JP4875470B2 - Color correction apparatus and color correction program - Google Patents

Description

  The present invention relates to a color correction apparatus and a color correction program that perform color correction of an image represented by image data.

  In recent years, techniques for capturing color image information as digital image data, such as digital cameras and scanners, have been developed, and software techniques for processing the obtained image data have been actively developed along with the development.

  As one of such software technologies, there is a technology that improves the appearance of an image by performing color correction of the image (see, for example, Patent Document 1). In general, in an image represented by image data obtained by a digital camera or a scanner, a hue may deviate from that desired by a user. Therefore, the image data obtained by the digital camera or the scanner can be brought close to the image desired by the user by performing the above color correction processing. In particular, in the case of a person image, such a color correction process is important because the person's impression received from the image differs considerably depending on the appearance of the person's face in the image.

In addition, when editing multiple images of each person, such as when creating a graduation album, if the background color is different for each image, the unsightly image group lacking a sense of unity It is easy to become. Therefore, recently, a main subject is detected using a technique (for example, see Patent Document 2) that detects a main subject (for example, a face region) in an image, and color correction corresponding to the main subject is performed. A technique for unifying background colors with an image area other than the subject as the background (see, for example, Patent Document 3), or contrary to Patent Document 3, the background is detected first, and the image area other than the background is used as the face area. A technique has been proposed in which the entire image is color-corrected with a correction amount obtained by weighted addition of the color correction amount and the face region color correction amount (see, for example, Patent Document 4).
Japanese Patent Laid-Open No. 7-210658 JP 2006-244385 A JP 2006-92127 A JP 2006-139368 A

  In the method described in Patent Document 3, a face detection technique for finding an image area that reflects a general feature of a face is applied when identifying a face area. However, with the current face detection technology, it is difficult to accurately detect the face position including the contour, and an image area (typically a rectangular area) around the face where the face is supposed to exist is specified. It remains to the extent. For this reason, even if the current face detection technology is applied as it is, it is difficult to perform appropriate color correction on both the face and background of a person.

In the method described in Patent Document 4, a reference background color (for example, a blue background color) is registered in advance, and if the color of the image area is close to the reference background color, the image area is the background area. It is determined. However, in general, in addition to a person's face, a part of the person's clothes often appears in the image, and in the method described in Patent Document 3, the color of the clothes in the image is significantly different from the reference background color. In this case, the clothes portion is also regarded as a face region and color correction is performed so as to match the face color. For this reason, even with the method described in Patent Document 4, it is difficult to perform appropriate color correction on both the face and background of a person.

  In view of the above circumstances, an object of the present invention is to provide a color correction apparatus and a color correction program that perform appropriate color correction on both a face and a background in an image.

In order to achieve the above object, the color correction apparatus of the present invention provides:
A face area detection unit for detecting a person's face area for each of a plurality of images each containing a person;
For each of the plurality of images, a face color correction unit that performs color correction according to each image such that the face region detected by the face region detection unit has a predetermined desired color;
For each of the plurality of images, a background region existing in a predetermined positional relationship with the face region or the entire image detected by the face region detection unit is obtained, and the color of the background region is unified among the plurality of images. A background color unification unit that applies color correction to each image,
A mask that indicates how much each point in the image belongs to the background or the face is determined based on the proximity of the color between each point and the background region and the proximity between each point and the face region. Based on each of the plurality of images, a mask creation unit,
Using the mask created by the mask creation unit, a first correction image obtained by correction by the face color correction unit and a second correction image obtained by correction by the background color unification unit, A point that strongly belongs to the face is provided with a combining unit that combines the components so that the component of the first corrected image is strong, and the point that strongly belongs to the background is that of the second corrected image.

  In the color correction apparatus according to the present invention, the first correction image and the second correction image are strongly attributed to the face, the first correction image component is strong, and the background is strongly attributed to the second. Since the components of the corrected image are combined so as to be strong, the color of the face region is close to the predetermined desired color, and the color of the background region is unified among the plurality of images. Is created. For this reason, in the color correction apparatus of the present invention, appropriate color correction is performed for both the face and the background in the image.

  In the color correction apparatus of the present invention, “the background color unifying unit is unified among the plurality of images based on the background region on the first corrected image obtained by the correction by the face color correcting unit. A form in which a unified background color to be obtained is obtained and color correction is performed on each of the plurality of images so that the color of the background region becomes the unified background color is a preferable form.

  According to such a form, since the unified background color is obtained based on the color of the background region in the image after the color of the face region is corrected, the color of the face region and the background in the synthesized image The color of the area is balanced.

  Further, in the color correction apparatus of the present invention, “the mask creation unit determines the distance in the predetermined color space between the color of the background region and the color of each point as the closeness of the color between each point and the background region. The form “used” is also a preferred form.

  According to such a form, it is easy to quantitatively represent how much each point in the image belongs to the background.

  In the color correction apparatus of the present invention, “the mask creating unit uses a distance on the image between the center point of each face area and each point as the proximity of the position between each point and the face area. The form “is” is also a preferable form.

  According to such a form, it is easy to quantitatively represent how much each point in the image belongs to the face region.

In order to achieve the above object, the color correction program of the present invention provides:
Built into a computer system, on that computer system,
A face area detection unit for detecting a person's face area for each of a plurality of images each containing a person;
For each of the plurality of images, a face color correction unit that performs color correction according to each image such that the face region detected by the face region detection unit has a predetermined desired color;
For each of the plurality of images, a background region existing in a predetermined positional relationship with the face region or the entire image detected by the face region detection unit is obtained, and the color of the background region is unified among the plurality of images. A background color unification unit that applies color correction to each image,
A mask that indicates how much each point in the image belongs to the background or the face is determined based on the proximity of the color between each point and the background region and the proximity between each point and the face region. Based on each of the plurality of images, a mask creation unit,
Using the mask created by the mask creation unit, a first correction image obtained by correction by the face color correction unit and a second correction image obtained by correction by the background color unification unit, A point that strongly belongs to the face has a strong component of the first correction image, and a point that strongly belongs to the background has a composition unit that composes so that the component of the second correction image becomes strong.

