JP4507078B2 - Automatic face image correction method and apparatus - Google Patents

Description

  The present invention relates to an automatic face image correction technique.

When a person is photographed on a photo print, in order to improve the image quality of the photo print, the human face is detected, the density is adjusted so that the detected face density is appropriate, sharpness correction, etc. There are things to do.
However, when the density of a photographic print is uniformly adjusted or sharpness correction is performed, for example, pixels that are converted into a white image are generated, and the white pixel region may be enlarged. If the white pixel area is enlarged, the photographic print becomes unnatural and the image quality may be degraded.
In order to reduce the number of pixels that are converted into white pixels, for example, in the sharpness correction, the technology for preventing the enlargement of the white pixel region is to separate the pixel values of pixels belonging to a predetermined white certification condition from the white pixel values. There is an image processing apparatus that converts a value that does not exceed a set intermediate value (see, for example, Patent Document 1).

JP 2002-158882 A

  However, the apparatus of Patent Document 1 can prevent the white area from being enlarged, but cannot reduce the white area originally present in the photographic print. At the time of shooting, for example, when a flash is used, so-called whiteout may occur in which a portion that should originally have a gradation is expressed in white. In the apparatus of Patent Document 1, white pixels cannot be corrected because white pixels become white pixels even if density adjustment or sharpness correction is performed. For this reason, it has not been possible to achieve a sufficient improvement in image quality even if image correction is performed on a photographic print in which whiteout occurs. In particular, in image data in which a person's face is photographed, if white jump occurs in a person's face, the finish of the photograph deteriorates, so an image processing technique for correcting white jump is desired.

  The present invention has been made in view of the above problems, and an object of the present invention is to provide an automatic face image correction technique capable of correcting whiteout of a person's face and correcting image data to a natural state. In the point.

In order to achieve this object, the feature configuration of the automatic face image correction method according to the present invention includes a face detection step for detecting a face area from image data by a face detection unit, and a white pixel recognition unit for the face detection step. The step of certifying the detected white pixels in the face area based on a predetermined white qualification condition, and when the admitted white pixels are continuously distributed over a predetermined range by a whiteout area detection unit A whitening area detection step for detecting a whitening area, and when the whitening area is detected in the whitening area detection step, the whitening area and a certain area around the whitening area are detected by a correction area setting unit. A correction area setting step for setting a correction area including a first luminance value of a boundary pixel between the correction area and the face area, and a second luminance value of a center pixel of the whiteout area by a face correction unit; And the luminance value of the pixel in the correction region gradually changes from the first luminance value to the second luminance value from the boundary between the correction region and the face region toward the center of the whiteout region. And a face correction step for correcting the difference.
That is, according to the above feature configuration, a so-called whiteout region in which a portion that should originally have a gradation is expressed in white is detected from the face region of the image data, and the white pixel is determined as the pixel value of the surrounding face region. Therefore, the white area can be corrected while making the human face natural, and the image quality of the photo print can be improved. In particular, since correction is performed so that the gradation of the pixels in the correction area gradually changes, an image in which the gradation difference between the whiteout area and the surrounding area is large and unnatural is made natural. Image quality can be improved.

Another feature of the automatic face image correction method according to the present invention for achieving this object is that the whiteout region detecting step detects a pupil from the face region, and the pupil is excluded from a whiteout region detection target. There is in point to do.
In the white eye part of the pupil, white pixels are continuously distributed over a certain range, but it is not necessary to correct the white eye part. For this reason, by removing the pupil from the whiteout detection target region, it is possible to prevent the white-eye portion of the pupil from being determined as the whiteout region and to prevent unnecessary correction.

The present invention also covers a program for causing a computer to execute the above-described automatic face image correction method and a medium on which the program is recorded, and such an automatic face image correction program uses a face from image data. A face detection function for detecting an area, and white pixels in the face area detected by the face detection function are certified based on a predetermined white certification condition, and the certified white pixels are continuously distributed over a predetermined range. A white-out area detection function for detecting as a white-out area when the white-out area is detected, and the white-out area and a certain area around the white-out area when the white-out area is detected by the white-out area detection function. A correction area is set, a first luminance value of a boundary pixel between the correction area and the face area, and a second luminance value of a center pixel of the whiteout area are detected, and a luminance value of a pixel in the correction area The Toward the boundary of the serial correction region and the face region in the center of the white跳領zone, and a face correction function for correcting such changed gradually from said first luminance value to said second luminance value. Naturally, such an automatic face image correction program can also obtain all the effects of the automatic face image correction method described above. It is also preferable to add the function according to the dependent claim described in the automatic face image correction method to the automatic face image correction program.

