JP2009105632A - Device and method for processing flesh color correction image - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a flesh color image correction device of a human being, which achieves correction not only for hue but also for saturation in correction processing in a region of a human flesh color and generates no existence of saturation discontinuousness and no color shade after correction in addition to acquiring rich expression power by bringing about a variation in a saturation correction pattern, and to provide its method. <P>SOLUTION: A flesh color correction image processor includes a polar coordinate system transformation means 10 for transforming color difference signals Cr and Cb of an orthogonal coordinate system into a polar coordinate system; a weighting function arithmetic means 12 for correcting a hue direction of hue by the polar coordinate system; and a weighting function arithmetic means 13 for correcting a saturation direction of the hue of the polar coordinate system, performs correction for converging the region of the flesh color on hue of the flesh color, reduces hue discontinuousness near a boundary in a hue direction in the region corrected by the weighting function arithmetic means for correcting a hue direction and reduces hue discontinuousness near a boundary in a saturation direction in the corrected region, further performs correction 16 and 17 of saturation in the region of the flesh color. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a skin color correction image processing apparatus and method, and more particularly to a technology related to image processing for correcting human skin color, which is mounted on video equipment such as a television, a camera, and a DVD.

ITU-R BT. In the 601 standard, the relationship between luminance Y and color difference Cr / Cb and red R, green G, and blue B is defined by the following formula (1). Are distributed in a region between red R and yellow Ye.

  When skin color correction described later is performed on the skin color region, for example, the human face and skin displayed on the TV image are effectively and clearly displayed. Moreover, as shown in the prior art to be described later, when the skin color area is further narrowed, the skin color of Asian people such as yellow races is displayed in reddish, and the effect of being displayed in a youthful manner. Conventionally, skin color correction image processing is installed in video equipment and the like.

図１６は、色差Ｃｂを横軸として、色差Ｃｒを縦軸とした直交座標系において、赤（Ｒ）、黄色（Ｙｅ）、緑色（Ｇ）、シアン（Ｃ）、青色（Ｂ）、マゼンタ（Ｍ）の色相を表す図である。
肌色補正を行うために、図１６に示すＣｂ−Ｃｒ座標系を下記の数式（３）に示す極座標系に変える。ここで、θは色の特性を表す色相で、ｒはその色相に対する彩度である。
カメラの場合はＣＣＤの色再現性、テレビの場合はパネルの色再現性に差が存在し、商品に設定する人肌の色データは、メーカによって異なるので、一般性を失わせないため、人肌の色相をθ１とする。

図１８は、デジタル画像処理において、従来の肌色補正の領域を示す図である。
肌色補正は、下記に示す数式（４）で行われる。補正エリアは、ｒ１＜ｒ＜ｒ２、θ１−Δθ１＜θ＜θ１＋Δθ１で定められた範囲の彩度ｒ、色相θである。

ここで、ｋは係数で、０から１までの範囲で設定できる。係数ｋは小さいほど、補正効果は強くなる。
図１７は、肌色補正前のエリアと補正後のエリア変化を示す図である。補正前の肌色は「ａｅｆｂｃｇｈｄａ」で囲まれたエリアに分布していたが、補正後の肌色は「ｅｆｇｈｅ」で囲まれたエリアに圧縮されている。補正後の肌色は、補正前より理想の肌色の色相θ１に集中したことが分かる。
しかしながら、「ａｅｈｄａ」で囲まれたエリアと「ｆｂｃｇｆ」で囲まれたエリアにある色は補正画像から無くなる。また補正後の画像にエリアの境目「ａｅｆｂ」と「ｄｈｇｃ」において、その境目の上と下に色の不連続が存在する。これらの原因で補正後の画像に色むらが生じるという問題点がある。 A conventional human skin color area correction process will be described below.
In the digital image, RGB is represented by the same number of bits. At this time, the range of values that the color differences Cr and Cb can take are different. Therefore, in order to keep Cr and Cb in the same range and to facilitate polar coordinate conversion to be described later, normalization processing is performed on the formula in CrCb of the above formula (1) to obtain the following formula (2).

FIG. 16 shows a red (R), yellow (Ye), green (G), cyan (C), blue (B), magenta (in the orthogonal coordinate system with the color difference Cb as the horizontal axis and the color difference Cr as the vertical axis. It is a figure showing the hue of M).
In order to perform skin color correction, the Cb-Cr coordinate system shown in FIG. 16 is changed to a polar coordinate system shown in the following mathematical formula (3). Here, θ is a hue representing a color characteristic, and r is a saturation with respect to the hue.
There is a difference in CCD color reproducibility in the case of cameras and panel color reproducibility in the case of TV, and the human skin color data set for the product differs depending on the manufacturer. Let the skin hue be θ1.

FIG. 18 is a diagram illustrating a conventional skin color correction region in digital image processing.
The skin color correction is performed by the following formula (4). The correction area is saturation r and hue θ in a range defined by r1 <r <r2, θ1−Δθ1 <θ <θ1 + Δθ1.