  By executing the color correction program of the present invention in a computer system, the color correction apparatus of the present invention can be easily realized.

  Note that the color correction program according to the present invention is shown only in its basic form here, but this is merely for avoiding duplication, and the color correction program according to the present invention includes only the above basic form. Instead, various forms corresponding to the respective forms of the color correction apparatus described above are included.

  Further, the elements such as the synthesis unit that the color correction program of the present invention constructs on the computer system may be one element constructed by one program part, and one element by a plurality of program parts. It may be constructed, or a plurality of elements may be constructed by one program part. In addition, these elements may be constructed so as to execute such an action by themselves, or may be executed by giving instructions to other programs and program components incorporated in the computer system. It may be constructed.

  According to the present invention, it is possible to perform appropriate color correction for both the face and the background in the image.

  Hereinafter, embodiments of the present invention will be described.

FIG. 1 is an overall configuration diagram of an image processing system to which an embodiment of the present invention is applied.
This image processing system is a system to which DTP (DeskTop Publishing) for creating a printed matter using a computer is applied. The user is a personal computer 100 and arranges an image group composed of a plurality of images in which a plurality of persons are captured. This is a system for editing and outputting with a color printer 2170. In the personal computer 100, read image data of a color person image generated by reading a document image with a color scanner (not shown), or color image representing a person shot with a digital camera (not shown). Image data or the like is input as image editing material. An image group 210 composed of images input to the personal computer 100 is an image group mainly for album editing. In any image, a person in the image faces a front with a substantially monochromatic background in the background. It is a color image. These image groups 210 are color-corrected so that the personal computer 100 can perform color correction of the face portion in each image and also obtain a sense of unity in the entire image group with respect to the background. Image data representing the image group 210A after color correction is converted into raster data that can be output by the color printer 2170 in a RIP (Raster Image Processor) (not shown) in the personal computer 100. The converted raster data is input to the color printer 2170, and the color printer 2170 creates an output image 210B based on the input raster data.

  As described above, in the image processing system shown in FIG. 1, the final output image group 210 </ b> B is created after the color correction processing of the image is performed by the personal computer 100. A feature of the image processing system as an embodiment of the present invention lies in the content of color correction processing executed by the personal computer 100, and the personal computer 100 is an embodiment of the color correction apparatus of the present invention. Operate. Hereinafter, the personal computer 100 will be described in detail.

  2 is an external perspective view of a personal computer that operates as an embodiment of the color correction apparatus of the present invention, and FIG. 3 is a hardware configuration diagram thereof.

  The personal computer 100 includes a main body 110 incorporating a CPU, a RAM memory, a hard disk, and the like which will be described later, an image display device 120 that displays a screen on the display screen 121 according to an instruction from the main body 110, and a user's computer in the personal computer 100. A keyboard 130 for inputting instructions and character information and a mouse 140 for inputting an instruction corresponding to the position by designating an arbitrary position on the display screen 121 are provided.

  The main body 110 further has an FD loading slot 111 into which a flexible disk (hereinafter abbreviated as FD) is loaded, and a CD-ROM loading slot 112 into which a CD-ROM is loaded. An FD drive and a CD-ROM drive (to be described later) for driving an FD and a CD-ROM are also incorporated.

  As shown in FIG. 3, the main body 110 includes a CPU 211 that executes various programs, a main memory 212 that reads programs stored in the hard disk device 213 and develops them for execution by the CPU 211, and various programs. And a hard disk device 213 and FD 2140 in which data and the like are stored are loaded, an FD drive 214 and a CD-ROM 2150 that access the loaded FD 2140 are loaded, and a CD-ROM drive 215 that accesses the loaded CD-ROM 2150 Is built-in. These various elements and the image display device 120, the keyboard 130, and the mouse 140 shown in FIG. 2 are also connected to each other via a bus 1200. Further, a color printer 2170 for outputting an image is provided outside the personal computer 100, and an output interface 217 is provided to the personal computer 100 in order to send output data from the personal computer 100 to the color printer 2170. Is built-in. The personal computer 100 also includes an I / O interface 216 that can be connected to an external device such as a scanner (not shown) to input image data. The connected external device is also controlled via the I / O interface 216. The output interface 217 and the I / O interface 216 are also connected to each other through the bus 1200 with the above-described various elements constituting the hardware inside the personal computer 100.

  In this personal computer 100, the user of the personal computer 100 can input image data to be processed from the I / O interface 216 and save it in the hard disk device 213, and the FD 2140 in which the image data is recorded, The data can also be stored in the hard disk device 213 from the CD-ROM 2150 via the FD drive 214 and the CD-ROM drive 215, respectively.

  Next, an embodiment of the color correction program of the present invention will be described.

  When one embodiment of the color correction program of the present invention is stored in, for example, the CD-ROM 2150, when the CD-ROM 2150 is loaded into the main body 110 from the CD-ROM loading slot 112, the CD-ROM 2150 is stored. The color correction program stored in the hard disk device 213 of the personal computer 100 is installed by the CD-ROM drive 215. When the color correction program installed in the hard disk device 213 is activated, the personal computer 100 operates as an embodiment of the color correction device of the present invention.

  FIG. 4 is a diagram showing an embodiment of the color correction program of the present invention.

  Here, this color correction program 900 is stored in the CD-ROM 2150. The storage medium for storing the color correction program 900 according to the present invention is not limited to the CD-ROM 2150 shown in FIG. 4, but also the hard disk device 213 and FD 2140 shown in FIG. 3, not shown in FIGS. Various storage media such as DVD and MO can be employed.

  The color correction program 900 is executed in the personal computer 100 shown in FIGS. 2 and 3 and operates the personal computer 100 as one embodiment of the present invention as described above. A face detection unit 21, a temporary correction unit 22, a target background color calculation unit 23, a unified background image creation unit 24, and an image composition unit 25.

  Detailed contents of each element of the color correction program 900 will be described later.

  FIG. 5 is a diagram showing an outline of elements constructed on the personal computer for operating the personal computer shown in FIGS. 2 and 3 as an embodiment of the color correction apparatus of the present invention, and actions by these elements. It is.

