JP3516786B2 - Face area extraction method and copy condition determination method - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a face area extracting method and a copy condition determining method, and more particularly to a face area extracting method for extracting an area estimated to correspond to a human face in a color original image, and the extracting method. The present invention relates to a copying condition determining method for determining an exposure amount based on an image feature amount in a face area extracted by the method.

[0002]

2. Description of the Related Art Since the face of a person is the most noticeable part when viewing a photograph of a person, for example, an original image recorded on a negative film or the like is used as a copying material such as photographic paper. When printing on, the area corresponding to the person's face in the original image is extracted so that the color of the person's face is printed in an appropriate color, and the exposure amount is based on the color and density of the extracted area. Is preferably determined.

Therefore, JP-A-52-156624, JP-A-52-156625, JP-A-53-12330, and JP-A-5-156624
3-145620, JP-A-53-145621, JP-A-53-
No. 145622 discloses a method of extracting data of a person's face by extracting skin color data by dividing a color original image into a large number of measurement points, and measuring each measurement point with R, G,
A cluster (group) of measurement points determined to be within the skin color range by determining whether the color of each measurement point is within the skin color range on the color coordinates based on the photometric data Is described as the face density. However, in the above, when the original image has a skin color area such as the ground or a tree trunk, the data of the measurement points in this area may be erroneously extracted as the density data of the face, and the appropriate exposure amount There was a problem that I could not get.

Further, the applicant of the present invention divides a color original image into a large number of pixels, decomposes each pixel into three colors of R, G, and B to perform photometry, and based on the photometric data, hue value (and saturation value). For each mountain, determine which pixel the divided mountain belongs to, and divide each pixel into groups corresponding to the divided mountains. It is proposed to divide an image into a plurality of areas (so-called clustering), estimate an area corresponding to a person's face in the plurality of areas, and determine the exposure amount based on the photometric data of the estimated area. (See JP-A-4-346333).

However, when the skin-colored non-human region such as the ground or the trunk of a tree existing in the original image is adjacent to the region corresponding to the human face, ideally, as an example, FIG. As indicated by different hatching in FIG. 9, it is preferable to extract as a region different from a region corresponding to a person's face and a non-human region (a region corresponding to the background in the figure).
As shown in (B), the area corresponding to the human face and the non-person area cannot be separated. If the hue and saturation in the area corresponding to the person's face are slightly different, a plurality of clusters are formed on the histogram by the data in the area corresponding to the person's face. (C)
As shown in, the area corresponding to a person's face may be excessively divided.

In the above technique, the processing is performed on the assumption that one of the divided areas is an area corresponding to the face of a person. Therefore, when the original image cannot be divided into areas within an appropriate range, , The area corresponding to the person's face is misjudged,
There is a problem that it is not possible to obtain an exposure amount that can properly print a person's face.

The present invention has been made in consideration of the above facts, and provides a face area extraction method capable of appropriately and accurately extracting an area corresponding to a human face existing in a color original image. Is the purpose.

It is another object of the present invention to obtain a copying condition determining method which can obtain with high accuracy a copying condition capable of properly printing an area corresponding to the face of a person existing in a color original image.

[0009]

In order to achieve the above object, a face area extracting method according to the invention of claim 1 separates a large number of points on a color original image into a plurality of component colors and performs photometry,
It is determined whether the hue at each position on the original image represented by the data obtained by the photometry is included in the skin color range, the original image is divided into a skin color region and a non-skin color region, and the light is obtained by the photometry. Detecting edges in the original image based on the data obtained, classifying each part in the original image into an edge part or a non-edge part, located in the skin color area, and classified as a non-edge part A region in the original image composed of a set of is extracted as a face candidate region, and it is determined whether the extracted face candidate region is a region corresponding to a human face,
A face candidate area determined to be an area corresponding to a person's face is extracted.

