JP2007094623A - Image correcting method and device, and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To easily and accurately extract the face and clothes of a photographed object on an image and to correct the image of each region. <P>SOLUTION: A face region extracting section 22 extracts a face region Rf from a face extraction object region Rxf specified by a user on a correction object image S, and a correlative information acquiring section 31 acquires correlative information Cc showing contents of correlative matters correlative to kinds of clothes of the photographed object. A clothes kind specifying section 33 acquires the kind of clothes with highest correlation to the contents of the correlative information Cc, and a clothes shape data acquiring section 33 acquires clothes shape data Ds showing the shape of the clothes. A clothes extraction region setting section 35 sets a clothes extraction object region Rxc modeled on the shape, into a position which should have the clothes of the photographed object determined on the basis of the face position of the photographed object, and a clothes region extracting section 36 extracts a clothes region Rc from the clothes extraction object region Rxc. A correcting section 40 corrects brightness, color, gradation, and the like of the respective images of the face region Rf and clothes region Rc. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to an image correction method and apparatus for correcting the brightness, color, gradation, and the like of a person's face and clothes in a digital image, and a program therefor.

  2. Description of the Related Art Conventionally, so-called digital dodging processing has been proposed in which a person's face is extracted from a digital photographic image having a person as a subject, and the brightness, color, gradation, and the like of the face are corrected. In this digital dodging process, if a face is extracted from information of the entire image without specifying the position of the face on the image, the face may not be accurately extracted if the background is complicated.

Therefore, the user (operator) manually designates a local area including the face on the image while looking at the target image, and extracts the face from the designated area based on the image information of the designated area. A method has been proposed. According to this method, it is possible to easily and accurately specify the approximate position of the face on the image, and to eliminate as much background elements as noise when extracting the face as much as possible, efficiently and with higher accuracy. Can be extracted.

  By the way, when correcting a digital photographic image, there are cases where it is desired to correct not only the face of a person who is a subject but also the clothes. For example, in a digital photo image of a woman wearing a wedding dress obtained by photographing at an event such as a wedding, not only the face of the woman but also the brightness, color, gradation, etc. of the wedding dress Often you want to correct.

  Even in such a case, as in the case of the face, the user manually specifies a local area including the clothes of the person on the image while viewing the target image, and information on the image of the specified area A method of extracting clothes from the designated area based on the above can be used.

  The following is disclosed as an image processing method related to the digital dodging.

  For example, in Patent Document 1, a subject is extracted from a digital image obtained by photographing, a background image suitable for the subject is selected based on the brightness and color of the subject, the subject image, the background image, In this case, the brightness is obtained from the clothing portion of the subject.

Further, for example, in Patent Document 2, the contour of a subject existing in a digital image is extracted, and white balance is adaptively adjusted based on image information at the boundary between the extracted subject region and other regions. A method for performing adjustment, saturation correction, and the like is disclosed. In this case, the extraction of a subject is an initial setting that is set based on the arrangement of facial parts constituting the face and the relative positional relationship of these facial parts. This is done by using boundary information.
JP 2003-163803 A JP 2003-216936 A

  However, in the digital dodging method for correcting the brightness, color, etc. of a person's face on the digital image and the clothes, as described above, in the method in which the user manually extracts the clothes to be extracted, the clothes are the faces. There are many cases where the shape is complicated, and in order to accurately specify the region from which the clothes are to be extracted, complicated work and skill of the user are required, and there is a problem that the burden on the user is large.

  In view of the above circumstances, the present invention accurately extracts a person's face and its clothing area on a digital image without requiring complicated work and user skills, and corrects the image of these areas to An object of the present invention is to provide an image correction method and apparatus capable of improving the image quality of an image, and a program therefor.

  Note that the method disclosed in Patent Document 1 merely selects a background image suitable for a subject, combines the subject image and the background image, and does not perform image processing on the subject image itself. . Moreover, the information on the clothing part is obtained by sampling from a plurality of locations on the image, and the clothing part is not extracted. Therefore, the method of Patent Document 1 is technically different from the image correction method of the present invention described later.

  Moreover, the method of the above-mentioned patent document 2 sets initial boundary information based on the arrangement of facial parts constituting the face and the relative positional relationship of these facial parts, and is compatible with various clothes. Do not mean. Therefore, the method of Patent Document 2 is also technically different from the image correction method of the present invention described later.

  The image correction method of the present invention includes a step of setting a local region including a face of the subject specified by an operator in a subject image including a subject that is a person as a face extraction target region, and an image of the face extraction target region. Extracting a face region representing the face from the region based on the information, obtaining correlation information representing the content of at least one correlation item correlated with the type of clothing of the subject, and the correlation The step of identifying the type of clothing having the highest correlation with the content of the correlation item represented by the acquired correlation information based on the relationship, and a plurality of clothing shape data each representing the shape of each clothing for each type , The step of acquiring clothes shape data corresponding to the specified type of clothes and the position of the face extraction target area or the face area in the subject image A region having a shape represented by the acquired clothing shape data at a position where the clothing of the subject should be, as a clothing extraction target region, and based on the image information of the clothing extraction target region, from the region Extracting a clothing region representing clothing of the subject; and performing image processing on each image of the face region and the clothing region in the subject image to obtain at least one of brightness, color, and gradation of each image And a correcting step.

  Here, the face extraction target area may be a rectangle or a circle.

  The level of the correlation is, for example, a correlation table in which the contents of each correlated item correlated for each type of clothes are prepared, and the number of items in which the contents of the correlated items represented by the acquired correlation information match. Can be determined accordingly. In this case, each correlation item or its content may be weighted, a point multiplied by the weight for each matched item may be calculated and determined according to the value of that point. In this way, even when there are few types of correlation items represented by the acquired correlation information, the type of clothing having the highest correlation can be specified. In addition, when there are a plurality of clothing types having the same correlation and the highest correlation, the list is displayed on the screen of the display device, and the operator selects one type from the list using the GUI. You may do it.

  The shape of the clothes may be a shape in which the outline of the clothes is modeled by a straight line or a curve, or may be an average shape of the type of clothes.

  It is preferable that the shape of the clothing extraction target region can be finely adjusted by the operator.

  In the image correction method according to the present invention, the step of extracting the clothing region is associated with a clothing color as a correlation item for the clothing type specified by the step of specifying the clothing type in the correlation table. In this case, a step of extracting a region of a color within a predetermined range close to the color of the clothing from the clothing extraction target region.

  In the image correction method of the present invention, the correlation items include the sex of the subject, the color or texture of the subject's clothing, the presence or absence of a related wearing device used simultaneously with a specific type of clothing, the type of the hairstyle of the subject, It may be at least one of a shooting time of the subject and an event when the subject is shot.