  When the color correction program 900 shown in FIG. 4 is installed in the personal computer 100 shown in FIGS. 2 and 3, the image data acquisition unit 10, the face detection unit 11, the temporary correction unit 12, and the target background color calculation unit shown in FIG. 13, each element of the unified background image creation unit 14 and the image composition unit 15 is constructed on the personal computer 100. With these elements, the personal computer 100 operates as an embodiment of the color correction apparatus of the present invention. The image data acquisition unit 10, the face detection unit 11, the temporary correction unit 12, the target background color calculation unit 13, the background unified image creation unit 14, and the image synthesis unit 15 acquire image data in the color correction program 900 illustrated in FIG. 4. The unit 20, the face detection unit 21, the temporary correction unit 22, the target background color calculation unit 23, the background unified image creation unit 24, and the image composition unit 25 are each constructed on the personal computer 100. Therefore, each element in FIG. 4 corresponds to each element in FIG. 5, and each element in FIG. 5 includes the hardware of the personal computer 100 shown in FIGS. 2 and 3, and the OS and application executed on the personal computer 100. 4 is different from the first embodiment in that each element of the color correction program 900 shown in FIG. 4 is configured only by an application program. Here, the face detection unit 21 corresponds to an example of a face area detection unit according to the present invention, and the temporary correction unit 22 corresponds to an example of a face color correction unit according to the present invention. The combination of the target background color calculation unit 23 and the unified background image creation unit 24 corresponds to an example of the unified background color unit according to the present invention, and the image synthesis unit 25 combines with the mask creation unit according to the present invention. This corresponds to an example of serving as a part. Below, each element shown in FIG. 5 and the outline of the effect | action by each element are demonstrated.

  The image data acquisition unit 10 shown in FIG. 5 shows, as shown in FIG. 3, an image data group made up of a group of image data each representing an image of a person before color correction as an object of color correction processing based on a user instruction. Obtained from the hard disk device 213. Hereinafter, an image (image data) before color correction is referred to as an original image (original image data) in order to distinguish it from an image after color correction. The face detection unit 11 detects a face area based on the original image data read by the image data acquisition unit 10, and the temporary correction unit 12 corrects the detected face area to have an appropriate color. Color correction is applied to the entire area on the original image by the amount. The target background color calculation unit 13, which is a subsequent processing unit of the face detection unit 11, uses the target background color as a unified background color for the entire image group based on the image data group after color correction is performed by the temporary correction unit 12. The target background color is calculated. The background unified image creation unit 14 performs color correction processing on the original image so as to realize the calculated target background color. The image compositing unit 15, which is a processing unit located downstream of both the temporary correction unit 12 and the background unified image creation unit 14, is color-corrected by the image corrected by the temporary correction unit 12 and the background unified image creation unit 14. Is combined with the selected image. At the time of this composition, the image composition unit 15 emphasizes the face portion in the image color-corrected by the temporary correction unit 12 and the background in the image color-corrected by the background unified image creation unit 14. Thus, two images are synthesized.

  Next, the flow of color correction processing by the computer 100 will be described using a flowchart.

  FIG. 6 is a flowchart showing the operation of color correction processing by the personal computer 100.

  When color correction processing is performed, original image data groups each representing an original image group designated by the user as a target for color correction processing are acquired by the image data acquisition unit 10 of FIG. 5 (step S1). . The original image group represented by such an original image data group is, for example, all original images existing in a folder specified by the user, or an original image existing in a file having a PDF data format specified by the user. Such as a group. In the following description, it is assumed that such an original image group consists of N original images in total. In addition, the processes in steps S2 to S5 described below are performed in order from the first original image data to the Nth original image data for N pieces of original image data. The processing of the second image data will be described as an example.

  The face detection unit 11 reads the i-th original image data from the N pieces of original image data (step S2). Based on the i-th original image data, the face detection unit 11 detects the person represented by the original image data. A face area is detected (step S3). In this face detection, the face detection technique described in Patent Document 2 is applied. In general, in the face area, the right eye is darker than the cheek, the left eye is darker than the eyebrows, and the nostril is darker than the cheek. In the personal computer 100, reference data characterizing such a face area is stored in advance, and the face detection unit 11 refers to the reference data, and the feature of the face area appears in the square detection area in the image. It is determined whether the detection area is a face area or a non-face area. The face detection unit 11 performs such determination while changing the position of the detection area, and detects the face area in the image. Further, the face detection unit 11 detects such a face region for each reduced image while reducing the image size at a predetermined rate. By detecting the face area of the reduced image in this way, the face area can be detected when the face size is reduced to the same extent as the size of the detection area. By this detection, the image before reduction is detected. The face area at is determined. By detecting the face area, information on the size and position of the face area (center position of the face area) is acquired.

  However, the face area detected in this way is a square area including the face, and the face position including the face outline is not accurately obtained. As will be described later, the personal computer 100 has a function for supplementing such a face detection function. By using both the function and the face detection function, both the face and the background in the image are used. Appropriate color correction can be performed.

  Face area data having face area information detected by the face detection unit 11 is input to the temporary correction unit 12, the target background color calculation unit 13, and the image synthesis unit 15 shown in FIG.

  The temporary correction unit 12 obtains a correction amount for making the color of the center of the face area of the i-th original image a desirable color as the color of the face part based on the face area data, and the i-th original with the correction amount Perform color correction on the entire image. In this color correction, the face portion is moderately color-corrected, but this color correction does not consider whether the color of the image other than the face portion in the original image is appropriate. In order to distinguish from background color correction described later, the correction performed by the temporary correction unit 12 is hereinafter referred to as temporary correction. By this temporary correction, a temporary corrected image in which the face portion is appropriately corrected is created based on the i-th original image data (step S4). Then, provisionally corrected image data representing the created provisionally corrected image is input to the target background color calculation unit 13 and the image composition unit 15 shown in FIG. As described above, the target background color calculation unit 13 also receives the face area data from the face detection unit 11, and the target background color calculation unit 13 calculates the background color of the temporarily corrected image based on the face area data. (Step S5). Here, the calculation of the background color will be specifically described.