In the present invention, an automatic face image correction apparatus that performs the above-described automatic face image correction method is also subject to rights. Such an automatic face image correction apparatus includes a face detection unit that detects a face area from image data, and , White pixels in the face area detected by the face detection unit are certified based on a predetermined white certification condition, and the whiteout area when the certified white pixels are continuously distributed over a predetermined range And a correction area setting for setting a correction area including the whitening area and a certain area around the whitening area when the whitening area is detected by the whitening area detection unit. A first luminance value of a boundary pixel between the correction area and the face area of the correction area, and a second luminance value of a center pixel of the whiteout area, and the luminance value of the pixel in the correction area is Before the boundary between the correction area and the face area Toward the center of the white跳領zone, and a face correcting unit for correcting such changed gradually from said first luminance value to said second luminance value. Naturally, such an automatic face image correction apparatus can also obtain all the effects of the automatic face image correction method described above. It is also preferable to add the configuration according to the dependent claim described in the automatic face image correction method to the automatic face image correction apparatus.
Other features and advantages of the present invention will become apparent from the following description of embodiments using the drawings.

A first embodiment of the present invention will be described with reference to the drawings.
FIG. 1 shows functional units for performing various operations of a controller of a photographic printing apparatus employing an automatic face image correction technique according to the present invention. Each function is constructed by hardware and / or software. ing.

  The photo printing apparatus includes an image input unit 31 that takes in image data read by the scanner 20 and the media reader 21 and performs preprocessing necessary for the next processing, creation of a graphic operation screen, a keyboard 24, a mouse 25, and the like. A GUI unit 33 for constructing a graphic user interface (hereinafter abbreviated as GUI) for generating a control command from an operation input by the control unit, a control command sent from the GUI unit 33, an operation command input directly from the keyboard 24, etc. A print management unit 32 for performing image processing on the image data transferred from the image input unit 31 to the memory 30 in order to generate desired print data, and a print source image and an expected finish at the time of pre-judge printing work such as color correction. Sent from the print image and the GUI 33 A video control unit 35 that generates a video signal for causing the monitor 23 to display the graphic data, a print data generation unit 36 that generates print data suitable for the print station 1 based on the image data that has undergone image processing, A formatter unit 37 that formats raw image data, image data that has been subjected to image processing, and the like into a format for writing to a CD-R according to a customer's request is provided.

  The print management unit 32 performs an image retouching process such as color correction and filtering (such as blurring and sharpness) on the image data developed in the print order processing unit 60 and the memory 30 for managing the print size and the number of prints. A processing unit 70 is provided.

  The image processing unit 70 includes face image correction means 90 that employs the automatic face image correction technique according to the present invention. As shown in FIG. 2, the face image correction means 90 includes an image data input unit 91, a face detection unit 92, a pupil detection unit 93, a white pixel recognition unit 94, a whiteout region detection unit 95, a correction region setting unit 96, A face correction unit 97 and a corrected image data output unit 98 are provided. The face detection unit 92 is a processing unit that performs a face detection step of detecting a face region from image data. The white pixel recognition unit 94 and the whiteout region detection unit 95 are white in the face region detected by the face detection unit 92. It is a processing unit that performs a whiteout region detection step of detecting pixels as whiteout regions when the pixels are recognized based on a predetermined white color recognition condition and the recognized white pixels are continuously distributed over a predetermined range. The face correcting unit 97 is a processing unit that performs a face correcting step of correcting the whiteout area based on the peripheral pixels of the whiteout area when the whiteout area is detected in the whiteout area detecting step. The pupil detection unit 93 is a processing unit that detects the pupil from the image data of the face region. The correction region setting unit 96 detects the whitening region when the whitening region detection unit 95 detects the whitening region. It is a processing unit that performs a correction area setting step for setting a correction area including a certain area around the area.

Next, details of processing in each processing unit (each step, each function) described above will be described. FIG. 3 is a flowchart showing a processing routine for whiteout correction according to the present invention.
First, the image data input unit 91 acquires image data from the memory 30 (# 100).
The face detection unit 92 receives image data from the image data input unit 91, detects a human face in the image data, and sets a face area (# 200). If no human face is detected, the image data is output to the corrected image data output unit 98 as it is (NO branch at # 250). When a human face is detected, the image data and the data of the face area are output to the pupil detection unit 93 (YES branch at # 250). The pupil detection unit 93 detects a pupil in the face area, and when a pupil is detected, sets an area obtained by removing the pupil from the face area as a whiteout detection target area (# 300).