  The gender of the subject is extracted, for example, by the conventional image recognition technology based on the brightness, hue, saturation, etc. of the image of the face extraction target area, and the subject's face contour, hair, skin, wrinkles, etc. It can be determined comprehensively by evaluating the shape of the face, the length of the hair, the brightness of the skin, the presence or absence of wrinkles, and the like.

  The color or texture of the clothing of the subject is, for example, obtained from the contour shape of the face region by determining the face inclination and the axial direction and the vertical length of the face, and downward from the face center along the axial direction of the face inclination. A local area at a position moved by the vertical width of the face is defined as an area where a part of the subject's clothes should be, and based on the image pattern, brightness, hue, saturation, etc. of the area, The color or texture of the clothes can be determined.

  As the related wearing goods, for example, Bunkin Takashimada and square corner used simultaneously with white solid, tiara and veil used simultaneously with wedding dress, etc., the presence or absence of these related wearing goods, for example, It can be obtained by a pattern matching method based on image information of an area within a predetermined range around the face extraction target area and the face area.

  The hairstyle of the subject can be obtained, for example, by extracting a hair region based on the brightness, hue, saturation, etc. of the image of the face extraction target region and evaluating its contour shape.

  The photographing time of the subject can be obtained by reading from incidental information of the subject image, for example.

  The event when the subject is photographed is displayed by, for example, displaying a list of candidate events prepared in advance on the screen of the display device, and the operator selecting from the list using a GUI (Graphical User Interface). Can be sought. In addition, for example, the shooting time is read from the incidental information of the subject image, and the shooting time matches a predetermined time when a general annual event such as an adult ceremony, Shichigosan, an entrance ceremony, a graduation ceremony, etc. is held. It can be determined by whether or not.

  In the image correction method of the present invention, the step of setting the clothing extraction target region is a step of adjusting the clothing extraction target region to a size corresponding to the size of the face extraction target region or the face region. It may be.

  In the image correction method of the present invention, the clothes shape data includes two types of data corresponding to the standing position and the sitting position of the subject, and the step of acquiring the clothes shape data includes the step of acquiring the face shape data. It may be a step of selecting and acquiring one of the two types of data according to the relative position of the extraction target area or the face area with respect to the subject image. Note that the standing position means a state where the subject is standing, and the sitting position means a state where the subject is sitting on a chair or the like.

  The image correction apparatus of the present invention includes a face extraction target region setting unit that sets a local region including a face of the subject specified by an operator in a subject image including a subject that is a person as a face extraction target region, and the face Based on image information of the extraction target area, face area extraction means for extracting the face area representing the face from the area, and correlation information representing the content of at least one correlation item correlated with the type of clothing of the subject A correlation information acquisition unit that acquires the content of the correlation item represented by the acquired correlation information based on the correlation, and a clothing type identification unit that identifies the type of clothing having the highest correlation. Clothing shape data acquisition means for acquiring clothing shape data corresponding to the specified clothing type from among a plurality of clothing shape data representing the shape of each clothing, and the face in the subject image Clothing extraction target region setting means for setting a region having a shape represented by the acquired clothing shape data at a position where the subject's clothing should be determined by the position of the extraction target region or the face region; Based on the image information of the clothes extraction target area, clothes area extracting means for extracting a clothes area representing the clothes of the subject from the area, and performing image processing on each image of the face area and the clothes area Correction means for correcting at least one of brightness, color, and gradation of the image.

  In the image correction apparatus of the present invention, the correlation items include the gender of the subject, the color or texture of the subject's clothing, the presence or absence of related wearing items used simultaneously with a specific type of clothing, the type of the subject's hairstyle, It may be at least one of a shooting time of the subject and an event when the subject is shot.

  In the image correction apparatus of the present invention, the clothes extraction target area setting means adjusts and sets the clothes extraction target area to a size corresponding to the size of the face extraction target area or the face area. May be.

  In the image correction apparatus of the present invention, the clothing shape data includes two types of data corresponding to each of the standing position and the sitting position of the subject, and the clothing shape data acquisition means includes the face extraction target region or One of the two types of data may be selected and acquired according to the height of the relative position of the face area with respect to the subject image.

  The program of the present invention includes a face extraction target region setting unit that sets a local region including a face of the subject specified by an operator in a subject image including a subject that is a person as a face extraction target region; Based on the image information of the face extraction target area, a face area extracting means for extracting a face area representing the face from the area, and a correlation indicating the contents of at least one correlation item correlated with the type of clothing of the subject Correlation information acquisition means for acquiring information, clothes type specifying means for specifying the type of clothes having the highest correlation with the content of the correlation item represented by the acquired correlation information based on the correlation, and each type Clothes shape data acquisition means for acquiring clothes shape data corresponding to the specified type of clothes from a plurality of clothes shape data representing the shape of each of the clothes, and the subject image A clothing extraction target region setting that sets, as a clothing extraction target region, a region having a shape represented by the acquired clothing shape data at a position where the clothing of the subject determined by the face extraction target region or the position of the face region should be Means for extracting a clothing area representing the clothing of the subject from the area based on the image information of the clothing extraction target area, and for each image of the face area and the clothing area in the subject image. This is a program for performing image processing to function as correction means for correcting at least one of the brightness, color, and gradation of each image.

  In the program of the present invention, the correlation items include the gender of the subject, the color or texture of the subject's clothing, the presence or absence of related wearing items used simultaneously with a specific type of clothing, the type of the subject's hairstyle, the subject And at least one of the events when the subject is photographed.

  In the program of the present invention, the clothes extraction target area setting means may adjust and set the clothes extraction target area to a size corresponding to the size of the face extraction target area or the face area. Good.

  In the program of the present invention, the clothing shape data includes two types of data corresponding to each of the standing position and the sitting position of the subject, and the clothing shape data acquisition unit is configured to detect the face extraction target region or the face. One of the two types of data may be selected and acquired according to the relative position of the region with respect to the subject image.

  According to the image correction method and apparatus and the program therefor according to the present invention, the face area is extracted from the face extraction target area designated by the operator on the subject image, and the correlation items correlated with the type of clothes of the subject. Correlation information representing the content is obtained, the type of clothing having the highest correlation with the content is obtained, and the clothing extraction target area imitated by the shape of the clothing is determined based on the position of the subject's face. Correlation information is used because it is set at a desired position, a clothing region is extracted from the clothing extraction target region, and at least one of the brightness, color, and gradation of each image of the face region and the clothing region is corrected. Therefore, it is possible to identify the type of clothes of the actual subject and set the clothes extraction target area to be extracted from the clothes area in an appropriate shape with an appropriate shape, which requires complicated work and operator skills. Need To, the face of the person on the digital image region of the clothing was precisely extracted, it is possible to improve the image quality of the image by correcting the image in these areas.

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

  FIG. 1 is a block diagram showing a configuration of an image correction apparatus according to an embodiment of the present invention. This image correction device is a device for obtaining the same effect on a digital image as a so-called dodging method in analog photographic technology. For example, the image correction device partially corrects a face and clothes that are darkly expressed on an image. is there.