Here, k is a coefficient and can be set in the range from 0 to 1. The smaller the coefficient k, the stronger the correction effect.
FIG. 17 is a diagram illustrating an area before skin color correction and an area change after correction. The skin color before correction is distributed in the area surrounded by “aefbcghda”, but the skin color after correction is compressed into the area surrounded by “efghe”. It can be seen that the skin color after the correction is concentrated on the hue θ1 of the ideal skin color than before the correction.
However, the colors in the area surrounded by “aehda” and the area surrounded by “fbcgf” disappear from the corrected image. In the corrected image, color discontinuities exist above and below the boundary between the areas “aefb” and “dhgc”. For these reasons, there is a problem in that color unevenness occurs in the corrected image.

The invention of Patent Document 1 is an invention made in view of this problem, and in a correction process in a human skin color area, there is no discontinuity after correction and color image correction apparatus for human skin that does not cause uneven color and its It aims to provide a method. The method of patent document 1 is described below.
FIG. 24 is a block configuration diagram showing an embodiment of a human skin color image correction apparatus according to the invention of Patent Document 1. In FIG. In the figure, a human skin color image correction apparatus receives a chroma signal Cr / Cb and converts it from a Cr / Cb orthogonal coordinate system shown in FIG. 16 to a polar coordinate system so that color correction described later can be easily performed. In order to determine the skin color correction area, the conversion means 110, a storage means 111 for storing the hue θ1 at the center of the ideal skin color, and a weight W (θ) calculation means 112 for the hue angle direction θ as shown in FIG. For the hue radial direction r weighting W (r) computing means 113, the hue angle direction θ weighting W (θ) computing means 112 and the hue radial direction r weighting W (r) computing means 113 Storage means 114 for storing a constant coefficient value k to be input.

ここで、色相の角度方向θにおける重み付け関数Ｗ（θ）は、例えば図１９に太線で示すような演算が簡易であるリニア関数の他、補正エリアが図２１に太線で示すような非対称エリアの場合は、例えば図２２に太線で示すような関数、さらにはこれらの線形関数以外にも、図２３に太線で示すような非線形関数を使用しても一般性が失われない。
また、彩度方向ｒにおける重み付け関数Ｗ（ｒ）は、例えば図２０に太線で示すような関数を使用する。
尚、解析式で表せない複雑な非線形関数や、ハードウェアで計算できない関数が必要な場合には、ＬＵＴ（Ｌｏｏｋ Ｕｐ Ｔａｂｌｅ）を使って重み付け関数を実現すれば、この問題を解決することができる。
このように、色相の角度方向θと彩度方向ｒにおける２つの重み関数Ｗ（ｒ）、Ｗ（θ）を設け、これらの関数を組み込んだ上記数式（５）を実行し、補正した画像は、θとｒの両方向における色の消滅や色の不連続によって生じた色むらを改善することができる。
特開２００６−２３７７９８号公報 Next, an operation for eliminating the color unevenness of the human skin color image correction apparatus having the above-described configuration will be described. By reflecting the weighting function W (θ) in the angular direction θ of the hue and the weighting function W (r) in the saturation direction r in the above mathematical expression (4), the following mathematical expression (5) is obtained. The means for calculating the function value of the function W (θ) is the weighting W (θ) calculating means 112, and the means for calculating the function value of the function W (r) is the weighting W (r) calculating means 113. .

Here, the weighting function W (θ) in the hue angle direction θ is, for example, a linear function that is easy to calculate as shown by a thick line in FIG. 19, and a correction area of an asymmetric area as shown by a thick line in FIG. In this case, for example, the generality is not lost even if a function as shown by a thick line in FIG. 22 or a nonlinear function as shown by a thick line in FIG. 23 is used besides these linear functions.
Further, as the weighting function W (r) in the saturation direction r, for example, a function shown by a thick line in FIG. 20 is used.
If a complicated nonlinear function that cannot be expressed by an analytical expression or a function that cannot be calculated by hardware is required, this problem can be solved by implementing a weighting function using a LUT (Look Up Table). .
Thus, two weighting functions W (r) and W (θ) in the hue angle direction θ and the saturation direction r are provided, and the corrected image is obtained by executing the above formula (5) incorporating these functions. It is possible to improve color unevenness caused by disappearance of colors and discontinuity of colors in both directions of θ, r.
JP 2006-237798 A

  However, although the hue correction is performed in the above-described prior art, the effect of the skin color correction may be insufficient when only the hue is corrected, and further improvement in expressive power is required.

  The present invention has been made in view of the above problems, and in the correction process in the human skin color region, not only the hue but also the saturation is corrected, and the saturation correction pattern is provided with variations to be rich. Another object of the present invention is to provide a skin color correction image processing apparatus and method that obtains a high level of expressive power and that does not have color saturation after correction and does not cause color unevenness.

  In order to solve the above-mentioned problems, the present invention provides a polar coordinate system converting means for converting the color difference signals Cr and Cb of the orthogonal coordinate system into a polar coordinate system, and a weighting function calculating means for correcting the hue direction of the hue by the polar coordinate system; A weighting function calculation unit that corrects the saturation direction of the hue by the polar coordinate system, and performs a correction that concentrates the skin color region on the hue of the ideal skin color and also performs the correction of the hue direction. This is a skin color correction image processing apparatus that reduces the discontinuity of the hue near the boundary in the hue direction in the corrected area and reduces the discontinuity of the hue near the boundary in the saturation direction in the corrected area. The skin color correction image processing apparatus further corrects the saturation of the skin color region.