In the above, the original image is divided into the flesh color region and the non-flesh color region by judging whether the hue at each location on the original image is included in the flesh color range or not, so that it corresponds to a human face. Even if the hue and saturation of the region to be partially shaded are slightly different, it is very likely that the hues at each point in the region are all included in the skin color range. The original image is divided into the flesh-colored area. Therefore, even if the hue or saturation in the area corresponding to the face of the person is partially different, this portion will not be excessively divided.

Further, a group of places which are located in the skin color area and are classified as non-edge parts by detecting edges in the original image and classifying each part in the original image into edge parts or non-edge parts Since a region in the original image composed of is extracted as a face candidate region, for example, in the original image, if a background region having the same hue or a similar hue as the human face is adjacent to the region corresponding to the human face. Also, since the boundary portion between the background area and the area corresponding to the person's face is classified as the edge portion, the background area and the area corresponding to the person's face are divided by the edge portion, and both are different. It will be extracted as a face candidate area.

Then, it is determined whether or not the face candidate region extracted as described above is a region corresponding to a human face, and a face candidate region determined to be a region corresponding to a human face is extracted. Therefore, the area corresponding to the face of the person existing in the color original image can be appropriately extracted with high accuracy without excessive division or mixture of background areas.

In the copy condition determining method according to the invention of claim 2, the face area extracting method according to claim 1 extracts and extracts an area corresponding to a human face from the original image recorded on the recording medium. It is characterized in that the copy condition of the original image on the copy material is determined based on the image feature amount in the area.

In the above, since the face area extracting method according to claim 1 is applied to extract the area corresponding to the human face from the original image recorded on the recording medium, like the above, the area corresponding to the human face is extracted. Area can be extracted accurately with high accuracy,
Further, since the copy condition for the copy material is determined based on the image feature amount in the extracted region, it is possible to obtain with high accuracy the copy condition capable of properly printing the region corresponding to the face of the person existing in the color original image. You can

[0015]

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, an example of an embodiment of the present invention will be described in detail with reference to the drawings. FIG. 1 shows a photographic printing apparatus 10 to which the present invention can be applied. The photo printing apparatus 10 includes a light source 16 that emits exposure light for printing an image recorded on the negative film 12. Light source 1
The color correction filter 1 such as a dimming filter is provided on the light emission side of 6.
8, the diffusion box 20, and the distribution prism 22 are arranged in order.

The conveying path of the negative film 12 as a photographic film is formed between the diffusion box 20 and the distribution prism 22, and the distribution prism 22 distributes the light transmitted through the negative film 12 in two directions. Conveying roller pairs 14A and 14B are provided on both sides of the optical axis of the light source 16 along the conveying path of the negative film 12. The conveying roller pairs 14A and 14B are connected to the drive shafts of the motors 52A and 52B, respectively, and are rotated by the driving force of the motors 52A and 52B being transmitted to convey the negative film 12.

A projection optical system 24, a black shutter 26, and a color paper (printing paper) 28 as a copying material are sequentially arranged on one optical path of the light distributed in two directions by the distribution prism 22. The projection optical system 30 and the CCD image sensor 32 are sequentially arranged on the other optical path. The CCD image sensor 32 divides the entire image (one frame) recorded on the negative film 12 into a large number of pixels (for example, 256 × 256 pixels), and each pixel is red.
(Red), G (green), and B (blue) are separated into three colors for photometry.

An amplifier 34 for amplifying the signal output from the CCD image sensor 32, an analog-digital (A / D) converter 36, and a sensitivity correction of the CCD image sensor 32 are provided at the signal output end of the CCD image sensor 32. 3x
Three matrix circuits 38 are connected in order. The 3 × 3 matrix circuit 38 is connected to the input / output port 40D of the control unit 40 including a microcomputer and its peripheral devices. The control unit 40 includes a CPU 40A and a ROM
40B, RAM 40C and input / output port 40D are provided, and these are connected to each other via a bus.