  As shown in FIG. 1, the image correction apparatus 1 includes an original image storage unit 11, an image selection unit 12, a face extraction target region setting unit 21, a face region extraction unit 22, a correlation information acquisition unit 31, a clothing type identification unit 32, Clothing shape data acquisition unit 33, correlation database 34, clothing extraction target region setting unit 35, clothing region extraction unit 36, correction unit 40, output control unit 50, display device 61, corrected image storage device 62, printing device 63, input operation Part 70 is provided. The correction unit 40 includes a mask creation unit 41, a correction parameter setting unit 42, an overall correction unit 43, and an image composition unit 44.

  The original image storage unit 11 stores a plurality of original images including a subject that is a person as digital data. As the original image storage unit 11, for example, a hard disk, a CD-ROM device, a DVD-ROM device, or the like can be considered. Here, the original image is an image that has not been subjected to a process for changing the image quality.

  The image selection unit 12 selects an original image as a correction target image (subject image) S from among a plurality of original images stored in the original image storage unit 11. In the present embodiment, the user (operator) designates a specific image using a GUI (graphical user interface) from among thumbnail images that are displayed on the screen of the display device 61 and shows an outline of the original image. Thus, the original image to be corrected is selected.

  The face extraction target area setting unit 21 sets a local area including the face of the subject specified by the user in the correction target image S as the face extraction target area Rxf. In the present embodiment, the user sets a face extraction target region Rxf by designating a specific region with the GUI in the correction target image S displayed on the screen of the display device 61. The face extraction target area Rxf may be rectangular or circular.

  The face area extraction unit 22 extracts the subject from the face extraction target area Rxf based on the image information of the face extraction target area Rxf set by the face extraction target area setting unit 21, for example, the brightness or color information of the image. A face region Rf representing a face is extracted. In addition, as a method of extracting the face area of the subject, for example, a small area having a high probability of being a skin part in the designated area as proposed in Japanese Patent Application No. 2005-162836 is obtained as a skin seed. It is possible to use a method in which a skin-like region is grown based on a predetermined index starting from the skin type, and finally a cluster of skin-like regions is extracted. Further, for example, the brightness, hue, saturation, etc. of the pixels constituting the image of the designated area are obtained, and an image area composed of pixels in which the index of these colors matches the characteristics of the skin color is simply selected from the designated area. In addition, a method of extracting as a region representing a person's skin part can also be used.

  The correlation information acquisition unit 31 acquires correlation information Cc representing the content of at least one correlation item correlated with the type of clothing of the subject. Here, the correlation items include, for example, the gender of the subject, the color or texture of the subject's clothing, the presence or absence of related accessories used at the same time as the specific type of clothing, the type of the subject's hairstyle, the shooting time of the subject, the subject You can think about the event when was taken.

  Here, the sex of the subject is extracted, for example, by the conventional image recognition technology based on the brightness, hue, saturation, etc. of the image of the face extraction target region Rxf, and the subject's face contour, hair, skin, wrinkles, etc. It can be determined comprehensively by evaluating the shape of the face, the length of the hair, the brightness of the skin, the presence or absence of wrinkles, and the like.

  The color or texture of the clothing of the subject is obtained, for example, from the contour shape of the face region Rf, the inclination of the face, the axial direction and the vertical width of the face, and from the center of the face along the axial direction of the face inclination. A local area at a position moved downward by 1 to 1.5 times the vertical width of the face is defined as an area where a part of the subject's clothes should be, and the pattern, brightness, hue, and saturation of the image in that area Based on the above, the color or texture of the clothing of the subject can be obtained.

  FIG. 2 is a view showing a woman wearing Japanese clothes. The shape of the clothing of the subject varies depending on the type, but most clothing is partly on the subject's chest, and the typical color of the clothing is one of the clothing on the chest. There is a tendency to match the color of the part. Accordingly, by detecting a rounded chin line from the contour shape of the extracted face region Rf, the inclination of the face of the subject is obtained, and from this, the axial direction of the face inclination and the vertical width of the face are obtained, A rectangular area having the same vertical width and horizontal width as the face area Rf around the position moved by 1.5 times the vertical width of the face downward along the axial direction of the face inclination from the center of the face. It is possible to determine the color of the subject's clothing based on the brightness and hue of the image in this region (clothing information detection region) Rd.

  In addition, as the above-mentioned related wearing goods, for example, you can think of Bunkin Takashimada and corner concealment used simultaneously with white solid, tiara and veil used simultaneously with wedding dress, etc. For example, it can be obtained by color determination and pattern matching based on image information of a region within a predetermined range centering on the face extraction target region Rxf and the face region Rf.

  FIG. 3 is a diagram showing a woman wearing Bunkin Takashimada on the subject's head. Bunkin Takashimada has a characteristic shape that is almost white and flat in the horizontal direction. Therefore, such a shape and color pattern is subjected to pattern matching in a partial region adjacent to the upper side of the extracted face region Rf, for example, in a region about three times the width of the face region Rf and twice the height. By searching, it is possible to detect the object region Rj representing Bunkin Takashimada and determine whether or not the alternate Takashimada is attached.

  In addition, the hairstyle of the subject can be obtained by, for example, extracting a hair region based on the brightness, hue, saturation, and the like of the image of the face extraction target region Rxf and evaluating the contour shape.

  Further, the photographing time of the subject can be obtained by reading from the supplementary information of the correction target image S, for example.

  The event when the subject is photographed can be obtained, for example, by displaying a list of event candidates prepared in advance on the screen of the display device, and the user selecting from the list using the GUI. Further, for example, the shooting time is read from the supplementary information of the correction target image S, and the shooting time matches a predetermined time at which a general annual event such as an adult ceremony, Shichigosan, entrance ceremony, graduation ceremony, etc. is held. It can be determined by whether or not. Moreover, an event can also be estimated using imaging reservation information.

  The clothing type identification unit 32 is based on the correlation between the clothing type of the subject and the correlation item, and the content of the correlation item represented by the correlation information Cc acquired by the correlation information acquisition unit 31 and the clothing item having the highest correlation. The type is specified as the type of clothing Vc of the subject. Here, the level of the correlation is represented by, for example, a correlation table T in which the contents of each correlated item correlated with each clothing type are associated, and the acquired correlation information Cc for each clothing type. The number of items that match the contents of the correlation items can be obtained and determined according to the number of items. In this case, each correlation item or its content may be weighted, a point multiplied by the weight for each matched item may be calculated and determined according to the value of that point. In this way, even when there are few types of correlation items represented by the acquired correlation information, the type of clothing having the highest correlation can be specified. If there are a plurality of clothing types having the same correlation and the highest correlation, the list is displayed on the screen of the display device, and the user selects one type from the list by using the GUI. It may be. In the present embodiment, the correlation table T is stored in the correlation database 34 in advance, and the clothing type identification unit 32 reads the correlation table T from the correlation database 34, and the correlation items represented by the correlation information Cc in the correlation table T are displayed. The type of clothing having the largest number of items matching the content is obtained, and the type of clothing is specified as the type of clothing Vc of the subject.