  The present invention is also the above skin color correction image processing apparatus that narrows the saturation region by correcting the saturation of the skin color region.

  And this invention is said skin color correction | amendment image processing apparatus which expands the area | region of a saturation by correction | amendment of the saturation of the said skin color area | region.

  Furthermore, the present invention is the above skin color correction image processing apparatus that increases the intensity of saturation by correcting the saturation of the skin color region.

  Further, the present invention is the above skin color correction image processing apparatus that weakens the intensity of saturation by correcting the saturation of the skin color region.

  The present invention further comprises weighting function calculation means for correcting the saturation direction of saturation by the polar coordinate system, and a boundary in the saturation direction in the region corrected by the weighting function calculation means for correcting the saturation direction. The skin color correction image processing apparatus described above that reduces discontinuity of saturation in the vicinity.

  Furthermore, the present invention includes weighting function calculation means for correcting the hue direction of saturation by the polar coordinate system, and a saturation near a boundary in the hue direction in the area corrected by the weighting function calculation means for correcting the hue direction. The skin color correction image processing apparatus described above that reduces discontinuity in degree.

  Further, in the present invention, in the region to be corrected, a correction region in which the saturation direction region is asymmetric is set as a correction target region, so that more flexible correction is performed on the characteristics of the device. It is said skin color correction image processing apparatus.

  The present invention is the above skin color correction image processing apparatus in which the weighting function calculation means for correcting the saturation direction of the saturation uses a linear weighting function symmetric with respect to the intensity of the saturation.

  Furthermore, the present invention is the above skin color correction image processing apparatus, wherein the weighting function calculation means for correcting the saturation direction of the saturation uses a linear weighting function that is asymmetric in the intensity of the saturation.

  Further, the present invention is the above skin color correction image processing apparatus, wherein the weighting function calculation means for correcting the saturation direction of the saturation uses a nonlinear weighting function in the intensity of the saturation.

  The present invention is the above skin color correction image processing apparatus using a LUT in which the weighting function calculation means for correcting the saturation direction of the saturation uses an LUT capable of realizing an arbitrary weighting function in the intensity of the saturation. .

  Furthermore, the present invention is the above-described skin color correction image processing apparatus, wherein the weighting function calculating means for correcting the hue direction of saturation by the polar coordinate system uses a trapezoidal function as the weighting function in the hue.

  Further, the present invention is the above skin color correction image processing apparatus using an LUT in which the weighting function calculating means for correcting the hue direction of the saturation by the polar coordinate system can realize an arbitrary weighting function in the hue.

  The present invention also includes a polar coordinate system conversion step for converting the color difference signals Cr and Cb of the orthogonal coordinate system into a polar coordinate system, a weighting function calculation step for correcting the hue direction of the hue by the polar coordinate system, and a hue by the polar coordinate system. A weighting function calculating step for correcting the saturation direction of the image, and performing correction for concentrating the skin color area on the hue of the ideal skin color, and performing the correction by the weighting function calculating step for correcting the hue direction. A skin color correction image processing method including reducing a hue discontinuity near a boundary in a hue direction in a corrected area and reducing a discontinuity in a hue near a boundary in a saturation direction in the corrected area. The skin color correction image processing method further includes correcting the saturation of the skin color region.

According to the skin color correction image processing apparatus and method of the present invention, by correcting the saturation in addition to the correction of the hue, the degree of freedom can be greatly expanded with respect to the conventional correction expression of only the hue.
Further, according to the skin color correction image processing apparatus and method according to the present invention, with regard to saturation correction, an expression that brings a person's skin color closer to ideal by narrowing the saturation to a region near the ideal saturation, saturation An expression that gives the skin color saturation of the person by broadening the area centered on the ideal saturation, an expression that makes the person's skin color brighter by increasing the saturation, and a decrease in saturation It is possible to express the person's skin color in general.
Further, according to the skin color correction image processing apparatus and method according to the present invention, by providing a weighting function for r and θ, natural correction can be performed without loss of color or discontinuity due to correction.
Further, according to the skin color correction image processing apparatus and method according to the present invention, by using an asymmetric correction area in the saturation of the ideal skin color, it is possible to perform correction in consideration of a difference in device characteristics and the like. The same function can be provided in other devices.
Further, according to the skin color correction image processing apparatus and method according to the present invention, these correction methods are not only realized by software, but also by a LUT in the weighting function in order to provide flexibility in hardware implementation. By providing, correction processing can be processed in a short time without using an arithmetic circuit for a complicated weighting function.

Hereinafter, an embodiment of a skin color correction image processing apparatus according to the present invention will be described in detail.
In the description of this embodiment, the skin color correction image processing apparatus according to the present invention will be described in detail. However, the skin color correction image processing method according to the present invention is employed by the skin color correction image processing apparatus according to the present invention. Since it is a method, the skin color correction image processing method according to the present invention is included in the following description.