The color correction filter 18 is connected to the input / output port 40D of the control section 40 via a driver 46 for driving the color correction filter 18, and the drivers 50A,
Motors 52A and 52B are connected to each other via 0B. Further, the input / output port 40D is provided with a display 42 as a display means composed of an LCD or a CRT, a keyboard 44 such as a numeric keypad for an operator to input various information, and negatives arranged on both sides of the optical axis of the light source 16 therebetween. A screen detection sensor 48 that detects the amount of transmitted light of the film 12 is connected.

Next, the operation of this embodiment will be described with reference to the flowchart of FIG. The flow chart of FIG. 2 shows the negative film 12 set in the photo printing apparatus 10.
Is carried out in a predetermined direction, the output signal from the screen detection sensor 48 is monitored, and this is executed every time the image recorded on the negative film 12 is positioned at the exposure position.

In step 100, the image positioned at the exposure position is measured by the CCD image sensor 32, and C
The R, G, and B photometric data output from the CD image sensor 32 via the amplifier 34, the A / D converter 36, and the 3 × 3 matrix circuit 38 is captured, and the captured photometric data is stored in a memory such as a RAM 40C. Remember once. In step 102, noise removal, that is, smoothing, of the photometric data stored in the memory is performed. In the next step 104, skin color area extraction processing is performed. This skin color area extraction processing will be described with reference to the flowchart in FIG.

In step 130, it is determined whether or not the hue of each pixel is included in the skin color range based on the photometric data for each pixel that has been smoothed as described above. For this determination, for example, an ellipsoidal solid described in JP-A-52-156624 can be used.

That is, the green density G at a large number of skin color points extracted from a large number of images recorded on a negative film,
The blue density B and red density R are the average values B _AVE, G of each color density.
_Since it has a three-dimensional Gaussian distribution centered on _{AVE and} R _AVE ,
The color densities B, G, R of the skin color point are centered on the average values B _AVE, G _AVE, R _AVE of the color densities defined on the three-dimensional color coordinates with the color densities B, G, R as coordinate axes. Ellipsoid (Ellipsoid 6 in FIG. 5, which is an ellipse because the distribution of each color is not equal)
(See 0)). The ellipsoid or ellipse is represented by the following general formula.

[0024]

[Equation 1]

However, α _ij is a constant, x ₁ = B−B _AVE , x
₂ = G-G _AVE, a _{x 3 = R-R AVE,} n is 3 if ellipsoid. Therefore, in the case of an ellipsoid, it can be set as dS ² = C ₁₁ dB ² + 2C ₁₂ dB · dG + C ₂₂ dG ² + 2C ₂₃ dG · dR + C ₃₃ dR ² + 2C ₃₁ dR · dB (2). dB = x ₁ , dG = x ² , dR =
because it is _{x 3, dB = B-B} AVE, dG = G-
G _AVE and dR = R−R _AVE . C _ij is an element of the inverse matrix of the following covariance / covariance matrix A.

[0026]

[Equation 2]

However, N represents the number of data, and B _i, G _{i, and} R _i represent the density at each measurement point. By the way, the size of the ellipsoid defined by the equation (2) changes depending on the value of dS ^{2 in} the equation (2). In order to prevent the data of measurement points whose hue is not skin color from being included in the ellipsoid, the majority of the data of many skin color points extracted from many images recorded on negative film (eg 95%) It is sufficient to set the threshold value of dS ² so that is included in the ellipsoid and obtain the above parameters.

Then, at step 130, the data of each pixel is substituted into the equation (2) to calculate dS ² , and the obtained dS
^If the value of ² is less than or equal to the previously set threshold value, it is determined that the hue of the pixel is included in the skin color range because the data of the pixel is located in the ellipsoid. When the value of dS ² is larger than the threshold value, the data of the pixel is located outside the ellipsoid, and thus it can be determined that the hue of the pixel is outside the skin color range.