  FIG. 4 shows an example of the correlation table T. In this correlation table T, the type of clothing and the correlation items, such as the subject's sex (gender), the color or texture of the subject's clothing (cloth color / texture), and the related wearing used simultaneously with the specific type of clothing. Items (related objects), shooting time of the subject (shooting time), contents such as an event (event) when the subject is shot are associated with each other.

  The clothing shape data acquisition unit 33 selects the clothing shape data Dc corresponding to the clothing type Vc specified by the clothing type specifying unit 32 from the plurality of clothing shape data, each representing the shape of each clothing for each type. Is to acquire. In the present embodiment, the plurality of clothing shape data are stored in the correlation database 34 in advance, and the clothing shape data acquisition unit 33 selects the clothing shape from the plurality of clothing shape data stored in the correlation database 34. Data Dc is read and acquired. The clothing shape data includes two types of data corresponding to the standing position and the sitting position of the subject, and the clothing shape data acquisition unit 33 corrects the face extraction target region Rxf or the correction target image S of the face region Rf. Depending on the relative height of the position, one of the two types of data corresponding to the standing position and the sitting position of the subject is selected and acquired. In addition, the shape of the clothes represented by the clothes shape data may be a shape obtained by modeling the outline of the clothes with a straight line or a curve, or may be an average shape of the type of clothes. Is a modeled shape formed by connecting a plurality of straight lines.

  As described above, the correlation database 34 stores the correlation table T and a plurality of clothing shape data. As the correlation database 34, for example, a hard disk, a CD-ROM device, a DVD-ROM device, or the like can be considered.

  The clothing extraction target region setting unit 35 is the clothing shape data Vc acquired by the clothing shape data acquisition unit 33 at the position where the subject's clothing should be determined by the position of the face extraction target region Rxf or the face region Rf in the correction target image S. A region having a shape represented by is set as a clothing extraction target region Rxc. The clothing extraction target area setting unit 35 adjusts and sets the clothing extraction target area Rxc to a size corresponding to the size of the face extraction target area Rxf or the face area Rf. Note that the shape of the clothing extraction target region Rxc set here is a shape as an initial setting, and the user can fine-tune this shape using the GUI.

  The clothing region extraction unit 36 extracts a clothing region Rc representing clothing of the subject from the clothing extraction target region Rxc based on image information of the clothing extraction target region Rxc, for example, the brightness, hue, saturation, and the like of the image. It is. In the correlation table T, when the clothing color is associated as the correlation item for the clothing type identified by the clothing type identifying unit 32, the clothing region extracting unit 36 is a predetermined value close to the clothing color. A color region within the range may be extracted from the clothing extraction target region Rxc.

  The correction unit 40 performs image processing on each image of the face region Rf and the clothing region Rc in the correction target image S to correct the brightness or color of each image. As described above, the mask creation unit 41 , A correction parameter setting unit 42, an overall correction unit 43, and an image composition unit 44.

  The mask creating unit 41 serves to mask areas other than the specific correction target area so that the correction covers only the correction target areas such as the face area Rf and the clothing area Rc extracted on the correction target image S. Mask data Mf and Mc are created. The mask data Mf and Mc represent individual mask data created for each correction target area. The mask data Mf is specified to be corrected only on the face area Rf, and the mask data Mc is applied to the clothes area Rc. It is specified so that only the correction can be made. Specifically, the mask data Mf includes the pre-correction image before the correction to be performed on the face region Rf and the correction image to be applied to the face region Rf on the entire image before the correction. The combination ratio in units of pixels is defined so that a partial correction image in which only the face region Rf is corrected can be obtained when the applied overall correction image is combined with pixels corresponding to each other. . That is, the composition ratio (the image before correction: the image after correction) of the pre-correction image and the entire correction image defined by the mask data Mf is normally 0 in the pixel corresponding to the face region Rf that is the correction target region. 1 and 1: 0 for pixels corresponding to outside the face region Rf. However, in order to smooth the inner and outer joints of the correction target area, the composition ratio is gradually increased from 1: 0 to 0: 1 along the inner and outer directions of the correction target area near the boundary of the correction target area. You may prescribe | regulate so that it may change. Like the mask data Mf, the mask data Mc is an image of a pre-correction image before the correction to be applied to the clothing region Rc and an image of the image before the correction to be applied to the clothing region Rc. Defines the composition ratio in units of pixels so that a partial correction image in which only the clothing region Rc is corrected can be obtained when the entire correction image applied to the whole is combined and the corresponding pixels are combined. It is. The mask data is basically created for each of the face region Rf and the clothing region Rc. However, when the same correction (image processing) is applied to the face region Rf and the clothing region Rc, These areas may be combined into one area group Rfc, and one mask data Mfc may be created for this area group Rfc.

  The correction parameter setting unit 42 defines parameters Pf and Pc (hereinafter referred to as correction) that define image processing conditions for correcting the brightness, color, and the like applied to each correction target area, that is, each image of the face area Rf and the clothes area Rc. Parameter). The correction parameter can be set according to a user input, or a parameter obtained by automatic calculation can be set. Here, the correction parameter determines the correction amount (shift amount) of the brightness, hue, and saturation of the image. For example, the correction parameter determines a value to be added to or subtracted from the pixel value of each pixel constituting the image. . The correction parameter setting unit 42 can update and set the correction parameter any number of times. That is, an image obtained by performing image processing defined by the correction parameters at the current time on the correction target image S is displayed on the screen of the display device 61 under the control of the output control unit 50, and the user observes it. The effect of correction can be confirmed and a desired correction parameter can be searched.

  The correction parameter setting unit 42 extracts a face region Rf and a clothing region Rc, which are correction target regions, and creates mask data Mf and Mc corresponding to the extracted regions, in advance, which is a target after correction of the correction target region. Based on the difference between the determined predetermined brightness and the average value or intermediate value of the brightness in the correction target area, a correction amount that is considered appropriate as a correction amount when correcting the correction target area is calculated. Set as initial value. Thereafter, a partial corrected image Sp_2, which is generated by the overall correction unit 43 and the image combining unit 44 and corrected only for these correction target areas, is displayed on the screen of the display device 61 under the control of the output control unit 50. However, the correction parameter setting unit 42 has a function of changing the correction parameter based on information for fine adjustment of the correction amount input by the user so that the user can further finely adjust the correction amount while viewing the screen. Have Then, the image is corrected again with the changed correction parameter, and the corrected image is displayed on the screen of the display device 61. Note that the components on the image that can be corrected are not particularly limited, but here, in order to improve the operability related to the correction, as components of at least one of hue, saturation, and lightness, several components are previously stored. The user can select any of the components and increase / decrease the component. For example, the components on the image that can be corrected are determined as four types of intensity of three colors of cyan C, magenta M, and aero Y and brightness D, and the screen of the display device 61 is controlled by the output control unit 50. These component items are displayed, a counter for displaying the correction level of the component stepwise next to each component item, and an increase / decrease button for increasing / decreasing the correction level one step at a time. Adjusts the correction level of each of these components using a GUI.