  FIG. 7 is a diagram showing an area where the background color is calculated.

  FIG. 7 shows a face area 70b which is a square area having a vertical width 2r and a horizontal width 2r. The target background color calculation unit 13 is located on both sides (left side and right side) of the face area 70b based on the position and size information of the face area 70b represented by the face area data. Two rectangular background color calculation areas 70c having a width of 3r and a width of 3r / 4 are obtained. The bottom side of the background color calculation area 70c having the vertical width 3r and the horizontal width 3r / 4 is at the same height as the bottom side of the face area 70b in the vertical direction in the figure. Of the two sides extending in the vertical direction in the figure in each of the background color calculation areas 70c, the side farther from the face area 70b is 2r away from the center of the face area 70b in the horizontal direction in the figure. . The target background color calculation unit 13 calculates the average value of the pixel values for the R, G, and B color components in the two background color calculation regions 70c, and stores the average value. This average value is regarded as the background color of this image. In this way, by obtaining the background color from the background color calculation area 70c located at a predetermined distance from the face area 70b using the size of the face area 70b as a reference of the distance, it is influenced by the color of the face area 70b. An average background color can be obtained. It should be noted that the area for which the background color is obtained needs only to be sufficiently away from the face area, and the shape thereof may be other than a rectangle. For example, the area for obtaining the background color may be an area on one side instead of an area on both sides of the face area, or may be located above or obliquely above the face area. Or the whole circumference | surroundings surrounding a face area | region may be sufficient. Further, the position of the area for obtaining the background color does not necessarily have to be a position separated from the face area by a predetermined relative distance based on the size of the face area. For example, a small predetermined area located in the upper left or upper right of the image that cannot normally overlap an image portion representing a person in the image may be used as the background color calculation area, or the four edges ( The background color may be calculated from a pixel group at the top, bottom, left, and right edges.

  Returning to the flowchart of FIG. 6, the description of the operation of the color correction processing by the personal computer 100 is continued.

  After the background color of the temporarily corrected image is calculated in step S5, the i-th original image data that is the source of the temporarily corrected image is the last original image data that has not been temporarily corrected. It is determined whether it is (that is, the Nth original image data) (step S6). When the i-th original image data is not the N-th original image data (step S6; Yes), the processing target is set to the (i + 1) -th image data (step S7), and the above-described steps S2 to S5 are performed. Processing is performed on the (i + 1) th original image data. When the i-th original image data is the N-th original image data (step S6; No), the target background color calculation unit 13 calculates the average value of the pixel values of the background colors of the N temporarily corrected images. Calculate (step S8). This average value is the pixel value of the target background color (target background color) for unifying the background colors of the N images. The pixel value data of the target background color is the background unified image of FIG. 5 together with the data of the position of the background color calculation area 70c of FIG. 7 used in the background color calculation of step S5 for each of the N images. Input to the creation unit 14. In FIG. 5, the pixel value data of the target background color and the data of the position of the background color calculation area 70c are shown together as target background color data. The processing of step S9 to step S11 described below after step S8 is performed in order from the first original image data to the Nth original image data for N pieces of original image data. Now, an i-th image data process will be described as an example.

  The unified background image creation unit 14 reads out the i-th original image data from the N pieces of original image data (step S9), and the original image represented by the i-th original image data in FIG. An image area located at the same position as the background color calculation area 70c is obtained. Then, the background unified image creation unit 14 obtains a correction amount for the average pixel value of the image region to be the pixel value of the target background color, and performs color correction of the entire i-th original image with the correction amount (step) S10). As described above, the input image group is mainly an image group for album editing, and the background color of any image is almost a single color (for example, the color of the wall behind the person). For this reason, the background color calculation area 70c used for calculating the background color is only a partial area in the entire background, but the entire background is corrected to the target background color by the color correction in step S10. A unified background image will be created. The background unified image data representing the unified background image is input to the image composition unit 15 shown in FIG. Here, i-th original image data and provisionally corrected image data representing a provisionally corrected image are also input to the image composition unit 15, and the image composition unit 15 sets a composition mask based on the i-th original image. Using the composite mask, the face portion of the temporary correction image and the image portion excluding the face portion in the unified background image are combined (step S11). Hereinafter, this image composition will be specifically described.

As described above, the face composition data created by the face detection unit 11 is input to the image composition unit 15. First, based on the face area data of the i-th original image, the image composition unit 15 determines the face area average pixel value that is the average of the pixel values for the R, G, and B color components in the face area for this original image. (Rf, Gf, Bf) and the pixel value of the background color in the original image (that is, the average pixel value of the background color calculation area in the original image) are obtained. Hereinafter, the pixel value of the background color in the original image is referred to as the original image background pixel value (Rb, Gb, Bb). Next, the image composition unit 15 obtains an area average chromaticity distance d ₀ between the face area average pixel value (Rf, Gf, Bf) and the original image background pixel value (Rb, Gb, Bb). This area average chromaticity distance d ₀ is obtained according to the following equation.
d ₀ ² = (Rf−Rb) ² / Rmax ² + (Gf−Gb) ² / Gmax ²
+ (Bf−Bb) ² / Bmax ² (1)
Here, Rmax represents the larger value of Rf and Rb (if it is equal, the value thereof). Similarly, Gmax and Bmax are also the larger value of Gf and Gb (if they are equal). Represents the larger value of Bf and Bb (or the value if equal).

Next, the image composition unit 15 represents the pixel value (R (x, y) of the pixel located at the coordinate point (x, y) when the pixel position in the image is represented by two-dimensional coordinates (x, y). , G (x, y), B (x, y)) and the chromaticity distance d (x, y) between the original image background pixel values (Rb, Gb, Bb) described above. This chromaticity distance is obtained in the same manner as in equation (1). This chromaticity distance d (x, y) is, by definition, a numerical value that is substantially close to zero in an image region close to the background color, and becomes a larger value as the hue differs greatly from the background color. Here, when there is a pixel having a chromaticity distance d (x, y) larger than the area average chromaticity distance d ₀ , the chromaticity distance d (x, y) is further set to the area average color. It is subjected to a process for replacing the time distance d _0. That is, the upper limit of the difference between the face area color and the background color is determined.