  The white pixel recognition unit 94 acquires image data and face area data from the face detection unit 92, and detects white pixels in the face area based on a predetermined white recognition condition (# 400). The white recognition condition here is a threshold value set for each of the R, G, and B luminance values, and the R, G, and B luminance values of the target pixel are all equal to or greater than the set threshold value. In this case, the pixel is recognized as a white pixel.

  Specifically, as shown in FIG. 4, the white pixel recognition unit 94 determines a target pixel for white recognition from the face area pixels (# 401), and acquires the R, G, and B luminance values of the target pixel. (# 402). Here, in this embodiment, 250 is set as the threshold value. Therefore, it is determined whether or not the R, G, and B luminance values of the target pixel are all 250 or more (# 403), and if the R, G, and B luminance values are all 250 or more, the pixel is recognized as a white pixel. Then, the pixel position is stored (# 404). The white pixel recognition unit 94 performs white recognition in steps # 401 to # 404 on all the pixels in the face area (# 405), and when the white recognition is completed for all the pixels in the face area, A white pixel map is created based on the pixel position (# 406). The white pixel map may be created at the step # 404 every time a white pixel is detected.

  The whiteout region detection unit 95 detects a whiteout region using the white pixel map created by the white pixel recognition unit 94 (# 500). Specifically, first, the white pixels are determined to be continuous or discontinuous, and the whiteout region to which each white pixel belongs is specified. As shown in FIG. 5, the whiteout region detection unit 95 obtains a white pixel for specifying a region from the white pixel map, and sets this as a target pixel (# 501). Then, a pixel that is recognized as a white pixel and that does not belong to any region is extracted from pixels in the vicinity of the periphery 8 of the target pixel (# 502). If a white pixel is extracted, it is stored as a pixel belonging to the same area as the target pixel (# 503). Similarly, the extracted white pixel is set as a target pixel, and white pixels that are recognized as white pixels and do not belong to any region are extracted from pixels in the vicinity of the surrounding 8 (in # 502). . This is repeated until there are no white pixels that do not belong to any region in the surrounding area, and a continuously distributed white region is specified. The range of pixels determined to be in the same region need not be limited to pixels in the vicinity of 8, but is set according to the size and resolution of image data, such as 4 neighborhoods. When pixels belonging to the same area are no longer detected (NO branch at # 502), white pixels that have not been identified are acquired from the white pixel map, and the process proceeds to step # 501 (YES branch at # 504). Steps # 501 to # 503 are repeated until there are no white pixels for which no area has been specified.

  Furthermore, the whiteout region detection unit 95 determines whether each whiteout region detected in step # 500 is a correction target based on a predetermined correction target condition (# 550). Specifically, the area of the white area and the area of the face area are calculated, and when the ratio of the area of the white area to the area of the face area is within a certain range, it is determined to be a correction target. The area of the overexposure area is smaller than the visible area, the area where the image quality cannot be improved even if the overexposure area is corrected to a very large size, etc. This is because the area is excluded from the correction target.

  Subsequently, the correction area setting unit 96 and the face correction unit 97 perform correction on the overexposure area determined as the correction target in Step # 550 (# 600). Here, the pixels in the correction area are corrected so as to equalize the pixels between the boundary between the correction area and the face area and the center of the whiteout area.

  Specifically, as shown in FIGS. 6 and 7, first, the correction area setting unit 96 sets a correction area for correcting the whiteout area (# 601). As shown in FIG. 7, the correction area here is an area surrounded by the boundary X, in which the whiteout area surrounded by the boundary W is enlarged at a predetermined ratio with the center pixel B of the whiteout area as the center.

  Next, the face correction unit 97 acquires the brightness value of the pixel on the boundary X between the correction area and the face area, that is, the pixel A in FIG. 7, and uses this brightness value as the first brightness value (# 602). . Further, the luminance value of the center pixel of the whiteout region, that is, the pixel B in FIG. 7 is acquired and set as the second luminance value (# 603). The luminance value of the pixels between the pixel A and the pixel B gradually changes at a constant rate from the first luminance value (the luminance value of the pixel A) to the second luminance value (the luminance value of the pixel B). Correct as follows. This is performed all around the pixel B. Specifically, the pixels on the boundary X are sequentially set as the pixel A, the luminance value of the pixel A is set as the first luminance value, and the pixels between the pixel A and the pixel B are corrected, so that all the pixels in the correction area are corrected. Are corrected (# 604).

  The corrected image data output unit 98 outputs the corrected image data to the memory 30 (# 700). As a result, it is possible to eliminate whiteout in an image printed at the print station 1 or an image written on a CD-R or the like.