  The correction parameters are usually set one by one for each correction target area, but a plurality of correction target areas are grouped into one area group, that is, a plurality of masks respectively corresponding to each of the plurality of correction target areas. It is also possible to combine the data into one mask data and set one correction parameter for this.

  The overall correction unit 43 performs image processing defined by one of the correction parameters Pf and Pc on the entire image of the correction target image S to correct the brightness, color, etc. of the entire image of this image. The first overall corrected image Sz_1 is obtained. In addition, when the first partial correction image Sp_1 in which one of the face region Rf and the clothing region Rc is corrected has already been obtained by the image composition unit 44, the first partial correction image. The entire image of Sp_1 is subjected to image processing defined by the correction parameter that is not yet used for correction, and the second entire corrected image Sz_2 in which the brightness, color, and the like of the entire image of this image are corrected. Is what you get. The first and second overall corrected images Sz_1 and Sz_2 are sent to the image composition unit 44.

  When the first overall correction image Sz_1 is sent from the overall correction unit 43, the image compositing unit 44 combines the correction target image S and the first overall correction image Sz_1 with pixels corresponding to each other. The first overall correction image Sz_1 is synthesized at a synthesis ratio defined by the mask data of the correction target region corresponding to the correction parameter used when obtaining the first overall correction image Sz_1, and only the correction target region is partially corrected. The partial correction image Sp_1 is obtained. Further, when the second correction image Sz_2 is sent from the whole correction unit 43, the first partial correction image Sp_1 and the second whole correction image Sz_2 are combined with each other corresponding pixels, and the first correction image Sz_2 is sent. A second portion in which only the correction target area is partially corrected by combining at a combining ratio defined by the mask data of the correction target area corresponding to the correction parameter used when obtaining the entire correction image Sz_2 of 2 A corrected image Sp_2 is obtained. By combining the images in this way, it is possible to obtain a partial correction image Sp_2 in which the images of the face region Rf and the clothing region Rc are corrected with the corresponding correction parameters.

The following equation (1) is a first overall correction obtained by subjecting the correction target image S and the entire correction target image S to image processing defined by the correction parameter Pf corresponding to the face region Rf. The image Sz_1 is synthesized with the synthesis ratio of the mask data Mf to obtain the first partial correction image Sp_1, the calculation formula (1), the first partial correction image Sp_1, and the first partial correction image Sp_1. The second entire corrected image Sz_2 obtained by performing the image processing defined by the correction parameter Pc corresponding to the clothing region Rc with respect to the entire image of the second image is synthesized at the synthesis ratio of the mask data Mc, and the second Expression (2) for obtaining the partial correction image Sp_2 is shown.

Here, x and y represent the x-coordinate and y-coordinate in each image, and the possible values of the mask data Mf and Mc are 1 ≧ Mf (x, y) ≧ 0, 1 ≧ Mc (x, y) ≧ 0.

  In addition to the original thumbnail image, the output control unit 50 inputs from the user mask images SMf and SMc, partial correction images Sp_2 and the like that represent the composition ratio in pixel units determined by the mask data Mf and Mc in shades. In response to or automatically, the image is displayed on the screen of the display device 61, the finally obtained partial correction image Sp_2 is stored as digital data in the correction image storage device 62, and is output to the printing device 63. The control which performs is performed.

  The display device 61 has a screen for displaying an image, and displays various images on the screen under the control of the output control unit 50. As the display device 61, a CRT monitor, a liquid crystal panel, or the like can be considered.

  The corrected image storage device 62 is a portion in which the image processing as described above is performed on the correction target image S, which is an original image, and the brightness and color of the image of the face region Rf and the clothing region Rc are corrected appropriately. The corrected image Sp_2 is stored as digital data. As the corrected image storage device 62, a hard disk, a CD-ROM device, a DVD-ROM device, and the like can be considered as in the original image storage unit 11.

  Under the control of the output control unit 50, the printing device 63 prints the partial correction image Sp_2 on a paper medium, a film, or the like.

  The input operation unit 70 inputs information indicating various user requests such as selection of an image to be corrected by the image selection unit 12 and selection of an image to be displayed on the screen of the display device 61 to each unit. As the input operation unit 70, for example, a mouse, a keyboard, a touch panel, or the like can be considered.

  Hereafter, the flow of processing in the image correction apparatus 1 will be described.

  FIG. 5 is a flowchart showing the flow of processing in the image correction apparatus 1. First, the output control unit 50 reads out a plurality of original images stored in the original image storage unit 11, creates thumbnail images by reducing the plurality of original images, and displays these thumbnail images on a display device. 61 are displayed side by side on the screen.

  Here, the user designates an image to be corrected by designating a specific image from the plurality of thumbnail images displayed on the screen of the display device 61 using the GUI. The image selection unit 12 selects an original image corresponding to the designated image as the correction target image S.

  FIG. 6A is a diagram illustrating an example of the correction target image S selected in this way. The correction target image S shown in FIG. 6A is a digital photograph image in which a woman wearing a wedding dress is a subject, and the brightness and color of the woman's face and the wedding dress are not appropriate.

  When the correction target image S is selected, the output control unit 50 displays the correction target image S on the screen of the display device 61.

  FIG. 7 is a diagram showing a screen of the display device 61. On the screen of the display device 61, as shown in FIG. 7, the area AR1 displaying either the correction target image S or the partial correction image Sp_2, the mask image SMf represented by the mask data Mf, and the image of the face region Rf are positioned. Area AR2 where the mask image SMc represented by the mask data Mc and the image of the clothing region Rc are displayed in a superimposed manner, and the above-described C, M in the correction target region of the correction target image S A window Win composed of an area AR3 displaying the levels of the respective components Y, D and D and the increase / decrease buttons of these components is displayed. At this stage, the correction target image S is displayed in the area AR1, and nothing is displayed in the area AR2.

  Here, when the user designates a local area including the face to be corrected on the displayed correction target image S using the GUI, the face extraction target area setting unit 21 sets the local area as the face extraction target area Rxf. (Step S1).

  When the face extraction target region Rxf is set by the face extraction target region setting unit 21, the face region extraction unit 22 determines the predetermined information based on the image information of the face extraction target region Rxf, for example, the brightness, hue, and the like of the image. The face area Rf is extracted by extracting the skin color area according to the algorithm (step S2).