  FIG. 8 is a diagram schematically showing how a composite mask is created.

  Part (a) in FIG. 8 is a drawing rule in which a pixel having a larger value of chromaticity distance d (x, y) is whiter (that is, a pixel having a smaller value of chromaticity distance d (x, y) is blacker). This is a schematic representation of the chromaticity distance d (x, y). As shown in part (a) of FIG. 8, the face portion 301 and the clothing portion 303 are whitish because the value of the chromaticity distance d (x, y) is large. A background portion 302 that is an image region other than the portion 303 is a dark area indicated by a horizontal line.

  Next, the image synthesizing unit 15 selects a portion of the image where the change in the value of the chromaticity distance d (x, y) is large so that the change in the value of the chromaticity distance d (x, y) becomes gentle. Processing to level the value is performed. Further, the image composition unit 15 cuts a small value of the chromaticity distance d (x, y) using the minimum value filter after the above chromaticity distance d (x, y) (the value is set to zero). Process). These processes are performed by scanning the image using a spatial filter having a size of about 1/8 of the size of the face area, that is, about r / 4 in FIG. With these processes, the boundary line is blurred in the diagram representing the chromaticity distance d (x, y).

  Part (b) of FIG. 8 shows a state of the chromaticity distance d ′ (x, y) after the minimum value filter is applied. As shown in this figure, in part (b) of FIG. 8, the boundary is ambiguous compared to part (a) of FIG.

Next, the image composition unit 15 acquires the center coordinates of the face area and the size of the face area from the face area data, and defines this information and a two-dimensional plane previously stored in the computer 100. In combination with a monotonically decreasing function for distance, the weight function is such that the vicinity of the center coordinate of the face area is 1, the distance from the center coordinate of the face area becomes smaller, and it approaches 0 on the outside of the face area. α (x, y) is determined. Then, using this weight function α (x, y), a function mask (x, y) that uses a position in the image as a variable and gives a composite mask is defined by the following equation.
mask (x, y) = α (x, y) × d ′ (x, y) / d ₀
Here, the region average chromaticity distance d ₀ is used for normalization of the chromaticity distance d (x, y), and the normalized chromaticity distance d ′ (x, y) / d ₀ is As can be seen from part (b) of FIG. 8, in the face part and clothes part, a large value (almost 1) is obtained, but for this normalized d ′ (x, y) / d ₀ , Furthermore, by applying the weight function α (x, y), only the face portion mainly remains as a region having a large value.

  Part (c) of FIG. 8 shows that the function mask (x, y) is whiter for pixels having a larger value of the function mask (x, y) (that is, a pixel having a smaller value of the function mask (x, y) is black). FIG. As shown in this figure, by weighting with a weighting function α (x, y), only the face portion is mainly extracted as a whitish region.

As described above, in this embodiment, the original image data is used to obtain the normalized chromaticity distance d ′ (x, y) / d ₀ and the weight function α (x, y) is determined (face Face area data is used to obtain the center position of the image. By using both data in this way, more appropriate color correction is performed for both the face and background in the image than when color correction is performed using only face area data obtained by face detection. Will be given.

The image composition unit 15 emphasizes the face portion extracted from the temporary correction image mainly in the image portion of the face portion, and the image region other than the face portion other than the face portion extracted mainly from the background unified image. The images are synthesized so that the image portion of the image is emphasized. Specifically, the pixel value (each RGB value) of the pixel located at the coordinates (x, y) in the composite image, for example, for the R component is determined as follows.
{1−mask (x, y)} × RB (x, y) + mask (x, y) × RF (x, y)
Here, in the above equation, the R component of the pixel value of the pixel located at the coordinate (x, y) in the temporary correction image and the R component of the pixel value of the pixel located at the coordinate (x, y) in the background unified image Are represented by RF (x, y) and RB (x, y), respectively. The G component and the B component are also determined by the same equation as the above equation.

  In this way, by combining the face part of the temporarily corrected image and the background part of the unified background image, the color of the face part is corrected, and the background is also colored so that a sense of unity can be obtained in the entire image group. Correction will be performed.

  Returning to the flowchart of FIG. 6, the description of the operation of the color correction processing by the personal computer 100 is continued.

  After the provisional correction image and the background unified image are combined in step S11, the i-th original image data that is the basis of the combined image is the last original image that has not been provisionally corrected. It is determined whether it is data (that is, the Nth original image data) (step S12). If the i-th original image data is not the N-th original image data (step S12; Yes), the processing target is set to the (i + 1) -th image data (step S13), and the above-described steps S9 to S11 are performed. Processing is performed on the (i + 1) th original image data. When the i-th original image data is the N-th original image data (step S12; No), the color correction of all the image groups is completed.

  Next, a screen displayed to the user during the color correction process will be described.

  FIG. 9 is a diagram showing a window when color correction is performed.

  When the user activates the color correction function of the computer 100 of FIG. 2, a color correction window 400 shown in part (a) of FIG. 9 is launched. The color correction window 400 displays an image group designation field 401 for designating an image group to be processed as shown in this figure. The user can specify the image group to be color-corrected by writing the name of the folder in which the image group is stored or the name of the file having the image group in the image group specifying field 401. Part (a) of FIG. 9 shows a state where an image group stored in the folder “org” existing in the folder “img” of the C drive is designated. Here, the user can display the image group by clicking the image display button 402 written as “Open” with the mouse 140 in FIG. 2.

  Part (b) of FIG. 9 shows a state in which an image group made up of images 501 to 506 stored in the folder “org” is displayed. On the right side of these image groups, a correction button 408, a re-correction button 409, and a manual mode button 410 are provided. By clicking the correction button 408, the user can cause these image groups to perform color correction processing according to the procedure described with reference to FIG. 6, and the image group displayed in the color correction window 400 is a color group. Switch to the corrected image group. The recorrection button 409 and the manual mode button 410 will be described later.