FIG. 1 is a block diagram showing an embodiment of a skin color correction image processing apparatus according to the present invention. In the figure, a skin color correction image processing apparatus receives a chroma signal Cr / Cb and converts it from a Cr / Cb orthogonal coordinate system shown in FIG. 16 to a polar coordinate system 10 so as to facilitate color correction described later. In order to determine the skin color correction area, a storage means 11 that stores the hue θ1 at the center of the ideal skin color, as shown by a thick line in FIG. 18, and a weighting Wθ (θ) calculation means 12 in the angular direction of the hue θ; The hue θ radial direction weighting Wθ (r) computing means 13, the hue angular direction weighting Wθ (θ) computing means 12, and the hue radial direction weighting Wθ (r) computing means 13. Storage means 14 for storing the constant coefficient value kθ, storage means 15 for storing the saturation r3 at the center of the ideal skin color, as shown by a thick line in FIG. 6, and weighting Wr (θ) in the angular direction of saturation r Calculation means 16 and saturation r Constant weighting Wr (r) calculating means 17, saturation angular direction weighting Wr (θ) calculating means 16 and saturation radial weighting Wr (r) calculating means 17. Storage means 18 for storing the coefficient value kr.
Since the correction of the hue θ is the same as that of the conventional example, the description is omitted here, and only the correction of the saturation r will be described in detail.
There is a difference in CCD color reproducibility in the case of cameras and panel color reproducibility in the case of TV, and the human skin color data set for the product differs depending on the manufacturer. Let the saturation of the skin be r3.

ここで、ｋｒは係数で、０から１までの範囲で設定できる。ｋｒは小さいほど、補正後の領域は狭くなる。
補正前の肌色は「ａｅｆｂｃｇｈｄａ」で囲まれたエリアに分布していたが、色相補正後の肌色は「ｅｆｇｈｅ」で囲まれたエリアに圧縮されている。補正後の肌色は、補正前より理想の肌色の色相θ１に集中したことが分かる。さらに彩度補正後の肌色は、「ｉｊｋｌｉ」で囲まれたエリアに圧縮されている。
しかしながら、「ｌｋｇｈｌ」で囲まれたエリアと「ｅｆｊｉｅ」で囲まれたエリアにある彩度は補正画像から無くなる。また補正後の画像にエリアの境目「ｅｉｌｈ」と「ｆｊｋｇ」において、その境目の両側で彩度の不連続が存在する。これらの原因で補正後の画像に彩度の不連続が生じる。
上述した数式（５）および（６）に、彩度ｒの角度方向θの重み付け関数Ｗｒ（θ）と彩度方向ｒの重み付け関数Ｗｒ（ｒ）を反映させて、下記に示す数式（７）が得られる。この関数Ｗｒ（θ）の関数値を演算する手段が、重み付けＷｒ（θ）演算手段１６であり、関数Ｗｒ（ｒ）の関数値を演算する手段が、重み付けＷｒ（ｒ）演算手段１７である。

ここで、彩度ｒのｒ方向における重み付け関数Ｗｒ（ｒ）は、例えば図７に太線で示すような演算が簡易であるリニア関数の他、補正エリアが図９に示すような非対称エリアの場合は、例えば図１０に太線で示すような関数、さらにはこれらの線形関数以外にも、図１１に太線で示すような非線形関数を使用しても一般性が失われない。
また、彩度ｒのθ方向における重み付け関数Ｗｒ（θ）は、例えば図８に太線で示すような関数を使用する。
尚、解析式で表せない複雑な非線形関数や、ハードウェアで計算できない関数が必要な場合には、ＬＵＴ（Ｌｏｏｋ Ｕｐ Ｔａｂｌｅ）を使って重み付け関数を実現すれば、この問題を解決することができる。
このように、色相θと彩度ｒにおける２つの重み関数を設け、これらの関数を組み込んだ上記数式（７）を実行し、補正した画像は、θとｒの両方向における色の消滅や色の不連続によって生じた色むらを改善することができる。 First, a method for narrowing the saturation of the skin color region to a region near the ideal saturation will be described.
FIG. 2 is a diagram illustrating an area change after the skin color correction of the hue is performed by superimposing the skin color correction for narrowing the region in the region near the ideal saturation r3.
The skin color correction is performed by the following formula (6) obtained by adding saturation r correction to the formula (4). The correction areas are r and θ within a range defined by r1 <r <r2 and θ1−Δθ1 <θ <θ1 + Δθ1. Note that r1 <r <r2 may be replaced by r3-Δr1 <r <r3 + Δr1.

Here, kr is a coefficient and can be set in a range from 0 to 1. The smaller the kr, the narrower the corrected area.
The skin color before correction is distributed in the area surrounded by “aefbcghda”, but the skin color after hue correction is compressed into the area surrounded by “efghhe”. It can be seen that the skin color after the correction is concentrated on the hue θ1 of the ideal skin color than before the correction. Furthermore, the skin color after saturation correction is compressed into an area surrounded by “ijkli”.
However, the saturation in the area surrounded by “lkghl” and the area surrounded by “efjie” disappears from the corrected image. In the corrected image, there is a discontinuity in saturation at both sides of the boundary between the areas “eilh” and “fjkg”. For these reasons, discontinuity of saturation occurs in the corrected image.
By reflecting the weighting function Wr (θ) in the angular direction θ of the saturation r and the weighting function Wr (r) in the saturation direction r in the above mathematical expressions (5) and (6), the following mathematical expression (7) Is obtained. The means for calculating the function value of the function Wr (θ) is the weighting Wr (θ) calculating means 16, and the means for calculating the function value of the function Wr (r) is the weighting Wr (r) calculating means 17. .