  When the face area Rf is extracted, the mask creating unit 41 creates mask data Mf that defines that correction by image processing is applied only to the face area Rf.

  Here, in the area AR2 in the window Win displayed on the screen of the display device 61, the mask image SMf and the image of the face region Rf in the correction target image S are superimposed in position under the control of the output control unit 50. Displayed. By viewing this image, the user can confirm whether or not the extracted face area Rf is a desired area.

  When the face area Rf is extracted from the correction target image S by the face area extraction unit 22, the correlation information acquisition unit 31 uses the above-described method to correlate items that are correlated with the type of clothing of the subject, that is, the subject's clothing. Shows the gender, the color or texture of the subject's clothing, the presence or absence of related accessories used at the same time as a specific type of clothing, the type of the subject's hairstyle, the shooting time of the subject, the event when the subject was shot, etc. Correlation information Cc is acquired (step S3).

  When the correlation information Cc is acquired by the correlation information acquisition unit 31, the clothing type identification unit 32 reads the correlation table T from the correlation database 34, and in this correlation table T, the correlation represented by the correlation information Cc for each type of clothing. The number of items that match the content of the item is obtained, and the type of clothing having the largest number of matching items is specified as the clothing type Vc of the subject. For example, the correlation table T is the correlation table shown in FIG. 4, and as the correlation information Cc, the color of the subject's clothes is “white”, the event is “wedding”, and the corner cover that is a related wearing item is If information indicating that there is no information is acquired, the type of clothes having the most items that match the contents of these correlation items is “dress”, and the type of clothes Vc of the subject is specified as “dress” (step S4). .

  When the clothing type identification unit 32 identifies the clothing type Vc of the subject, the clothing shape data acquisition unit 33 identifies the identified clothing type Vc from the plurality of clothing shape data stored in the correlation database 34. Is read and acquired (step S5). At this time, the clothing shape data acquisition unit 33 obtains the relative position of the face of the subject with respect to the correction target image S based on the position of the face extraction target region Rxf or the position of the extracted face region Rf. If the position is higher than the predetermined height, it is determined that the subject is in a standing position (standing posture), and clothing shape data corresponding to the standing position is selected and acquired. On the other hand, when this position is lower than a predetermined height, it is determined that the subject is in a sitting position (sitting posture), and clothing shape data corresponding to the sitting position is selected and acquired.

  When the clothing shape data acquisition unit 33 acquires the clothing shape data Dc corresponding to the type of clothing of the identified subject, the clothing extraction target region setting unit 35 includes the face extraction target region Rxf in the correction target image S. Alternatively, the partial area of the shape represented by the acquired clothes shape data Dc is set as the clothes extraction target area Rxc at a position where the clothes of the subject determined by the position of the face area Rf should be (step S6). The position of the subject's clothes should be, for example, the position that is moved downward from the lower end of the face extraction target area Rxf or the face area Rf by a predetermined width corresponding to the subject's neck, and the upper end of the clothes. It can be a position. Note that the size of the clothing extraction target region Rxc is adjusted to a size corresponding to the size of the face of the subject by estimating the size of the face of the subject from the size of the face extraction target region Rf or the face region Rf.

  FIG. 8 is a diagram illustrating a state in which the clothing extraction target region Rxc is set in the shape of the clothing shape data corresponding to the standing position of the “dress” so as to surround the clothing of the subject. Note that the outline of the clothing extraction target region Rxc is displayed on the correction target image S displayed on the screen of the display device 61 by the output control unit 50. The user can change the shape of the clothing extraction target region Rxc by using the GUI, and can adjust the shape to match the actual clothing shape of the subject. Thereby, the background component which becomes noise when extracting the clothing region Rc can be eliminated as much as possible, and the extraction accuracy of the clothing region Rc can be further increased.

  When the clothing extraction target region Rxc is set by the clothing extraction target region setting unit 35, the clothing region extraction unit 36 converts the image information of the clothing extraction target region Rxc, for example, the pattern, brightness, hue, and saturation of the image. Based on this, by detecting an edge in the image or extracting a region of the same color having the largest occupied area, a clothing region Rc representing the clothing of the subject in the correction target image S is extracted (step S7). ).

  When the clothing region Rc is extracted, the mask creating unit 41 creates the mask data Mc that is defined so that the correction by image processing extends only to the clothing region Rc. Here, in the area AR2 in the window Win displayed on the screen of the display device 61, the mask image SMc and the image of the clothing region Rc in the correction target image S are superimposed in position under the control of the output control unit 50. Displayed. By viewing this image, the user can confirm whether or not the extracted clothing region Rc is a desired region.

  When the face region Rf and the clothing region Rc are extracted, the correction unit 36 performs image processing on each image of the face region Rf and the clothing region Rc, and corrects the brightness or color of each image (step S8). ). Specifically, this correction is performed as follows.

  The correction parameter setting unit 42 sets initial values of correction parameters Pf and Pc that are conditions for image processing performed on the images of the face region Rf and the clothing region Rc. First, a pure skin color area is detected from the extracted face area Rf, and a correction amount is calculated so that the brightness of the skin color area is appropriate for the skin. That is, the saturation of each pixel in the face region Rf is obtained, and other regions are extracted from the face region Rf excluding pixels whose saturation is below a predetermined threshold. As a result, the skin color area excluding the hair, eyebrows, black eyes, and shadows in the face area Rf is extracted. Then, the brightness of each pixel in this skin color region is obtained, and the average value Dave or intermediate value Dmid of the brightness is calculated. Then, the lightness correction amount Dmod is obtained by subtracting the above-mentioned Dave or Dmid from the lightness that is preferable as the lightness of the skin of the person in the photographic image, that is, the target lightness value after correction (Dmod = Dideal− (Dave or Dmid)). Based on the value of the brightness correction amount Dmod, the correction amount of each component on the image is set as the initial value of the correction parameter Pf. For example, a correction amount to be corrected for each component of C, M, and Y (may be R, G, and B) is defined as a correction amount level, and a signal value for each color component of each pixel is set to this correction amount level. Conversion is performed by a corresponding conversion curve (for example, a one-dimensional lookup table). Similarly, a correction amount is calculated such that the lightness of the extracted clothing region Rc becomes appropriate lightness as the clothing type specified by the clothing type specifying unit 32. That is, the brightness of each pixel in the clothing region Rc is obtained, and the average value Dave or intermediate value Dmid of the brightness is calculated. Then, the lightness correction amount Dmod is obtained by subtracting the above-mentioned Dave or Dmid from the lightness that is preferable as the lightness of the clothing type in the photographic image, that is, the target lightness value after correction (Dmod = Dideal). -(Dave or Dmid)). Based on the value of the brightness correction amount Dmod, the correction amount of each component on the image is set as the initial value of the correction parameter Pc.