  In the above description, the image group to be subjected to color correction is a set of images stored in the folder. However, in this computer 100, color correction processing can also be performed on the image group in the file. . In this case, the user can designate a file by writing the file name in the image group designation field 401. Furthermore, the user can also click all page radio buttons 403 by clicking the mouse 140 to set all image groups in the specified file as image groups to be processed. In addition, an image group can be specified by entering a page having an image group to be processed in the page column 405. When a page is designated in this way, a black mark is automatically entered in the page designation radio button 404.

  FIG. 10 is a diagram showing a color correction window when all image groups in a file are designated as image groups to be processed.

  This figure shows a state where the file “org.pdf” existing in the folder “img” of the C drive is designated in the image group designation column 401. Here, the file “org.pdf” is a PDF file having a plurality of pages, and each page includes an image group in which images of a plurality of persons are arranged. In this figure, the all page radio button 403 is selected (a black mark is entered), and all the image groups in the color correction file “org.pdf” are designated as color correction targets. Has been. When the image display button 402 is clicked with the mouse 140 in FIG. 2 in this state, all image groups in the file “org.pdf” are displayed in the same manner as in part (b) of FIG. Then, the user can perform color correction on the displayed image group in the same manner as described above.

  FIG. 11 is a diagram showing a color correction window when an image group existing on the second page of a file is designated as an image group to be processed.

  In this figure, the page designation radio button 404 is selected (a black mark is entered), and “2” representing the second page of the file is entered in the page column 405. By this designation, the user can designate the image group existing on the second page of the file as the color correction target image group. When this page is designated, the user may directly enter a number in the page field 405, or by clicking the next page button 406, the page is increased by one page to the target page. Also good. By clicking the previous page button 407, it is possible to return to the previous page. When page designation is performed as in the case of all page designation, the image display button 402 is clicked with the mouse 140 in FIG. 2, and the image group in the page is displayed in the same manner as part (b) in FIG. The Even when the image is displayed, the user can switch the displayed image group by changing the page and clicking the image display button 402, and can select the image group for which color correction is desired. Then, by clicking the correction button 408, color correction processing is performed on the specified page image group.

  In the color correction described above, the user simply specifies the image group and clicks the correction button 408, and color correction is performed on all the images constituting the specified image group. In most cases, the color correction is performed so that the color of the face region is optimized for each image and the background color is uniform for all images by color correction of all the images. However, rarely, as a result of erroneously detecting a face area, an image that is not subjected to color correction as intended by the user, or an image that cannot be detected and that is not subjected to color correction may appear. is there. The computer 100 has a function that allows the user to manually specify a face area and a background color calculation area for such an image. Hereinafter, this function will be described. Hereinafter, the color correction for the entire image group, which has been described so far, is referred to as “all image color correction” in order to distinguish it from the color correction of individual images described later.

  First, a case will be described in which the face color is not corrected as intended by the user as a result of erroneously detecting the face area. In the following, as an example, the user finds that the color correction of the face of the upper left image 501 is not sufficient from the image group displayed by performing all image color correction in part (b) of FIG. explain.

  The user who has found that the face color correction of the image 501 in the upper left is not sufficient, clicks the area information radio button 501a provided near the image 501 with the mouse 140 in FIG. Can enter.

FIG. 12 is a diagram showing a state of an image when the region designation mode is entered.
In the image 501 that has entered the area designation mode, a rectangular (square) face area 501b and a rectangular background color calculation area 501c located on both sides of the face area 501b, which are detected at the time of all image color correction, are included. Displayed as default. At this time, these two types of rectangular frames are displayed in different colors so that the user can easily identify these two types of regions. In the drawing, in order to show that the frame color is different between the face area 501b and the background color calculation area 501c, the frame of the face area 501b is represented by a dotted line, and the frame of the background color calculation area 501c is displayed by a solid line. ing. The user can change the position and size of the face area 501b and the background color calculation area 501c with the mouse 140 of FIG. After the face area 501b and the background color calculation area 501c having appropriate positions and sizes are designated, the user clicks on the recorrection button 409 shown in part (b) of FIG. The color correction is performed again only on the performed image. In this re-correction, detection of the face area in step S3 in FIG. 6 and calculation of the background color by obtaining the background color calculation area in step S5 are omitted, and the manually specified face area 501b and background color calculation are omitted. Region 501c is used instead. For the target color calculation color, the background color correction is performed so as to match the target background color at the time of the first all image color correction without calculating the target color calculation color again. As described above, the color correction is performed again only on the image for which the manual area designation has been performed, because the color correction of the entire image group is performed for the few images that have failed the overall image color correction. This is because the processing efficiency is poor. In particular, when the number of images increases, as described above, it is more efficient to correct only the images that have failed to be corrected for the entire image as compared to performing color correction for the entire image group.

  Next, a case where a face cannot be detected during the entire image color correction will be described. In the following, in the part (b) of FIG. 9, a case where face detection of the upper left image 501 has failed during all image color correction will be described as an example.

  FIG. 13 is a diagram showing a color correction window when face detection has failed in some images in the image group during the color correction of all images.

  An image for which face detection has failed during color correction for all images is displayed with a red frame surrounding the image. In the figure, a state in which the periphery of the upper left image 501 in which face detection has failed is surrounded by a hatched frame representing the red frame. In order to notify the user that face detection has failed, a warning message “WARNING: face cannot be detected in red frame image” is displayed as shown in the upper left image 501. For the remaining images excluding the upper left image 501 for which face detection has failed, the above-described all-image color correction is performed using the image group including these remaining images as the image correction target image group. However, the upper left image 501 in which face detection has failed is in a state where color correction has not been performed due to face detection failure. The user recognizes that it is necessary to perform color correction again on the upper left image 501 by displaying the red frame and the warning message. Here, the user enters the area designation mode by clicking the area information radio button 501a provided near the image 501 with the mouse 140 of FIG. Can do.

  FIG. 14 is a diagram illustrating a state when an area designation mode is entered for an image for which face detection has failed.