The distribution of the skin color points on the three-dimensional color coordinates changes according to the type of the light source when the image of the skin color point extraction source is photographed. As an example, in FIG. 5, an ellipsoid obtained from data of skin color points extracted from an image captured using a fluorescent lamp as a light source is shown as an ellipsoid 62, and a skin color extracted from an image captured using a tungsten lamp as a light source. The ellipsoid obtained from the point data is shown as an ellipsoid 64. Therefore, the values of dS ² and the parameters of the equation (2) are obtained for each type of light source, and it is determined whether the hue of each pixel is included in the skin color range defined by each ellipsoid. You may do it. Further, instead of the ellipsoid described above, a polygonal area shown in FIG. 3 of JP-A-53-145620 may be used.

In step 132, based on the determination result in step 130, the original image is composed of a skin color area formed only of pixels determined to be pixels within the skin color area and a pixel determined to be a pixel outside the skin color area. The non-skin color area is composed of only the non-skin color area, and a binarized binary image is generated according to the pixel in the skin color area or the pixel in the non-skin color area. The skin color area is regarded as a single skin color area, and numbering is performed to identify each of the skin color areas.

In the next step 134, minute areas are removed by judging the areas of the divided areas, and the numbering is performed again. In step 136, contraction processing is performed to delete all the boundary pixels in the skin color area to remove one skin, and conversely to the contraction processing, expansion processing is performed to expand the boundary pixels in the direction of the background pixels to enlarge the skin by one skin, thereby forming a large area. Separate large areas from large areas. In step 142, as in step 134, minute areas are removed and renumbering is performed. In the next step 140, step 136,
It is determined whether the processing of 138 has been executed a predetermined number of times (for example, twice). If the determination is negative, the procedure returns to step 136, and step 13 is performed until the determination of step 140 is positive.
Repeat steps 6 and 138. As a result, the flesh color areas that are weakly coupled are separated from each other.

When the determination at step 140 is affirmative, the routine proceeds to step 142, where the skin color area is filled by closing, which is one of the four basic morphological transformations. closing is defined by the following expression (3).

[0033]

[Equation 3]

B is a bounded set called a component. In the morphological transformation, various transformations can be obtained by changing the constituent element B. Generally closing
Is to perform smoothing from the outside by removing noise (irregular pattern determined by the component B) from the set X. Therefore, as shown in FIG. 6A, if an original image (binary image) to be processed is set as X and a regular component B is selected, the solid line is shown in FIG. 6B. As described above, the set X ′ after the morphological transformation (closing) has a smooth boundary with respect to the set X, and narrow gaps, elongated cracks, and small holes are filled.

In step 142, the skin color area is applied as the set X and the above conversion is performed to fill in the skin color area. From the above, as an example, from the original image shown in FIG. 8A (this original image is assumed to be the hue in which the hue of the area corresponding to the background is included in the skin color range as an example), The skin color area is extracted as shown in B). As is clear from FIG. 8B, in the above-described skin color region extraction processing, when the hue of the background region adjacent to the face region is a hue included in the skin color range, the face region and the background are regarded as the skin color region. A region in which the region and the region are mixed is extracted, but instead, even if the hue or saturation in the face region is slightly different, all the face regions are extracted as the same skin color region. And step 14
After the binary image as the final processing result in 2 is stored in the memory as the image A, the process proceeds to step 106 in FIG.

In step 106, edge extraction processing is performed. This edge extraction processing will be described with reference to the flowchart in FIG. In step 190, the brightness value is calculated for each pixel based on the photometric data captured in step 100 of the flowchart of FIG. 2, and brightness image data representing a brightness image is generated. In step 192, for each pixel, the direction toward each of eight surrounding pixels (so-called eight neighboring pixels) (total eight directions: shown in FIG. 7 as eight arrows different in direction at 45 ° intervals). ), The brightness change value (edge strength) is calculated using a differential filter such as Sobel. FIG. 7 shows an example of eight differential filters for calculating the edge intensities along the eight directions.

