JP2006237798A - Device for correcting color image of skin of human being and its method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To conduct the color-image correction processing of the skin of a human being generating no existence of discontinuousness and no color shade after correction in the correction processing in a region of a color of the skin of the human being such as a TV image. <P>SOLUTION: A device for correcting the color image of the skin of the human being is constituted by being provided with a polar coordinate system transformation means transformed into a polar coordinate system so as to easily correct a color as mentioned later, a weighting W(θ) arithmetic means in the angular direction of a hue, and a weighting W(r) arithmetic means in the radial direction of the hue from a Cr/Cb orthogonal coordinate system. A formula 5 to which a weighting function W(θ) in the angular direction θ of the hue is reflected is carried out to the formula 4. The annihilation of colors in an area aehda and the area fbcg shown in Fig. 2 can be avoided, and the color shade can be removed. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to image processing that is mounted on video equipment such as a television, a camera, and a DVD and corrects a human skin color area.

ITU-R BT. In the 601 standard, the relationship between luminance Y, color difference Cr / Cb and red R, green G, and blue B is defined by Equation 1 below, but the color of human skin varies depending on the race, but is usually red. It is distributed in the region between 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.

  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 Formula 1 to obtain Formula 2 shown below.

  FIG. 8 shows a red (R), yellow (Ye), green (G), cyan (C), blue (B), magenta (in a rectangular 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. 8 is changed to the polar coordinate system shown in the following Equation 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. 9 is a diagram showing a conventional skin color correction region in digital image processing.

The skin color correction is performed by Equation 4 shown below. The correction area is 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 k is, the stronger the correction is.

  FIG. 2 is a diagram illustrating an area before skin color correction and an area change after correction.

  Conventional correction skin color is distributed in the area surrounded by aefbcghda, but the corrected skin color 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.

  However, since the colors in the area surrounded by “aehda” and the area surrounded by “fbcgf” shown in FIG. 2 disappear from the corrected image, the corrected image has an area boundary “aefb” and “dhgc”. , There are discontinuities in color above and below the border. For these reasons, there is a problem in that color unevenness occurs in the corrected image.

  Further, the area to be corrected on both sides of the flesh color ideal position θ1 may not be symmetric depending on the characteristics of the CCD and the panel. Therefore, it is desirable to correct the asymmetric area on both sides of θ1.

  Similarly, correction of asymmetric areas is also required depending on race.

  The present invention has been made in view of the above-described conventional problems, and in a correction process in a human skin color region, there is no discontinuity after correction and no color unevenness occurs in the human skin color image correction device and its It aims to provide a method.

  In order to solve the above problems, a human skin color image correction apparatus and method according to the present invention have the following features.

  A human skin color image correction apparatus according to the present invention is a skin color correction image processing apparatus that performs correction to narrow a skin color area to an area close to an ideal skin color, and uses color difference signals Cr and Cb in an orthogonal coordinate system. A polar coordinate system converting means for converting to a polar coordinate system; and a weighting function calculating means for correcting the hue direction by the polar coordinate system; and a hue direction in the region corrected by the weighting function calculating means for correcting the hue direction. It is characterized by reducing color unevenness in the vicinity of the boundary.

  Further, the human skin color image correction device according to the present invention further comprises weighting function calculation means for correcting the saturation direction by the polar coordinate system, and color unevenness near the boundary in the saturation direction in the corrected region. It is characterized by reducing.

  Further, the human skin color image correction apparatus according to the present invention uses the correction area in which the angle in the hue direction is asymmetric in the area to be corrected as a correction target area. In contrast, correction with higher flexibility is performed.

  In the human skin color image correction apparatus according to the present invention, the weighting function calculation means for correcting the hue direction uses a linear weighting function that is symmetrical with respect to the hue angle.

  In the human skin color image correction apparatus according to the present invention, the weighting function calculation means for correcting the hue direction uses a linear weighting function that is asymmetric in the hue angle.

  In the human skin color image correction apparatus according to the present invention, the weighting function calculation means for correcting the hue direction uses a nonlinear weighting function for the hue angle.

  In the human skin color image correction apparatus according to the present invention, the weighting function calculation means for correcting the hue direction uses an LUT capable of realizing an arbitrary weighting function in the hue angle.