Next, a second embodiment of the present invention will be described with reference to FIGS.
In the first embodiment, the white area is corrected by the correction area setting unit 96 and the face correction unit 97. The pixels in the correction area are pixels between the boundary between the correction area and the face area and the center of the white area. In this embodiment, correction is performed so that the pixels around the whiteout area in the correction area are expanded into the whiteout area as shown in FIG.

  Specifically, in step # 602, the face correction unit 97 acquires the luminance value of the pixel on the boundary X between the correction region and the face region, the pixel A in FIG. 8, and uses this luminance value as the first luminance value. . In step # 603, the face correcting unit 97 sets the pixel located at the intersection of the line connecting the pixel A and the center pixel B of the whiteout area and the boundary W of the whiteout area as the pixel C, and obtains this luminance value. Thus, the second luminance value is set. In step # 604, the luminance value of the pixels from pixel A to pixel B is constant from the first luminance value (the luminance value of pixel A) to the second luminance value (the luminance value of pixel C). Correction is made so that the rate gradually changes. As in the first embodiment, all pixels in the correction area are corrected by performing the entire circumference around the pixel B.

  In the first embodiment, since the brightness value of the substantially white pixel B is set as the second brightness value, the center of the correction area is substantially white. However, in the present embodiment, the substantially skin-colored pixel is used as the second brightness value. Since the luminance value of C is set, the center of the correction area is substantially skin color. According to image data, it is also a preferred embodiment that the correction in the first embodiment and the correction in the second embodiment can be selectively performed.

Configuration diagram of a photo printing apparatus adopting automatic face image correction technology according to the present invention Configuration diagram of face image correcting means according to the present invention The flowchart which shows the processing routine of white-out correction by this invention The flowchart which shows the recognition subroutine of the white pixel by this invention Flowchart showing a whiteout region detection subroutine according to the present invention. Flowchart showing a whiteout correction subroutine according to the present invention. Schematic diagram showing correspondence between whiteout area and correction area Schematic diagram showing correspondence between whiteout area and correction area

Explanation of symbols

92 face detection unit 93 pupil detection unit 95 white area detection unit 96 correction region setting unit 97 face correction unit

Claims (4)

A face detection step of detecting a face region from the image data by the face detection unit ;
A white pixel recognition step for recognizing a white pixel in the face area detected in the face detection step based on a predetermined white recognition condition by a white pixel recognition unit ;
A whiteout region detection step of detecting a whiteout region when the recognized white pixels are continuously distributed over a predetermined range by a whiteout region detection unit ,
A correction area setting step for setting a correction area including the whitening area and a certain area around the whitening area when the whitening area is detected in the whitening area detection step ;
A face correction unit detects a first luminance value of a boundary pixel between the correction area and the face area and a second luminance value of a center pixel of the whiteout area, and determines a luminance value of the pixels in the correction area. And a face correction step for correcting so as to gradually change from the first luminance value to the second luminance value from the boundary between the correction area and the face area toward the center of the whiteout area. Face image correction method.   The automatic face image correction method according to claim 1, wherein the whitening region detection step detects a pupil from the face region and excludes the pupil from being a whitening region detection target. A face detection function for detecting a face area from image data;
A white pixel in the face area detected by the face detection function is recognized based on a predetermined white recognition condition, and when the recognized white pixel is continuously distributed over a predetermined range, it is determined as a whiteout area. A white area detection function to detect,
When the white spot area is detected by the white spot area detection function, a correction area including the white spot area and a certain area around the white spot area is set, and a boundary pixel between the correction area and the face area is set. A first luminance value and a second luminance value of a central pixel of the whiteout area are detected, and a luminance value of a pixel in the correction area is determined from the boundary between the correction area and the face area. An automatic face image correction program for causing a computer to execute a face correction function for correcting so as to gradually change from the first luminance value to the second luminance value toward the center . A face detection unit for detecting a face area from image data;
A white pixel in the face area detected by the face detection unit is certified based on a predetermined white certification condition, and when the certified white pixel is continuously distributed over a predetermined range, the white pixel is defined as a whiteout area. A white area detection unit to detect,
A correction area setting unit that sets a correction area including the whitening area and a certain area around the whitening area when the whitening area is detected by the whitening area detection unit; and the face area of the correction area And the second luminance value of the center pixel of the whiteout region is detected, and the luminance value of the pixel in the correction region is determined from the boundary between the correction region and the face region. An automatic face image correction apparatus including a face correction unit that performs correction so as to gradually change from the first luminance value to the second luminance value toward the center of the whiteout region .

Images