For example, when calculating the edge strength along a predetermined direction, a differential filter indicated by the arrow pointing in the predetermined direction out of the eight arrows shown in FIG. 7 is used, and the density value of the pixel to be calculated and The edge strength along the predetermined direction can be obtained by multiplying the density values of the eight pixels existing around the pixel to be calculated by multiplying the numerical values of the differential filter as coefficients and calculating the sum thereof. . By performing the above calculation using eight differential filters corresponding to each direction, the edge strength along each direction in a single pixel can be obtained. In step 192,
For each pixel, compare the edge strength along each of the above directions,
The maximum absolute value of the edge strength is stored as the edge strength of the pixel to be processed.

In step 194, the edge strength of each pixel obtained above is compared with a predetermined reference value,
If the edge strength is equal to or higher than the reference value, it is determined that the pixel corresponds to an edge portion of the image. If the edge strength is equal to or higher than the reference value, the pixel is equivalent to a non-edge portion of the image. By repeating the determination, each pixel forming the image is separated into an edge portion and a non-edge portion. Then, a binarized binary image is generated according to whether each pixel is an edge portion or a non-edge portion.

In the next step 196, the well-known thinning process for converting the edge portion of the binary image into a line of one pixel width is applied to the binary image generated above, and the obtained image is stored as the image B in the memory. Remember. As a result, as an example, an edge (shown by a thick line in the drawing) as shown in FIG. 8C is extracted from the original image shown in FIG. 8A. The edge extraction processing is completed as described above, and the process proceeds to step 108 in the flowchart of FIG.

In step 108, the data of the image A and the image B stored in the memory is taken out, and is composed of pixels which are determined to be the skin color area by the skin color area extraction processing and are determined to be the non-edge portions in the edge extraction processing. Extracted areas. Then, in step 110, the area surrounded by the pixels determined to be the edge portion is regarded as one face candidate area for the area extracted in step 108, and labeling is performed to identify the face candidate area.

As described above, even if the original image has a non-human region (for example, a region corresponding to the background) of a hue included in the skin color range adjacent to the region corresponding to the human face, A region corresponding to the face of the person and a region corresponding to the background are divided by an edge portion corresponding to the contour of the face of the person, and as shown by different hatching in FIG. The area corresponding to the face and the non-human area having the hue included in the skin color range are extracted as different face candidate areas. This makes it possible to properly extract a region corresponding to a person's face as a face candidate region.

In the next steps 112 to 118, of the labeled face candidate areas, the area which is estimated to have the highest probability of being the area corresponding to the human face is determined. That is, in step 112, one is extracted from the face candidate areas extracted in the previous step 108, and in step 114, the area is adjusted so that the horizontal fillet diameter and the vertical fillet diameter of the extracted face candidate areas become predetermined values. Is performed to standardize the size of the area, and also to standardize the density value or the brightness value according to the following equation (4).

[0043]

[Equation 4]

However, dmax: maximum density value (or brightness value) in the area dmin: minimum density value (or brightness value) in the area ds: full scale density value (or brightness value) of the image sensor d: density value before normalization (Or luminance value) dr: normalized density value (or luminance value) In step 116, a plurality of types (10 types in the present embodiment) of standard face images stored in advance (face image viewed from the front, horizontal direction) The correlation coefficient r of the attention area with respect to the face image (left and right), the downward face image, the upward face image, etc. seen from
The correlation coefficient is calculated by using an equation and is used as a feature amount. This standard face image, even if only the data of the face outline,
It may be data obtained by adding the data of the internal structure (eyes, nose, mouth, etc.) of the face to the data of the outline of the face.

[0045]

[Equation 5]

Where T is the horizontal and vertical fillet diameters of the image (here, the fillet diameters are the same), f
(X, y) represents a face candidate area, and g (x, y) represents a standard face image.