  In the human skin color image correction apparatus according to the present invention, the weighting function calculation means for correcting the saturation direction by the polar coordinate system uses an inverted trapezoidal function as the weighting function in the saturation. And

  In the human skin color image correction apparatus according to the present invention, the weighting function calculation means for correcting the saturation direction by the polar coordinate system uses an LUT that can realize an arbitrary weighting function in saturation. To do.

  A human skin color image correction method according to the present invention is a skin color correction image processing method for performing a correction to narrow a skin color area to an area close to an ideal skin color, and the color difference signals Cr and Cb of an orthogonal coordinate system are obtained. A polar coordinate system converting step for converting to a polar coordinate system; and a weighting function calculating step for correcting the hue direction by the polar coordinate system, and the hue direction in the region corrected by the weighting function calculating step for correcting the hue direction. It is characterized by reducing color unevenness in the vicinity of the boundary.

  In addition, the human skin color image correction method according to the present invention further includes a weighting function calculation step for correcting the saturation direction by the polar coordinate system, and the color unevenness near the boundary of the saturation direction in the corrected region. It is characterized by reducing.

  Further, the human skin color image correction method according to the present invention uses the correction area in which the angle in the hue direction is asymmetric in the area to be corrected as a correction target area. In contrast, correction with higher flexibility is performed.

  In the human skin color image correction method according to the present invention, the weighting function calculation step for correcting the hue direction is performed using a linear weighting function symmetric with respect to the hue angle.

  In the human skin color image correction method according to the present invention, the weighting function calculation step for correcting the hue direction is performed using an asymmetric linear weighting function in the hue angle.

  In the human skin color image correction method according to the present invention, the weighting function calculation step for correcting the hue direction may be performed using a nonlinear weighting function at the hue angle.

  In the human skin color image correction method according to the present invention, the weighting function calculation step for correcting the hue direction uses an LUT capable of realizing an arbitrary weighting function in the hue angle.

  In the human skin color image correction method according to the present invention, the weighting function calculation step of correcting the saturation direction by the polar coordinate system may be performed by using a function of an inverted trapezoidal weighting function in the saturation. It is characterized by.

  In the human skin color image correction method according to the present invention, the weighting function calculation step for correcting the saturation direction by the polar coordinate system uses an LUT that can realize an arbitrary weighting function in saturation. To do.

  As described above, according to the human skin color image correction apparatus and method according to the present invention, it is possible to eliminate the color loss and discontinuity by providing the weighting functions in θ and r.

  Further, according to the human skin color image correction apparatus and method according to the present invention, it is possible to perform correction in consideration of differences in race and device characteristics by using an asymmetric correction area in the hue of the ideal skin color. it can.

  Further, according to the human skin color image correction apparatus and method according to the present invention, in addition to implementation by software, a complex nonlinear weighting function or the like is generated in order to provide flexibility in hardware implementation. In doing so, correction processing can be performed in a short time using a look-up table (“LUT”).

  Hereinafter, an embodiment of a human skin color image correction apparatus according to the present invention will be described in detail with reference to the drawings.

  In the description of this embodiment, the human skin color image correction apparatus according to the present invention will be described in detail. However, the human skin color image correction method according to the present invention will be described. Since the method is adopted by the apparatus, the human skin color image correction method according to the present invention is included in the following description.

  FIG. 1 is a block diagram showing a first embodiment of a human skin color image correcting apparatus according to the present invention.

  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. 8 into 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 10, a storage means 11 for storing the ideal hue θ1 of the center of the skin color, as shown in FIG. 9, a hue angle direction weighting W (θ) calculation means 12, The hue radial weight W (r) computing means 13, the hue angular weight W (θ) computing means 12, and the hue radial weight W (r) computing means 13 are input to the constant. Storage means for storing the coefficient value k.

  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) of the hue angle direction θ in the above-described Expression 4, the following Expression 5 is obtained. The means for calculating the function value of the function W (θ) is the weighting W (θ) calculation means 12.

  The image corrected by executing Equation 5 shown above can avoid the disappearance of the colors of the areas aehda and area fbcg shown in FIG.

  Here, FIG. 3 is a diagram showing an example of the weighting function at θ, and shows a linear function that is easy to calculate.

Further, the following Expression 6 is obtained by reflecting the weighting function in the saturation direction r in Expression 4. The means for calculating the function value of the function W (r) is the weighting W (r) calculation means 13.