Here, the weighting function Wr (r) in the r direction of the saturation r is, for example, when the correction area is an asymmetric area as shown in FIG. For example, the generality is not lost even if a function as shown by a thick line in FIG. 10 or a nonlinear function other than these linear functions as shown by a thick line in FIG. 11 is used.
Further, as the weighting function Wr (θ) in the θ direction of the saturation r, for example, a function shown by a thick line in FIG. 8 is used.
If a complicated nonlinear function that cannot be expressed by an analytical expression or a function that cannot be calculated by hardware is required, this problem can be solved by implementing a weighting function using a LUT (Look Up Table). .
In this way, two weighting functions for hue θ and saturation r are provided, and the above formula (7) incorporating these functions is executed, and the corrected image is the disappearance of the color in both directions of θ and r, Color unevenness caused by discontinuity can be improved.

ここで、前述の数式（８）の彩度ｒのｒ方向における重み付け関数Ｗｒ（ｒ）は、例えば図１２に太線で示すような演算が簡易であるリニア関数の他、補正エリアが図９に太線で示すような非対称エリアの場合は、例えば図１４に太線で示すような関数、さらにはこれらの線形関数以外にも、図１５に太線で示すような非線形関数を使用しても一般性が失われない。
また、彩度ｒのθ方向における重み付け関数Ｗｒ（θ）は、例えば図１３に太線で示すような関数を使用する。
尚、解析式で表せない複雑な非線形関数や、ハードウェアで計算できない関数が必要な場合には、ＬＵＴ（Ｌｏｏｋ Ｕｐ Ｔａｂｌｅ）を使って重み付け関数を実現すれば、この問題を解決することができる。
このように、θとｒ方向における２つの重み関数Ｗｒ（ｒ）、Ｗｒ（θ）を設け、これらの関数を組み込んだ上記数式（８）を実行し、補正した画像は、θとｒの両方向における色の消滅や色の不連続によって生じた色むらを改善することができる。 Next, a method for expanding the saturation of the skin color area around the ideal saturation will be described.
FIG. 3 is a diagram showing an area change after the skin color correction of the hue and the skin color correction for extending the saturation region centered on the ideal saturation r3 are performed repeatedly.
The skin color correction is performed by the above-described equation (6). The correction area is
r1 <r <r2
θ1−Δθ1 <θ <θ1 + Δθ1
R and θ in the range determined by.
Here, the coefficient kr can be set to a value of 1 or more. The larger the coefficient kr, the wider the corrected area.
The skin color before correction is distributed in the area surrounded by “aefbcghda”, but the skin color after hue correction is compressed into the area surrounded by “efghhe”. It can be seen that the skin color after the correction is concentrated on the hue θ1 of the ideal skin color than before the correction. Further, the skin color after saturation correction is extended to an area surrounded by “emnfgophe”.
However, the saturation in the area surrounded by “hgoph” and the area surrounded by “emnfe” doubles with the saturation before correction in the corrected image. In the corrected image, there are discontinuities in saturation on both sides of the boundary between the areas “mehp” and “nfgo”. For these reasons, discontinuity of saturation occurs in the corrected image.
The following formula (8) is obtained by reflecting the case where the coefficient kr is 1 or more in the formula (7).

Here, the weighting function Wr (r) in the r direction of the saturation r in Equation (8) described above is, for example, a linear function that is easy to calculate as indicated by a thick line in FIG. In the case of an asymmetric area as indicated by a thick line, for example, a function as indicated by a thick line in FIG. 14 and, in addition to these linear functions, a non-linear function as indicated by a thick line in FIG. Not lost.
Further, as the weighting function Wr (θ) in the θ direction of the saturation r, for example, a function as shown by a thick line in FIG. 13 is used.
If a complicated nonlinear function that cannot be expressed by an analytical expression or a function that cannot be calculated by hardware is required, this problem can be solved by implementing a weighting function using a LUT (Look Up Table). .
As described above, the two weight functions Wr (r) and Wr (θ) in the θ and r directions are provided, the above formula (8) incorporating these functions is executed, and the corrected image is obtained in both the θ and r directions. Color unevenness caused by color disappearance or color discontinuity in can be improved.