  If the face detection fails in all image color correction, the face area and the background color calculation area are not specified. Therefore, even if the image for which face detection has failed is entered in the area designation mode, As shown in part (a), rectangular areas representing the face area and the background color calculation area are not displayed by default. In this state, the user can specify a face area using the mouse 140 of FIG. Part (b) of FIG. 14 shows a state when a user designates a face area. As shown in this figure, when the face area 501b ′ is designated, the red frame shown in part (a) of FIG. 14 disappears, while the background color calculation area 501c ′ is changed according to the designated face area. 7 is automatically displayed in the positional relationship described in FIG. Once the background color calculation area 501c 'is displayed, the user can change the position and size of the background color calculation area 501c' with his / her preference using the mouse 140 in FIG. Thus, when the face area 501b ′ and the background color calculation area 501c ′ having appropriate positions and sizes are finally designated, the user clicks on the recorrection button 409 shown in FIG. Similarly to the above, color correction is performed again only for an image for which manual region designation has been performed.

  As described above, by providing the manual designation mode for the face area, it is possible to perform appropriate color correction by re-correction even when the face area is erroneously detected or when face detection cannot be performed. Become.

  In this manner, in addition to the function for manually specifying the area of the face area and the background color calculation area, the computer 100 in FIG. A manual color correction function for performing color correction is also provided. Hereinafter, the manual color correction function will be described.

  When the user feels that the color of some of the images in the image group subjected to the entire image color correction is uncomfortable as compared with the other images, the color correction shown in part (b) of FIG. 9 is performed. By clicking the manual correction mode button 410 in the window 400, the manual correction mode can be entered. In the following, in part (b) of FIG. 9, as a result of the user examining the color condition of the image after all image color correction, the upper row is matched with the background color of the middle image 505 in the lower row. A case where it is determined to correct the background color of the middle image 502 and the background color of the lower right image 506 will be described as an example. As a case where the background color is not unified as intended by the user in this way, for example, a case where the background color is not uniform in one image can be considered.

  FIG. 15 is a diagram illustrating a state of the color correction window 400 when the manual correction mode button illustrated in part (b) of FIG. 9 is clicked.

  In performing manual color correction, first, the user selects and clicks the pixel P having a background color as a color correction target in the middle image 505 in the lower row with the mouse 140 of FIG. Next, the user selects the pixel Q and the pixel R whose colors are to be matched with those of the pixel P in the lower right image 506 and the upper middle image 502, which are subject to background color correction. Click with the mouse 140 in FIG. By such an operation, the pixel value of the pixel clicked first is designated as the target pixel value for color correction, and the pixel value of the pixel clicked after that is a color that matches the pixel value with the target pixel value. Used to calculate the correction amount. In executing the color correction, the user performs the first color correction in which the color correction is uniformly performed with the calculated correction amount for all the pixels in the image to be subjected to the color correction, and the correction amount calculation. Two types of color correction, ie, a second color correction in which only a pixel having a pixel value close to the pixel value of the pixel clicked for is corrected by a correction amount corresponding to the difference in pixel value from the target pixel value One of these can be selected.

  Of the two types of color correction, the first color correction is used for the color correction used to create the provisionally corrected image in step S4 in FIG. 6 and the background unified image in step S10 in FIG. Same as the color correction used. The first color correction is executed by clicking the overall correction button 411 displayed as “Align pixel values” in FIG. When the overall correction button 411 is clicked, the image 506 on the right side of the lower row indicates that the pixel value of the pixel Q in this image is the target pixel value (that is, the background pixel P in the middle image 505 in the lower row). The entire image is uniformly color-corrected with a correction amount such that the pixel value of At the same time, the image 502 in the middle of the upper stage is uniformly color-corrected with the correction amount so that the pixel value of the pixel R in the image becomes the target pixel value. This first color correction has an advantage that the entire image is corrected in a well-balanced manner, and is a color correction selected by a user who places emphasis on color balance.

  Of the two types of color correction, the second color correction is executed by clicking a local correction button 412 displayed as “Align pixel values of approximate colors” in FIG. When this local correction button 412 is clicked, only the pixel having a pixel value close to the pixel value of the pixel Q in the image 506 on the right side of the lower stage is the difference in pixel value from the target pixel value. At the same time, only the pixel having a pixel value close to the pixel value of the pixel R in this image is the difference between the pixel value and the target pixel value. Color correction is performed with a corresponding correction amount. Details of the second color correction will be described below. Here, a description will be given taking correction of the right image 506 in the lower row as an example.

Before the second color correction is selected, the target pixel value (the pixel value of the background pixel P in the middle image 505 in the lower row) and each pixel value in the right image 506 in the lower row Are expressed as coordinate values in the RGB color space as numerical values in the image data, but when the second color correction is performed, the pixel values expressed in the RGB color space are temporarily It is converted into a pixel value expressed in an HSL color space (a color space described with three degrees of freedom of hue H, saturation S, and lightness L). In the second color correction, the hue, saturation, and lightness are corrected in the same manner, and therefore, the lightness correction will be described below as an example. Assuming that the lightness of the target pixel value is Lp and the lightness of the pixel Q in the lower right image 506 is Lq, in the second color correction, first, the lightness Lp of the target pixel value and the lightness Lq of the pixel Q From this, the difference ΔLq = Lp−Lq of the brightness difference is acquired. Next, among the pixels in the image 506, a pixel having the lightness L that satisfies the following condition is detected.
| L−Lq | ≦ Δq
Here, Δq is a predetermined value stored in advance in the computer 100 of FIG. By this detection, only pixels having a brightness close to the brightness Lq of the pixel Q are picked up as correction targets. Then, the detected lightness L of the pixel is replaced with a new lightness L ′ determined by the following equation.
L ′ = L + ΔLq × f (| L−Lq |)
Here, f (| L−Lq |) is a predetermined monotonically decreasing function with the absolute value | L−Lq | of the brightness difference as an argument, and the value of this function is such that | L−Lq | Becomes 1, and when | L−Lq | becomes Δq, it becomes 0. The above correction expression indicates that a pixel having a lightness close to the lightness Lq of the pixel Q is subjected to a larger correction, and in a pixel having a lightness equal to the lightness Lq of the pixel Q, correction is performed with Lq which is the maximum correction amount. Will be. Such a predetermined monotone decreasing function can be adopted as long as it is a smooth monotonic decreasing function having the above-described properties. For example, the following functions can be adopted.
f (x) = (Δq−x) ^1/2 × (1 / Δq) ^1/2
The same correction as the brightness correction described above is executed for the remaining hue and saturation. The corrected pixel values in the image 506 are converted into coordinate values in the RGB color space. Based on the image data composed of the converted pixel values, the color-corrected image 506 is displayed.