At the next step 118, it is determined whether or not the face candidate region is a region corresponding to a person's face by a linear discriminant analysis using the above-mentioned characteristic amount as a variable, and the determination result is stored.
In step 120, it is determined whether or not the above processing has been performed on all the extracted face candidate areas, and steps 112 to 120 are repeated until the determination in step 120 is affirmative.

In the above, the correlation coefficient is used as the feature amount used to determine whether or not the face is a person's face.
As described in No. 346333, a region corresponding to a human face is used by using an invariant derived from the normalized central moment around the center of gravity, an autocorrelation function, or a geometric invariant as a feature amount. It may be determined whether or not.

In step 122, the exposure amount Ej is calculated based on the photometric data of the face area determined above and the average density Dj of the entire screen of the image. The exposure amount Ej can be calculated by the following equation (6).

[0050] logEj = LMj · CSj · (DNj -Dj) + PBj + LBj + MBj + NBj + K 1 + K 2 ... (6) However, the meaning of each symbol is as follows.

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

CS: Color slope coefficient. It is prepared for each type of negative film, and there are one for underexposure and one for overexposure. It is determined whether the average density of the image frame to be printed is under or over the standard negative density value, and either under exposure or over exposure is selected.

DN: Standard negative density value. D: Density value of image frame to be printed.

PB: Correction balance value for standard color paper. It is determined according to the type of color paper.

LB: Corrected balance value for the standard printing lens. It is determined according to the type of lens used for printing.

MB: A correction value (master balance value) for fluctuations in the light source light quantity and paper development performance.

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

K ₂ : Color correction amount. K ₁ : A density correction amount represented by the following formula.

[0059]

[Equation 6]

Here, Ka and Kb are constants, and FD is the face area average density. As a result, an exposure amount Ej capable of properly printing the face area can be obtained.

Further, the density correction amount K ₁ in the equation (6) may be used as a correction value obtained by the film inspection apparatus, and the color correction amount K ₂ may be expressed by using the face area average density as follows.

[0062]

[Equation 7]

However, Kc is a constant. Further, the density correction amount K ₁ and the color correction amount K ₂ in the above equation (1) are used as the correction amounts obtained by the film inspection device, and the average density Dj of the print frame in the equation (1) is calculated as the average density FDj of the face area.
Alternatively, the exposure amount may be calculated. Further, in the above, the three-color average density of the face area is used as the feature amount of the area corresponding to the face of the person, but instead of this, the density histogram of the face area, the maximum density, the minimum density, the dispersion of the density, etc. Various characteristic amounts may be used.

At the next step 124, the exposure amount Ej calculated above is output to the driver 46. As a result, the driver 46 moves the color correction filter 18 to the position corresponding to the exposure amount Ej based on the input exposure amount Ej. Further, the control unit 40 opens the black shutter 26 for a predetermined time determined by the exposure amount Ej. As a result, the image positioned at the exposure position is printed on the photographic printing paper 28 with the above exposure amount Ej.

In the above description, the edge extraction processing is performed after the skin color area extraction processing. However, any processing may be performed first, and the processing unit and the edge for performing the skin color area extraction processing may be executed. It is also possible to provide a processing unit that performs extraction processing separately and perform processing in parallel in both processing units.

Further, in extracting the skin color area, as described in JP-A-4-346333, etc., a histogram of hue values is obtained based on the data obtained by photometrically measuring the original image, Divide the obtained histogram, determine which of the divided mountains each pixel belongs to, divide each pixel into groups corresponding to the divided mountains, and divide the color original image into a plurality of regions for each group, You may make it judge the area | region whose hue is included in the range of the skin color among these several areas. However, in the division of the histogram in the above processing, in order to prevent the face area from being excessively divided as shown in FIG. 9C, the division is performed so that each divided portion includes a plurality of peaks. It is preferable.