ここで、係数ｋｒは１以上の値で設定できる。係数ｋｒは大きいほど、彩度ｒは強く補正される。
補正前の肌色は「ａｅｆｂｃｇｈｄａ」で囲まれたエリアに分布していたが、色相補正後の肌色は「ｅｆｇｈｅ」で囲まれたエリアに圧縮されている。補正後の肌色は、補正前より理想の肌色の色相θ１に集中したことが分かる。さらに彩度補正後の肌色は、「ｉｊｇｏｐｈｉ」で囲まれたエリアにシフトされている。
しかしながら、「ｅｆｊｉｅ」で囲まれたエリアにある彩度は補正画像から無くなる。また「ｈｇｏｐｈ」で囲まれたエリアにある彩度は補正画像において補正前からあった彩度と二重に存在してしまう。また補正後の画像にエリアの境目「ｅｉｈｐ」と「ｆｊｇｏ」において、その境目の上と下に彩度の不連続が存在する。これらの原因で補正後の画像に彩度の不連続が生じる。
上述した数式（９）に、色相の角度方向θの重み付け関数Ｗθ（θ）と彩度方向ｒの重み付け関数Ｗθ（ｒ）と彩度の角度方向θの重み付け関数Ｗｒ（θ）と彩度方向ｒの重み付け関数Ｗｒ（ｒ）を反映させて、下記に示す数式（１０）が得られる。この関数Ｗθ（θ）の関数値を演算する手段が、重み付けＷθ（θ）演算手段１２であり、関数Ｗθ（ｒ）の関数値を演算する手段が、重み付けＷθ（ｒ）演算手段１３であり、関数Ｗｒ（θ）の関数値を演算する手段が、重み付けＷｒ（θ）演算手段１６であり、関数Ｗｒ（ｒ）の関数値を演算する手段が、重み付けＷｒ（ｒ）演算手段１７である。

ここで、前述の数式（１０）の彩度ｒのｒ方向における重み付け関数Ｗｒ（ｒ）は、例えば図１２に太線で示すような演算が簡易であるリニア関数の他、補正エリアが図９に示すような非対称エリアの場合は、例えば図１４に太線で示すような関数、さらにはこれらの線形関数以外にも、図１５に太線で示すような非線形関数を使用しても一般性が失われない。
また、彩度ｒのθ方向における重み付け関数Ｗｒ（θ）は、例えば図１３に太線で示すような関数を使用する。
尚、解析式で表せない複雑な非線形関数や、ハードウェアで計算できない関数が必要な場合には、ＬＵＴ（Ｌｏｏｋ Ｕｐ Ｔａｂｌｅ）を使って重み付け関数を実現すれば、この問題を解決することができる。
このように、θとｒにおける２つの重み関数を設け、これらの関数を組み込んだ上記数式（１０）を実行し、補正した画像は、θとｒの両方向における色の消滅や色の不連続によって生じた色むらを改善することができる。 Next, a method for strongly correcting the saturation of the skin color area will be described.
FIG. 4 is a diagram illustrating an area change after the skin color correction for increasing the saturation is superimposed on the skin color correction for the hue.
The skin color correction is performed by the following formula (9). The correction area is
r1 <r <r2
θ1−Δθ1 <θ <θ1 + Δθ1
R and θ in the range determined by.

Here, the coefficient kr can be set to a value of 1 or more. The greater the coefficient kr, the stronger the saturation r is corrected.
The skin color before correction is distributed in the area surrounded by “aefbcghda”, but the skin color after hue correction is compressed into the area surrounded by “efghhe”. It can be seen that the skin color after the correction is concentrated on the hue θ1 of the ideal skin color than before the correction. Furthermore, the skin color after saturation correction is shifted to the area surrounded by “ijgophi”.
However, the saturation in the area surrounded by “efjie” is lost from the corrected image. Further, the saturation in the area surrounded by “hgoph” is doubled with the saturation that existed before the correction in the corrected image. In the corrected image, there is a discontinuity of saturation above and below the boundary between the areas “eihp” and “fjgo”. For these reasons, discontinuity of saturation occurs in the corrected image.
The weighting function Wθ (θ) for the hue angle direction θ, the weighting function Wθ (r) for the saturation direction r, the weighting function Wr (θ) for the saturation angle direction θ, and the saturation direction Reflecting the weighting function Wr (r) of r, the following formula (10) is obtained. The means for calculating the function value of the function Wθ (θ) is the weighting Wθ (θ) calculation means 12, and the means for calculating the function value of the function Wθ (r) is the weighting Wθ (r) calculation means 13. The means for calculating the function value of the function Wr (θ) is the weighting Wr (θ) calculating means 16, and the means for calculating the function value of the function Wr (r) is the weighting Wr (r) calculating means 17. .

Here, the weighting function Wr (r) in the r direction of the saturation r in Equation (10) described above is, for example, a linear function that is easy to calculate as shown by a thick line in FIG. In the case of an asymmetric area as shown in FIG. 14, for example, the function shown by a thick line in FIG. 14 and, in addition to these linear functions, the use of a nonlinear function as shown by a thick line in FIG. Absent.
Further, as the weighting function Wr (θ) in the θ direction of the saturation r, for example, a function as shown by a thick line in FIG. 13 is used.
If a complicated nonlinear function that cannot be expressed by an analytical expression or a function that cannot be calculated by hardware is required, this problem can be solved by implementing a weighting function using a LUT (Look Up Table). .
In this way, two weight functions in θ and r are provided, and the above formula (10) incorporating these functions is executed, and the corrected image is caused by color disappearance or color discontinuity in both directions of θ and r. The resulting color unevenness can be improved.