  The corrected image obtained by executing Equation 6 shown above can maintain the continuity of the colors above and below the boundary aefb and the boundary dhgc shown in FIG. 2, so that color unevenness can be reduced.

  Here, FIG. 4 is a diagram illustrating an example of a weighting function in r. As shown in FIG. 4, Δr1 and Δr2 are positive numbers between 0 and (r2-r1) / 2.

Further, Equation 5 and Equation 6 are combined into the same equation, and the following Equation 7 is obtained by reflecting both the weighting functions of θ and r.

  In this way, two weighting functions in θ and r are provided, and the above formula 7 in which these functions are incorporated is executed, and the corrected image is caused by the disappearance of color and discontinuity in both directions of θ and r. Color unevenness can be improved.

  Next, correction processing in the case of an asymmetric correction area with respect to θ1 will be described.

  FIG. 5 is a diagram showing an asymmetric correction area having different Δθ1 and Δθ2 on both sides of θ1.

The asymmetric correction area shown in FIG.

Although the weighting function W (r) in the asymmetric correction area shown in FIG. 5 does not change, the weighting function W (θ) in θ changes to a function shown in FIG. 7 with respect to FIG. 3, for example. By reflecting the weighting functions W (r) and W (θ) in Equation 8, Equation 9 shown below is obtained.

  When RGB is 0-255 (8 bits), the range of the luminance signal Y obtained from Equation 2 is 0-255, and the range of the color difference signal Cr / Cb is -128 to +127. Also, the range of θ obtained from Equation 3 is 0-360, and the range of r is 0-180. Here, for simplicity, the calculation range is a positive number.

  The ideal hue of skin color is θ1 = 123 °. However, Δθ2 is set wider than Δθ1 because Asian skin has a lot of yellow.

Here, other parameters to be substituted in Expression 7 or Expression 9 are summarized in Expression 10.

  Next, specific shapes of the weighting functions W (θ) and W (r) when correction is performed in the asymmetric correction area will be described below.

The weighting function W (θ) at θ shown in FIG. 6 can be expressed by Equation 11 shown below.

Further, the weighting function W (r) in r of FIG. 4 can be expressed by the following formula 12.

The weighting function at θ shown above uses a linear function, but generality is not lost even if a nonlinear function is used. For example, a non-linear weighting function using a cosine function in θ can be expressed as Equation 13 shown below. The weighting function in this case has the shape shown in FIG.

  By substituting the related parameter values of Equation 10 into Equations 11 and 12, specific weighting functions can be obtained respectively. Similarly, the weighting coefficients W (θ) and W (r) obtained in this manner and the related parameters of Expression 10 are substituted into Expression 9 to obtain an expression that realizes specific skin color correction.

  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, CrCb coordinate system of ITU-R BT.709).

  Further, the weighting function for θ may require a complicated nonlinear weighting function that cannot be expressed by an analytical expression or a function that cannot be calculated by hardware. In such a case, this problem can be solved by implementing a weighting function using a LUT (Look Up Table). For example, when the expression 13 is realized using an LUT having 128 elements, (Δθ1 + Δθ2) between θ1−Δθ1 and θ1 + Δθ2 is first divided into 128 equal parts, and one equal part is represented by Δθ. If this LUT is written as Lθ (i), i is a 0-127 positive number. If the value of W (θ) at iΔθ is stored in Lθ (i), creation of the LUT is completed. When referring to the LUT, Lθ (i) may be used with i of iΔθ.

  Similarly, the weighting function in saturation (r direction) can also be converted into an LUT.

1 is a block configuration diagram showing a first embodiment of a human skin color image correction device according to the present invention. FIG. It is a figure which shows the area before skin color correction, and the area change after correction | amendment. It is an example of a weighting function in θ, and is a diagram showing a linear function that is easy to calculate. It is the figure which showed the example of the weighting function in r. It is a figure which shows the asymmetrical correction area which has (DELTA) (theta) 1 and (DELTA) (theta) 2 which are different in the both sides of (theta) 1. It is a figure which shows weighting function W ((theta)) in the case of correct | amending in an asymmetrical correction area. It is a figure which shows the weighting function W ((theta)) using the cosine function which is a nonlinear function in the case of correct | amending in an asymmetric correction area. Hue of red (R), yellow (Ye), green (G), cyan (C), blue (B), and magenta (M) in an orthogonal coordinate system with the color difference Cb as the horizontal axis and the color difference Cr as the vertical axis FIG. It is a figure which shows the area | region of the conventional skin color correction.