Further, in the above, the correlation coefficient r with the standard face image is obtained for the extracted face candidate area, and it is determined whether or not the area corresponds to a person's face based on this correlation coefficient r. However, the present invention is not limited to this, and corresponds to the positional relationship between the face of the person and a specific portion (for example, the head or the body) of the person in the extracted face candidate area. Based on whether or not a position has a size and orientation corresponding to the face candidate area and a shape pattern unique to the specific portion (for example, a shape pattern representing the contour of the head, the body, or the like). Alternatively, it may be determined whether or not the area corresponds to a person's face.

In the above, the area corresponding to a person is extracted from the image recorded on the negative film 12 as a photographic film, and the exposure amount for printing the image on the photographic paper 28 is determined. As the image recording medium, various recording media such as photographic film such as reversal film and paper can be applied, and the present invention can be used for determining copying conditions when copying an image by an electrophotographic method or the like. It is also possible to apply.

[0069]

As described above, according to the first aspect of the present invention, it is determined whether the hue at each position on the original image is included in the skin color range, and the original image is set to the skin color area and the non-skin color area. In addition to dividing into, the edges in the original image are detected and each part in the original image is classified into an edge part or a non-edge part, which is located in the skin color area and is classified as a non-edge part. An area in the original image composed of a collection is extracted as a face candidate area, and a face candidate area determined to be an area corresponding to a person's face is extracted from the extracted face candidate areas. This has an excellent effect that a region corresponding to a person's face existing in the image can be appropriately extracted with high accuracy.

According to a second aspect of the present invention, an area corresponding to a person's face is extracted from the original image by the face area extraction method of the first aspect, and a copy is made on a copy material based on the image feature amount in the extracted area. Since the conditions are determined, there is an excellent effect that it is possible to obtain with high accuracy a copying condition capable of properly printing an area corresponding to the face of a person existing in the color original image.

[Brief description of drawings]

FIG. 1 is a schematic configuration diagram of a photographic printing apparatus according to this embodiment.

FIG. 2 is a flowchart illustrating a photo printing process according to the present embodiment.

FIG. 3 is a flowchart illustrating a skin color area extraction process.

FIG. 4 is a flowchart illustrating edge extraction processing.

FIG. 5 is a diagram showing an ellipsoid defined by color coordinates for extracting a skin color region.

6A and 6B are image diagrams for explaining closing, which is one of morphological transformations.

FIG. 7 is a conceptual diagram showing an example of a differential filter for detecting edges.

8A and 8B show, as an example, (A) an original image, (B) a result of extracting a skin color region from the original image, (C) a result of extracting an edge from the original image, and (D) extracting a face candidate region. It is an image figure which respectively shows the result.

9A and 9B are diagrams for explaining the problems of the related art, where FIG. 9A is a case where a face candidate area can be properly extracted, FIG. 9B is a case where a face area and a background area cannot be separated, and FIG. FIG. 3 is an image diagram showing an example of a case where a face area is excessively divided.

[Explanation of symbols]

10 Photo printing equipment 12 Negative film 28 photographic paper 32 CCD image sensor 40 control unit 60 ellipsoid

─────────────────────────────────────────────────── ─── Continuation of the front page (58) Fields surveyed (Int.Cl. ⁷ , DB name) G03B 27/72-27/80 H04N 1/38-1/393

Claims (2)

(57) [Claims] 1. Whether or not the hue at each position on the original image represented by the data obtained by the photometry is included in the skin color range The original image is divided into a flesh-colored area and a non-flesh-colored area, and an edge in the original image is detected based on the data obtained by the photometry. Areas in the original image that are classified into edge parts and that are located within the skin color area and that are composed of the parts classified as non-edge parts are extracted as face candidate areas, and the extracted face candidate areas are Face area extraction method that determines whether or not the area corresponds to the human face, and extracts the face candidate area determined to be the area corresponding to the person's face. 2. The face area extracting method according to claim 1,
An area corresponding to a person's face is extracted from the original image recorded on the recording medium, and the copy condition of the original image on the copy material is determined based on the image feature amount in the extracted area. How to determine copying conditions.