  For the middle image 502 in the upper row, the second color correction is executed in the same manner as the color correction of the right image 506 in the lower row in FIG. 15, and the corrected image is displayed. The second color correction described above is particularly useful when the user wants to correct only a specific color (background color in the above example).

  As described above, the computer 100 in FIG. 2 includes two types of manual color correction means, that is, the first color correction and the second color correction. By including these color correction means, The result of all image color correction described with reference to FIG. 6 can be modified to suit the user's preference.

  In the manual color correction described above, one of the images in the image group gives the target color. However, other than such a color correction method, it has been prepared in advance. It is also possible to employ a method in which the user selects a target color from among a plurality of types of colors and performs color correction using the selected color as the target color.

1 is an overall configuration diagram of an image processing system to which an embodiment of the present invention is applied. 1 is an external perspective view of a personal computer that operates as an embodiment of a color correction apparatus of the present invention. FIG. 3 is a hardware configuration diagram of the personal computer whose external perspective view is shown in FIG. 2. It is a figure which shows one Embodiment of the color correction program of this invention. It is a figure showing the outline of the effect | action by those elements constructed | assembled on this personal computer in order to operate the personal computer shown in FIG. 2 and FIG. 3 as one Embodiment of the color correction apparatus of this invention. FIG. 4 is a flowchart showing an operation of color correction processing by the personal computer shown in FIGS. 2 and 3. It is a figure showing the area | region where calculation of a background color is performed. It is the figure which represented a mode that the synthetic | combination mask was created. It is a figure showing the window at the time of color correction being performed. It is a figure showing the color correction window when all the image groups in a file are designated as an image group used as a candidate for processing. It is a figure showing the color correction window when the image group which exists in the 2nd page of a file is designated as an image group used as a process target. It is a figure showing the mode of the image when it entered into area | region designation | designated mode. FIG. 10 is a diagram illustrating a color correction window when face detection fails in some images in an image group during full image color correction. It is a figure showing a mode when it enters into area specification mode about an image in which face detection failed. FIG. 10 is a diagram illustrating a state of a color correction window 400 when a manual correction mode button illustrated in part (b) of FIG. 9 is clicked.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Image data acquisition part 11 Face detection part 12 Temporary correction part 13 Target background color calculation part 14 Background unified image creation part 15 Image composition part 20 Image data acquisition part 21 Face detection part 22 Temporary correction part 23 Target background color calculation part 24 Background Unified image creation unit 25 Image composition unit 70b, 501b, 501b 'Face region 70c, 501c, 501c' Background color calculation region 100 Personal computer 110 Main unit 111 FD loading port 112 CD-ROM loading port 120 Image display device 121 Display screen 130 Keyboard 140 Mouse 1200 Bus 210, 210A, 210B Image group 211 CPU
212 Main memory 213 Hard disk device 214 FD drive 2140 FD
215 CD-ROM drive 2150 CD-ROM
216 I / O interface 217 Output interface 301 Face portion 302 Background portion 303 Clothing portion 400 Color correction window 401 Image group designation column 402 Image display button 403 All page radio button 404 Page designation radio button 405 Page column 406 Next page button 407 Previous page Button 408 Correction button 409 Re-correction button 410 Manual mode buttons 501, 502, 503, 504, 505, 506 Image 501a, 502a, 503a, 504a, 505a, 506a Area information radio button 2170 Color printer

Claims (5)

A face region detector for detecting a face area of a person for each of the plurality of images that is reflected a person in each
For each of the plurality of images, the face-color correction unit that performs the like facial area detected by the face area detection unit becomes a predetermined desired color, the color correction according to each image,
To the entire area of the detected face area or the image by the face region detector, a background region present in a given positional relationship determined for each of the plurality of images, the color of the background region is uniform between the plurality of image A background color unification unit that applies color correction to each image,
A mask that indicates how much each point in the image belongs to the background or the face is determined based on the closeness of the color between each point and the background region and the closeness between the position of each point and the face region. Based on each of the plurality of images, a mask creation unit,
Using the mask created by the mask creation unit, a first correction image obtained by correction by the face color correction unit and a second correction image obtained by correction by the background color unification unit, A color characterized in that a point strongly belonging to the face has a first correction image component, and a point strongly belonging to the background has a composition unit for synthesizing the second correction image component to be strong. Correction device.   The background color unification unit obtains a unified background color to be unified among the plurality of images based on the background region on the first corrected image obtained by the correction by the face color correction unit, and 2. The color correction apparatus according to claim 1, wherein color correction is performed on each of the images so that the color of the background region becomes the unified background color.   The mask creation unit uses a distance in a predetermined color space between the color of the background region and the color of each point as the color closeness between each point and the background region. The color correction apparatus according to 1.   The mask generation unit uses a distance on the image between the center point of each face region and each point as the proximity of the position between each point and the face region. Color correction device. Built into a computer system, on that computer system,
A face region detector for detecting a face area of a person for each of the plurality of images that is reflected a person in each
For each of the plurality of images, the face-color correction unit that performs the like facial area detected by the face area detection unit becomes a predetermined desired color, the color correction according to each image,
To the entire area of the detected face area or the image by the face region detector, a background region present in a given positional relationship determined for each of the plurality of images, the color of the background region is uniform between the plurality of image A background color unification unit that applies color correction to each image,
A mask that indicates how much each point in the image belongs to the background or the face is determined based on the closeness of the color between each point and the background region and the closeness between the position of each point and the face region. Based on each of the plurality of images, a mask creation unit,
Using the mask created by the mask creation unit, a first correction image obtained by correction by the face color correction unit and a second correction image obtained by correction by the background color unification unit, A color is characterized in that a point that strongly belongs to the face has a strong component of the first corrected image, and a point that strongly belongs to the background has a composition unit that combines so that the component of the second corrected image becomes strong Correction program.

Images