  The overall correction unit 43 performs the image processing specified by the correction parameter Pf on the entire image of the correction target image S, and corrects the brightness, color, and the like of the entire image of this image. Sz_1 is obtained. The first entire corrected image Sz_1 is sent to the image composition unit 44.

  When the first overall corrected image Sz_1 is sent to the image synthesis unit 44, the image synthesis unit 44 combines the correction target image S and the first overall correction image Sz_1 together with the pixels corresponding to each other. The first partial corrected image Sp_1 is obtained by combining at a combining ratio defined by the mask data Mf of Rf and partially correcting only the face region Rf. The first partial correction image Sp_1 is sent to the overall correction unit 43.

  When the first partial correction image Sp_1 is sent to the overall correction unit 43, the overall correction unit 43 corrects the correction parameter that has not been used for correction for the entire image of the first partial correction image Sp_1. Image processing defined by Pc is performed to obtain a second overall corrected image Sz_2 in which the brightness, color, etc. of the entire image of this image are corrected. The second entire corrected image Sz_2 is further sent to the image composition unit 44.

  When the second corrected image Sz_2 is sent to the image composition unit 44, the image composition unit 44 combines the first partial correction image Sp_1 and the second overall correction image Sz_2 with corresponding pixels. A second partial correction image Sp_2 in which the clothing region Rc is corrected in addition to the face region Rf is obtained by combining at a composition ratio defined by the mask data Mc of the clothing region Rc. By synthesizing the images in this way, it is possible to obtain a second partial correction image Sp_2 in which the images of the face region Rf and the clothing region Rc are corrected with the corresponding correction parameters. The second partial correction image Sp_2 is displayed on the screen of the display device 61 under the control of the output control unit 40.

  FIG. 6 is a diagram illustrating a state in which the first partial correction image Sp_1 is generated by combining the correction target image S and the first overall correction image Sz_1 at a combination ratio defined by the mask data Mf. FIG. 6A is a diagram illustrating a correction target image S that is a digital photographic image with a woman wearing a wedding dress as a subject. FIG. 6B is a diagram illustrating an example of the first overall correction image Sz_1, and the first overall correction image Sz_1 is set so that the skin area of the face of the woman who is the subject has appropriate brightness. This is an image in a state where the brightness of the entire image of the correction target image S is corrected. FIG. 6 (3) is a diagram showing an example of the mask image SMf represented by the mask data Mf, and the boundary of the face area is blurred. FIG. 6 (4) is a diagram illustrating an example of the first partial correction image Sp_1. The first partial correction image Sp_1 is a female face region Rf so that the female face region has appropriate brightness. Only the figure is corrected.

  FIG. 9 is a diagram illustrating an example of a screen of the display device 61 when the face region Rf and the clothing region Rc are extracted. As shown in FIG. 9, in the area AR2, the mask image SMf of the face area Rf and the mask image SMc of the clothes area Rc are displayed side by side, and the second partial correction image Sp_2 is displayed in AR1. ing.

  While viewing the screen, the user selects a correction target area to be corrected, that is, either the face area Rf or the clothes area Rc, in order to make the correction amount appropriate, and the GUI displays a window Win on the screen of the display device 61 by the GUI. A correction amount increase / decrease button for each of the C, M, Y, and D components displayed in the area AR3 is pressed to finely adjust the correction amount for each component. The correction parameter setting unit 42 resets the correction parameters Pf and Pc in accordance with the fine adjustment instruction from the user.

  The overall correction unit 43 and the image composition unit 44 perform a series of processes using the reset correction parameter, and obtain a new second partial correction image Sp_2. The second partial correction image Sp_2 is displayed on the screen of the display device 61 under the control of the output control unit 50.

  Such correction parameter setting is repeated as necessary, and when the user determines that the desired correction parameter can be set, the correction image Sp_2 is stored or an instruction to print is issued to the image correction apparatus. 1, the output control unit 50 stores the corrected image Sp_2 as digital data in the corrected image storage device 62 or causes the printing device 63 to print in accordance with the command.

  According to such an image correction apparatus of the present invention, the face area is extracted from the face extraction target area designated by the user (operator) on the correction target image (subject image), and the type of clothes of the subject is determined. Correlation information representing the contents of correlated items with correlation is obtained, the type of clothing having the highest correlation with this content is obtained, and the clothing extraction target area imitated by the shape of the clothing is used as a reference for the position of the subject's face Since the clothes of the determined subject should be set to the position where the clothes should be, the clothes area is extracted from the clothes extraction target area, and at least one of the brightness, color and gradation of each image of the face area and the clothes area is corrected By using the correlation information, it is possible to identify the type of clothes of the actual subject and set the clothes extraction target area to be extracted from the clothes area at an appropriate position in an appropriate shape, which is complicated. Requires work and operator skills And without, the face of the person on the digital image region of the clothing was precisely extracted, it is possible to improve the image quality of the image by correcting the image in these areas.

  In the present embodiment, there is only one correlation table T. However, a plurality of different correlation tables may be prepared, and the correlation table to be used may be switched according to circumstances. For example, a correlation table for excursions that lists the types of clothes related to amusement parks and sightseeing spots is prepared, and specific animals and anime characters at amusement parks and sightseeing spots are displayed on the correction target image S. When these characters are detected by performing detection processing, the type of clothing of the subject may be specified using a correlation table for vacation.

  Further, in the present embodiment, when the clothing extraction target region Rxc is set, the shape represented by the clothing shape data is adjusted and set according to the size of the face of the subject. In this case, a function capable of linearly and continuously changing the size of the clothing extraction target region Rxc may be provided and adjusted, or a plurality of clothing shape data having different shape sizes in several stages may be provided. Prepared in advance, according to the size of the face of the subject, select clothing shape data of an appropriate size, and set a region having the same shape as the clothing extraction target region Rxc Also good.

  In the present embodiment, the clothes extraction target region Rxc determines the position where clothes should be found by determining the inclination of the face, and is set to that position. The position and direction may be designated, and the setting position of the clothing extraction target area Rxc may be determined accordingly.

  Further, in the process of extracting the clothing region Rc from the clothing extraction target region Rxc by the clothing region extraction unit 22, the following processing may be used. First, the color of the image of the clothing extraction target region Rxc is determined, and if the image of the clothing extraction target region Rxc contains white or black, threshold determination is performed based on lightness, hue, and saturation. However, regarding the brightness, the brightness threshold is adjusted by looking at the histogram of the entire image. For example, in the brightness histogram, the brightness value is from the minimum value to the overall intermediate value in the brightness histogram, and in white, the brightness value is from the maximum value to the overall intermediate value in the brightness histogram. On the other hand, if the image of the clothing extraction target region Rxc does not include white and black, the region designation by the hue is performed. Considering the case where there are fine patterns and one-point patterns, the expansion process and the contraction process of the morphology process are performed once.