ここで、前述の数式（１１）の彩度ｒのｒ方向における重み付け関数Ｗｒ（ｒ）は、例えば図７に太線で示すような演算が簡易であるリニア関数の他、補正エリアが図９に示すような非対称エリアの場合は、例えば図１０に太線で示すような関数、さらにはこれらの線形関数以外にも、図１１に太線で示すような非線形関数を使用しても一般性が失われない。
また、彩度ｒのθ方向における重み付け関数Ｗｒ（θ）は、例えば図８に太線で示すような関数を使用する。
尚、解析式で表せない複雑な非線形関数や、ハードウェアで計算できない関数が必要な場合には、ＬＵＴ（Ｌｏｏｋ Ｕｐ Ｔａｂｌｅ）を使って重み付け関数を実現すれば、この問題を解決することができる。
このように、θとｒにおける２つの重み関数を設け、これらの関数を組み込んだ上記数式（５）を実行し、補正した画像は、θとｒの両方向における色の消滅や色の不連続によって生じた色むらを改善することができる。 Next, a method for correcting the saturation of the skin color area to be weak will be described.
FIG. 5 is a diagram illustrating an area change after the skin color correction for hue is performed by performing the skin color correction for reducing the saturation.
The skin color correction is performed by the above-described equation (10). The correction area is
r1 <r <r2
θ1−Δθ1 <θ <θ1 + Δθ1
R and θ in the range determined by.
Here, the coefficient kr can be set in the range from 0 to 1. The smaller the coefficient kr, the weaker the saturation r is corrected.
The skin color before correction is distributed in the area surrounded by “aefbcghda”, but the skin color after hue correction is compressed into the area surrounded by “efghhe”. It can be seen that the skin color after the correction is concentrated on the hue θ1 of the ideal skin color than before the correction. Further, the skin color after saturation correction is shifted to an area surrounded by “emnfkle”.
However, the saturation in the area surrounded by “lkghl” is lost from the corrected image. In addition, the saturation in the area surrounded by “emnfe” is doubled with the saturation that existed before the correction in the corrected image. In the corrected image, there is a discontinuity of saturation above and below the boundary between the areas “melh” and “nfkg”. For these reasons, discontinuity occurs in the saturation of the corrected image.
The formula (11) shown below is obtained by reflecting the case where the coefficient kr is from 0 to 1 in the formula (10).

Here, the weighting function Wr (r) in the r direction of the saturation r in the above equation (11) is, for example, a linear function that is easy to calculate as shown by a thick line in FIG. In the case of an asymmetric area as shown, for example, the function shown by a thick line in FIG. 10 and, in addition to these linear functions, generality is lost even if a nonlinear function as shown by a thick line in FIG. 11 is used. Absent.
Further, as the weighting function Wr (θ) in the θ direction of the saturation r, for example, a function as shown by a thick line in FIG. 8 is used.
If a complicated nonlinear function that cannot be expressed by an analytical expression or a function that cannot be calculated by hardware is required, this problem can be solved by implementing a weighting function using a LUT (Look Up Table). .
In this way, two weight functions in θ and r are provided, and the above formula (5) incorporating these functions is executed, and the corrected image is caused by the disappearance of color or discontinuity in both directions of θ and r. The resulting color unevenness can be improved.

  As described above, ITU-R BT. Although the skin color correction method is described using the Cr, Cb coordinate system of 601, generality is not lost even in other coordinate systems (for example, IQ coordinate system, Cr, Cb coordinate system of ITU-R BT.709). .

According to the skin color correction image processing apparatus and method of the present invention, the degree of freedom of correction expression can be expanded by correcting the saturation in addition to the correction of the hue.
Further, according to the skin color correction image processing apparatus and method of the present invention, for saturation correction, the saturation is narrowed to a region near the ideal saturation, and the region is expanded with the saturation centered on the ideal saturation. It is possible to increase the saturation or decrease the saturation.
In addition, according to the skin color correction image processing apparatus and method of the present invention, it is possible to provide a weighting function for r and θ to eliminate color loss and discontinuity due to correction.
In addition, according to the skin color correction image processing apparatus and the method thereof according to the present invention, it is possible to perform correction in consideration of differences in device characteristics by using an asymmetric correction area in the saturation of the ideal skin color.
In addition, according to the skin color correction image processing apparatus and method of the present invention, in order to generate a complicated non-linear weighting function or the like in order to realize the hardware in addition to the realization by software, An uptable (“LUT”) can be used.

1 is a block configuration diagram showing a first embodiment of a skin color correction image processing apparatus according to the present invention. Explanatory drawing of the color area change before and behind the skin color correction | amendment which narrows the area | region of saturation. Explanatory drawing of the color area change before and after the skin color correction which expands the area | region of saturation. Explanatory drawing of the color area change before and after the skin color correction which makes a saturation strong. Explanatory drawing of the color area change before and behind the skin color correction | amendment which makes a saturation weak. Explanatory drawing of the symmetry correction area in r. Explanatory drawing of the linear weighting function (symmetric) in r. Explanatory drawing of the weighting function in (theta). Explanatory drawing of the asymmetric correction area in r. Explanatory drawing of the linear weighting function (asymmetric) in r. Explanatory drawing of the nonlinear weighting function (asymmetric) in r. Explanatory drawing of the linear weighting function (symmetric) in r. Explanatory drawing of the weighting function in (theta). Explanatory drawing of the linear weighting function (asymmetric) in r. Explanatory drawing of the nonlinear weighting function (asymmetric) in r. Explanatory drawing of the positional relationship of each color in a Cr / Cb coordinate system. Explanatory drawing of the color area change before and after skin color correction. Explanatory drawing of the symmetry correction area in (theta). Explanatory drawing of the linear weighting function (symmetric) in (theta). Explanatory drawing of the weighting function in r. Explanatory drawing of the asymmetric correction area in (theta). Explanatory drawing of the linear weighting function (asymmetric) in (theta). Explanatory drawing of the nonlinear weighting coefficient (asymmetric) in (theta). The block block diagram which shows a prior art.

