc is a constant.

Furthermore, the density correction amount K1 of the above equation (8),
Let the color correction amount K2 be the correction amount determined by the film verification device, and the average density D of the print frames in equation (8)
The exposure amount may be determined by replacing i with the average density FDi of the face area.

[0071] In this embodiment, since judgment is made using the outline and internal structure of the region, face data can be extracted even from an image containing a mixture of faces, ground, trees, etc. that have similar hues. .

FIG. 9 shows a modification in which the present invention is applied to an exposure amount determination device separate from the printer or printer processor. Note that in FIG. 9, parts corresponding to those in FIG. 1 are designated by the same reference numerals, and description thereof will be omitted. Furthermore, although the average density calculation circuit 38 is not necessarily required, an integrated transmission density detection circuit that detects the LATD of the entire screen may be used instead.

In FIG. 10, the face extraction circuit of FIG. 9 is constructed of a plurality of face extraction circuits 361, 362, . . . , 36n, and the exposure amount is calculated by parallel processing. Face extraction circuit 3
61, 362, . . . 36n read the image according to the time chart in FIG. 11, calculate the exposure amount, and output the result. In FIG. 11, t1 is the image reading time for one frame, t2 is the exposure calculation time for one frame, t3 is the transfer time of the exposure calculation result for one frame, and t2 >> t1
, t3. The face extraction circuit 361 reads one frame image at time t1, calculates the exposure amount at time t2, and transfers the calculation result at time t3. At the same time that the face extraction circuit 361 finishes reading one frame of the image, the film is advanced by one frame, and the face extraction circuit 362 starts reading one frame of the image, and the face extraction circuit 361 calculates the exposure amount and the face extraction circuit 362
image reading is performed in parallel, and the subsequent processing is similarly performed in parallel by the face extraction circuits 363, 364, . . . , 36n.

The time Tp required to process mxn frames in parallel is Tp=m(t1 +t2 +t3)+(n-1)t1
It is. On the other hand, the processing time T when parallel processing is not performed
s is Ts=m·n(t1 +t2 +t3). Therefore,

[0075]

[Math. 15]

[0076] Double speeding up is possible.

Note that this parallel processing device can also be applied to the printer shown in FIG. In addition to determining the exposure amount of a photo printing device, the present invention also determines the exposure amount of a digital color printer, the copying conditions of a copier, the exposure amount of a camera, the display conditions of a CRT screen, and the creation of hard copies from magnetic image data. It can also be applied to determining the amount of light when

[0078]

As described above, according to the present invention, at least one of the outline and internal structure of a divided area is judged based on a hue value histogram to determine whether or not it is a human face. Therefore, even if the color tone and color range of the original color image change due to changes in film type and light source type, changes in film characteristics over time, differences in film development, etc., data on human faces can be extracted with high accuracy. This effect can be obtained.

Furthermore, since it is determined whether or not it is a human face by determining at least one of the outline and internal structure of the divided area based on the two-dimensional histogram of hue values and saturation values, it is possible to It is possible to extract data on a person's face even if there are parts of the face that are the same or similar in hue.
This effect can be obtained.

[Brief explanation of drawings]

FIG. 1 is a schematic diagram showing a printer according to one embodiment of the present invention.

FIG. 2 is a flowchart showing a face extraction routine of a face extraction circuit.

FIG. 3 is a diagram showing color coordinates.

FIG. 4 (1) is a diagram showing a two-dimensional histogram of hue values and saturation values. (2) is a diagram showing a state in which the original image is divided. (3) is a diagram showing a state in which a single peak is cut out from a two-dimensional histogram.

FIG. 5 is a diagram showing details of step 106 of FIG. 2;

FIG. 6 is a diagram showing a histogram and an evaluation function.

7 is a diagram showing details of step 108 of FIG. 2; FIG.

FIG. 8 is a diagram showing details of step 110 of FIG. 2;

FIG. 9 is a schematic diagram of an exposure calculation device to which the present invention is applied.

FIG. 10 is a schematic diagram of an exposure calculation device that performs parallel processing using a plurality of face extraction circuits.

FIG. 11 is a diagram showing a time chart of parallel processing.

[Explanation of symbols]

28 CCD image sensor 30
Amplifier 36 Face extraction circuit

Claims (3)

[Claims] [Claim 1] Divide a color original image into a large number of pixels, separate each pixel into three colors of red light, green light, and blue light, perform photometry, and obtain a histogram of hue values based on the data obtained by photometry. , divide the obtained histogram into peaks, determine which peak each pixel of the color original image belongs to, and divide the pixels into groups corresponding to the peaks.
The color original image is divided into each group, and at least one of the outline and internal structure of each divided area is judged to determine whether it is a human face or not, and the data of the area determined to be a human face is collected. A method for extracting human face data. 2. Divide the color original image into a large number of pixels, separate each pixel into three colors of red light, green light, and blue light, perform photometry, and calculate hue and saturation values based on the data obtained by photometry. Find a two-dimensional histogram of At the same time, the color original image is divided into each group, and at least one of the outline and internal structure of each divided area is judged to determine whether or not it is a human face. A human face data extraction method that extracts region data. 3. An exposure amount determining method for determining the amount of exposure to a copying material based on the extracted human face data according to claim 1, 2, or 3.