  The image correction apparatus according to the embodiment of the present invention has been described above, but a program for causing a computer to execute each process in the image correction apparatus is also an embodiment of the present invention. A computer-readable recording medium that records such a program is also one embodiment of the present invention.

The block diagram which shows the structure of the image correction apparatus 1 Illustration representing a woman in kimono A figure representing the upper body of a woman wearing Bunkin Takashimada The figure which shows an example of the correlation table T The flowchart which shows the flow of the process performed in the image correction apparatus 1. The figure which shows a mode that the correction | amendment object image S and 1st whole correction image Sz_1 are synthesize | combined, and 1st partial correction image Sp_1 is produced | generated. The figure which shows an example of the screen layout of the display apparatus 61 The figure which shows a mode that the clothing extraction object area | region Rxc was set with the shape of the clothing shape data corresponding to standing of "dress" The figure which shows an example of the screen layout of the display apparatus 61 when the face area | region Rf and the clothing area | region Rc are extracted.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Image correction apparatus 11 Original image storage part 12 Image selection part 21 Face extraction object area | region setting part 22 Face area extraction part 31 Correlation information acquisition part 32 Clothing type specific | specification part 33 Clothing shape data acquisition part 34 Correlation database 35 Clothes extraction object area | region setting Unit 36 Clothing region extraction unit 40 Correction unit 41 Mask creation unit 42 Correction parameter setting unit 43 Overall correction unit 44 Image composition unit 50 Output control unit 61 Display device 62 Correction image storage device 63 Printing device 70 Input operation unit

Claims (12)

Setting a local region including a face of the subject designated by an operator in a subject image including a subject that is a person as a face extraction target region;
Extracting a face area representing the face from the area based on image information of the face extraction target area;
Obtaining correlation information representing the content of at least one correlation item correlated with the type of clothing of the subject;
Based on the correlation, identifying the type of clothing having the highest correlation with the content of the correlation item represented by the acquired correlation information;
Obtaining clothes shape data corresponding to the specified clothes type from a plurality of clothes shape data each representing the shape of each clothes for each type;
Setting a region having a shape represented by the acquired clothing shape data at a position where the clothing of the subject, which is determined by the position of the face extraction target region or the face region in the subject image, should be a clothing extraction target region; ,
Extracting a clothing region representing clothing of the subject from the region based on image information of the clothing extraction target region;
And performing image processing on each image of the face region and the clothing region in the subject image to correct at least one of brightness, color, and gradation of each image. Method.   The correlation items include the sex of the subject, the color or texture of the subject's clothing, the presence or absence of a related accessory used simultaneously with a specific type of clothing, the type of the subject's hairstyle, the shooting time of the subject, and The image correction method according to claim 1, wherein the image correction method is at least one of events when the subject is photographed.   The step of setting the clothing extraction target region is a step of adjusting and setting the clothing extraction target region to a size corresponding to the size of the face extraction target region or the face region. Item 3. The image correction method according to Item 1 or 2. The clothes shape data includes two types of data corresponding to the standing position and the sitting position of the subject,
The step of acquiring the clothing shape data is a step of selecting and acquiring one of the two types of data according to the relative position of the face extraction target region or the face region with respect to the subject image. The image correction method according to claim 1, 2 or 3. Face extraction target region setting means for setting a local region including the face of the subject specified by the operator in a subject image including a subject that is a person as a face extraction target region;
A face area extracting means for extracting a face area representing the face from the area based on image information of the face extraction target area;
Correlation information acquisition means for acquiring correlation information representing the content of at least one correlation item correlated with the type of clothing of the subject;
Based on the correlation, clothes type specifying means for specifying the type of clothes having the highest correlation with the content of the correlation item represented by the acquired correlation information;
Out of a plurality of clothing shape data each representing the shape of each clothing for each type, clothing shape data acquisition means for acquiring clothing shape data corresponding to the specified clothing type;
Clothing extraction for setting, as a clothing extraction target region, a region having a shape represented by the acquired clothing shape data at a position where the clothing of the subject is determined by the face extraction target region or the position of the face region in the subject image Target area setting means;
Clothes area extraction means for extracting a clothes area representing clothes of the subject from the area based on image information of the clothes extraction target area;
And correction means for performing image processing on each image of the face area and the clothes area in the subject image to correct at least one of brightness, color, and gradation of each image. Image correction device.   The correlation items include the sex of the subject, the color or texture of the subject's clothing, the presence or absence of a related accessory used simultaneously with a specific type of clothing, the type of the subject's hairstyle, the shooting time of the subject, and The image correction apparatus according to claim 5, wherein the image correction apparatus is at least one of events when the subject is photographed.   6. The clothes extraction target area setting means adjusts and sets the clothes extraction target area to a size corresponding to the size of the face extraction target area or the face area. Or the image correction apparatus of 6. The clothes shape data includes two types of data corresponding to the standing position and the sitting position of the subject,
The clothing shape data acquisition means selects and acquires one of the two types of data according to the relative position of the face extraction target area or the face area with respect to the subject image. The image correction apparatus according to claim 5, 6 or 7. Computer
Face extraction target region setting means for setting a local region including the face of the subject specified by the operator in a subject image including a subject that is a person as a face extraction target region;
A face area extracting means for extracting a face area representing the face from the area based on image information of the face extraction target area;
Correlation information acquisition means for acquiring correlation information representing the content of at least one correlation item correlated with the type of clothing of the subject;
Based on the correlation, clothes type specifying means for specifying the type of clothes having the highest correlation with the content of the correlation item represented by the acquired correlation information;
Out of a plurality of clothing shape data each representing the shape of each clothing for each type, clothing shape data acquisition means for acquiring clothing shape data corresponding to the specified clothing type;
Clothing extraction for setting, as a clothing extraction target region, a region having a shape represented by the acquired clothing shape data at a position where the clothing of the subject is determined by the face extraction target region or the position of the face region in the subject image Target area setting means;
Clothes area extraction means for extracting a clothes area representing clothes of the subject from the area based on image information of the clothes extraction target area;
A program for performing image processing on each image of the face area and the clothes area in the subject image to function as correction means for correcting at least one of brightness, color, and gradation of each image.   The correlation items include the sex of the subject, the color or texture of the subject's clothing, the presence or absence of a related accessory used simultaneously with a specific type of clothing, the type of the subject's hairstyle, the shooting time of the subject, and The program according to claim 9, wherein the program is at least one of events when the subject is photographed.   10. The clothes extraction target area setting means adjusts and sets the clothes extraction target area to a size corresponding to the size of the face extraction target area or the face area. Or the program of 10. The clothes shape data includes two types of data corresponding to the standing position and the sitting position of the subject,
The clothing shape data acquisition means selects and acquires one of the two types of data according to the relative position of the face extraction target area or the face area with respect to the subject image. The program according to claim 9, 10 or 11.