Explanation of symbols

10 Polar coordinate conversion means 11 Storage means 12 for storing the ideal hue θ1 of the flesh color center Hue direction weighting Wθ (θ) calculation means 13 Hue saturation direction weighting Wθ (r) calculation means 14 Hue correction Storage means 15 for storing the constant coefficient value kθ Storage means 16 for storing the saturation r3 at the center of the ideal skin color Weighting Wr (θ) for saturation hue direction Weighting Wr (θ) for saturation direction of saturation r) Calculation means 18 Storage means 110 for storing a constant coefficient value kr for saturation correction Polar coordinate conversion means 111 Storage means 112 for storing the hue θ1 at the center of an ideal skin color Weighting W (θ) calculation means for hue direction of hue 113 Hue saturation direction weighting W (r) calculation means 114 Storage means for storing a constant coefficient value k for hue correction

Claims (15)

Polar coordinate system conversion means for converting the color difference signals Cr and Cb of the orthogonal coordinate system to the polar coordinate system, weighting function calculation means for correcting the hue direction of the hue by the polar coordinate system, and correction of the saturation direction of the hue by the polar coordinate system A weighting function calculating means for performing correction for concentrating the skin color area on the hue of an ideal skin color, and a boundary in a hue direction in the area subjected to the correction by the weighting function calculating means for correcting the hue direction A skin color correction image processing apparatus that reduces discontinuity of a nearby hue and reduces discontinuity of a hue near a boundary in a saturation direction in the corrected region,
Furthermore, the skin color correction image processing apparatus which corrects the saturation of the skin color region.   The skin color correction image processing apparatus according to claim 1, wherein the saturation region is narrowed by correcting the saturation of the skin color region.   The skin color correction image processing apparatus according to claim 1, wherein the saturation region is expanded by correcting the saturation of the skin color region.   The skin color correction image processing apparatus according to claim 1, wherein the intensity of saturation is increased by correcting the saturation of the skin color region.   The skin color correction image processing apparatus according to claim 1, wherein the intensity of saturation is reduced by correcting the saturation of the skin color region.   Weighting function calculation means for correcting the saturation direction of saturation by the polar coordinate system is provided, and saturation in the vicinity of the boundary of the saturation direction in the area corrected by the weighting function calculation means for correcting the saturation direction is not included. The skin color correction image processing apparatus according to claim 1, which reduces continuity.   Weighting function calculating means for correcting the hue direction of saturation by the polar coordinate system is provided, and the discontinuity of saturation near the boundary of the hue direction in the corrected region is reduced by the weighting function calculating means for correcting the hue direction. The skin color correction image processing apparatus according to claim 1 or 6.   2. The correction with higher flexibility for the characteristics of the device or the like is performed by setting a correction area in which the saturation direction area is asymmetric in the area to be corrected as a correction target area. 6. The skin color correction image processing apparatus according to 6.   The skin color correction image processing apparatus according to claim 6, wherein the weighting function calculation unit that corrects the saturation direction of the saturation uses a linear weighting function that is symmetrical with respect to the intensity of the saturation.   The skin color correction image processing apparatus according to claim 6 or 8, wherein the weighting function calculation means for correcting the saturation direction of the saturation uses a linear weighting function that is asymmetric in the intensity of the saturation.   The skin color correction image processing apparatus according to claim 6 or 8, wherein the weighting function calculation means for correcting the saturation direction of the saturation uses a nonlinear weighting function for the intensity of the saturation.   The skin color correction image processing apparatus according to claim 6 or 8, wherein the weighting function calculation means for correcting the saturation direction of the saturation uses an LUT that can realize an arbitrary weighting function in the intensity of the saturation. .   The skin color correction image processing apparatus according to claim 7, wherein the weighting function calculation unit that corrects the hue direction of saturation by the polar coordinate system uses a trapezoidal function as the weighting function in the hue.   The skin color correction image processing apparatus according to claim 7, wherein the weighting function calculation unit that corrects the hue direction of saturation by the polar coordinate system uses an LUT that can realize an arbitrary weighting function in the hue. Polar coordinate system conversion step for converting the color difference signals Cr and Cb of the orthogonal coordinate system to the polar coordinate system, a weighting function calculation step for correcting the hue direction of the hue by the polar coordinate system, and correction of the saturation direction of the hue by the polar coordinate system A weighting function calculating step for performing correction for concentrating the skin color region on the hue of an ideal skin color, and performing correction in the hue direction in the region subjected to the correction by the weighting function calculating step for correcting the hue direction. A skin color correction image processing method including reducing discontinuity of hue near a boundary and reducing discontinuity of hue near a boundary in a saturation direction in the corrected region,
A skin color correction image processing method, further comprising correcting the saturation of the skin color region.

Images