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 W (θ) calculation means 13 Saturation direction weighting W (r) calculation means 14 Constant coefficient value k Storage means for storing

Claims (18)

A skin color correction image processing apparatus that performs correction to narrow a skin color area to an area close to an ideal skin color,
Polar coordinate system conversion means for converting the color difference signals Cr and Cb of the orthogonal coordinate system into a polar coordinate system;
Weighting function calculation means for correcting the hue direction by the polar coordinate system,
A skin color-corrected image processing apparatus, characterized in that color unevenness near a boundary in a hue direction in the corrected region is reduced by a weighting function calculation means for correcting the hue direction. Furthermore, a weighting function calculating means for correcting the saturation direction by the polar coordinate system is provided,
The skin color correction image processing apparatus according to claim 1, wherein color unevenness near a boundary in a saturation direction is reduced in the corrected region.   In a region to be corrected, a correction region in which the angle in the hue direction is asymmetric is set as a correction target region, so that more flexible correction is performed on the skin color of the race and the characteristics of the device. The skin color correction image processing apparatus according to claim 1 or 2.   The skin color correction image processing apparatus according to claim 1, wherein the weighting function calculation unit that corrects the hue direction uses a linear weighting function that is symmetric with respect to the hue angle.   4. The skin color correction image processing apparatus according to claim 3, wherein the weighting function calculating means for correcting the hue direction uses a linear weighting function that is asymmetric with respect to the hue angle.   4. The skin color correction image processing apparatus according to claim 1, wherein the weighting function calculation means for correcting the hue direction uses a non-linear weighting function for the hue angle.   The skin color correction image processing apparatus according to claim 1 or 3, wherein the weighting function calculation means for correcting the hue direction uses an LUT that can realize an arbitrary weighting function in the hue angle.   3. The skin color correction image processing apparatus according to claim 2, wherein the weighting function calculation means for correcting the saturation direction by the polar coordinate system uses an inverted trapezoidal function as the weighting function for the saturation.   The skin color correction image processing apparatus according to claim 2, wherein the weighting function calculation means for correcting the saturation direction by the polar coordinate system uses an LUT that can realize an arbitrary weighting function in saturation. A skin color correction image processing method for performing correction for narrowing a skin color area to an area close to an ideal skin color,
A polar coordinate system converting 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 by the polar coordinate system,
A skin color correction image processing method characterized by reducing color unevenness near a boundary in a hue direction in the corrected region by a weighting function calculation step for correcting the hue direction. Furthermore, a weighting function calculation step for correcting the saturation direction by the polar coordinate system is provided,
The skin color correction image processing method according to claim 10, wherein color unevenness near a boundary in a saturation direction in the corrected region is reduced. Furthermore, a weighting function calculating means for correcting the saturation direction by the polar coordinate system is provided,
The skin color correction image processing apparatus according to claim 1, wherein color unevenness near a boundary in a saturation direction is reduced in the corrected region.

In a region to be corrected, a correction region in which the angle in the hue direction is asymmetric is set as a correction target region, so that more flexible correction is performed on the skin color of the race and the characteristics of the device. The skin color correction image processing method according to claim 10 or 11.   The skin color correction image processing method according to claim 10, wherein the weighting function calculation step of correcting the hue direction is performed using a linear weighting function that is symmetric with respect to the hue angle.   The skin color correction image processing method according to claim 12, wherein the weighting function calculation step of correcting the hue direction is performed using a linear weighting function that is asymmetric with respect to the hue angle.   13. The skin color correction image processing method according to claim 10, wherein the weighting function calculation step for correcting the hue direction is performed using a nonlinear weighting function at the hue angle.   13. The skin color correction image processing method according to claim 10, wherein the weighting function calculation step for correcting the hue direction uses an LUT that can realize an arbitrary weighting function in the hue angle.   12. The skin color correction image processing method according to claim 11, wherein in the weighting function calculation step of correcting the saturation direction by the polar coordinate system, the weighting function in the saturation is calculated using an inverted trapezoidal function. .   12. The skin color correction image processing method according to claim 11, wherein the weighting function calculation step of correcting the saturation direction by the polar coordinate system uses an LUT that can realize an arbitrary weighting function in saturation.

Images