JP2010154494A - Device, method and program for video conversion - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To convert colors while suppressing excessive conversion of a contrast when a video is color-converted. <P>SOLUTION: An area division unit 22 divides the video into areas of a plurality of objects, and an RGB conversion parameter generation unit 23 generates histograms of the respective areas to generate RGB conversion parameters for distributing the histograms of the respective areas to a target distribution range. An RGB conversion function generation unit 25 generates color conversion functions by using the RGB conversion parameters. A video conversion unit 27 applies the color conversion functions to the video to convert the video, calculates conversion synthesis coefficients of respective pixels of an input video, and adjusts intensity of conversion of the video according as whether the conversion synthesis coefficients are large or small when generating a video from the input video and the video generated by the conversion. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a technique for automatically converting the gradation of an image, and more particularly to a technique for converting the gradation of an input image in a digital image processing apparatus.

  The histogram of the video shows the distribution of light and dark values for each pixel in the video. Contrast stretching is a technique for improving the contrast of an image by widely distributing the light and dark values from a low value to a high value for an image with a skewed light and dark value distribution. Through contrast stretching, an image that is too dark becomes brighter, and an image that is too bright becomes darker to maintain an appropriate brightness value. That is, the overall contrast balance of the video is improved by correcting the luminance value distribution of the video.

  FIG. 19 is a block diagram showing a configuration of a conventional contrast stretching apparatus. As shown in the figure, the conventional contrast stretching apparatus has a distribution calculation unit 42 and a stretching unit 43.

  The distribution calculation unit 42 counts the brightness value from the input video to obtain a histogram value or a histogram function. The target value input unit receives the target minimum value and the target maximum value, and outputs the input values to the stretching unit 43. The stretching unit 43 obtains a light / dark value having a minimum value and a maximum value using the histogram function obtained by the distribution calculation unit 42, and a target value in which the minimum value and the maximum value of the brightness of the input video are input to the target distribution input unit. Transform the histogram to match the lowest and target highest values. In the processing of the stretching unit 43, if the target minimum value is 0 and the target maximum value is 255, the image has brightness values from 0 to 255, and the contrast of the image increases.

  Further, the above-described histogram enlarging process is similarly applied to the R component, the G component, and the B component of the video, so that the contrast can be increased while maintaining the color. Alternatively, it is possible to create a video with an increased contrast with little color bias by performing histogram enlargement processing independently on the R component, the G component, and the B component. As prior art documents related to the present application, there are the following.

Japanese Patent Laid-Open No. 4-257082

  However, in the case of contrast stretching according to the conventional technology, when the contrast enlargement process is performed on the R component, the G component, and the B component in order to convert the color, the color of the background is excessively converted. The color balance may be lost. In particular, when only the skin color is converted, the color balance with the hair dyed in brown may be lost, and it cannot be adjusted as desired.

  The present invention has been made to solve the above-mentioned problems, and the problem is that the color can be converted while suppressing excessive contrast conversion during the color conversion of the image, and in particular, only the skin color can be converted. An object of the present invention is to provide a video conversion device that can adjust as desired without causing a loss of color balance with brown hair or the like even after conversion.

  According to a first aspect of the present invention, there is provided a video conversion device, a storage unit for storing an input video, a region dividing unit for reading the video, dividing the video into a plurality of regions, and storing the divided video in a storage unit, and a desired color component Histogram creation means for creating a histogram of each region; target range designating means for designating a target distribution range for each histogram; and the video so that each histogram is distributed in the target distribution range. Conversion parameter creation means for creating a conversion parameter for conversion, color conversion function creation means for creating a color conversion function using the conversion parameter, and video for converting the video by applying the color conversion function to the video Conversion means; conversion synthesis coefficient calculation means for calculating a conversion synthesis coefficient of each pixel in the input video; and the input video and the conversion Conversion adjustment image creating means for adjusting the intensity of image conversion according to the magnitude of the conversion synthesis coefficient when creating a conversion adjustment image from the image of the image, and the region dividing means And the histogram creating means creates a histogram of R component, G component, and B component for each region, and the conversion synthesis coefficient calculating means calculates a distance from the skin color.

  In the present invention, the image is divided into a plurality of regions, a conversion parameter is created to distribute the histogram of each region in the target distribution range, a color conversion function is created using the conversion parameter, and the color conversion function A composite coefficient for combining the image is calculated, a plurality of color conversion functions are combined using the composite coefficient, the image is converted by applying the combined color conversion function, and each pixel in the input image is converted. Color conversion of the video by calculating the conversion synthesis coefficient and adjusting the intensity of the video conversion according to the size of the conversion synthesis coefficient when creating the conversion adjustment video from the input video and the converted video In this case, the color can be converted while suppressing excessive contrast conversion.In particular, even if only the skin color is converted, the color balance with the hair that has been dyed in brown will not be lost. Can be adjusted.

  In the video conversion device, the conversion adjustment video creation means does not adjust the intensity of luminance conversion by using the luminance of the video after conversion, but adjusts the intensity of color conversion. . In the present invention, only the strength of color conversion is adjusted.

  In the video conversion apparatus, the conversion adjustment video creation means may multiply the difference between the R value, the G value, and the B value of the input video pixel and the converted video pixel by the conversion synthesis coefficient. A high-speed conversion table is created, and a difference between the luminance of the pixel of the input video and the luminance of the pixel of the converted video is calculated using the high-speed conversion table. In the present invention, the speed of video conversion is increased.

  In the video conversion device, the conversion adjustment video creation means includes means for converting the input video using the conversion synthesis coefficient, and at least R, G, and B values of pixels of the video after the conversion. Means for adding a predetermined value to one, and means for creating a conversion adjustment video from the video based on the R value, G value, and B value added with the predetermined value and the input video. It is characterized by having. In the present invention, for example, the skin color region is strongly converted.

  In the video conversion apparatus, the target range designation is performed in response to adding the predetermined value to at least one of an R value, a G value, and a B value of a pixel of a video after conversion using the conversion synthesis coefficient. The target distribution range designated by the means is changed. In the present invention, for example, the effect of strengthening the conversion of the skin color area is prevented from affecting the areas other than the skin color.

  A video conversion method according to a second aspect of the present invention is a video conversion method performed by a video conversion device, the step of storing an input video in a storage means, and the video is read out and divided into a plurality of areas and stored. Means for storing, a step of creating a histogram of each region in a desired color component, a step of designating a target distribution range for each histogram, and each histogram as the target distribution range Creating a conversion parameter for converting the video to be distributed, creating a color conversion function using the conversion parameter, creating a color conversion function using the conversion parameter, and Applying a color conversion function to the video to convert the video, and conversion synthesis coefficients of each pixel in the input video Calculating a conversion adjustment video from the input video and the converted video, and adjusting a conversion strength of the video according to the size of the conversion synthesis coefficient; In the step of dividing the video into a plurality of regions, a face region is detected, and in the step of creating the histogram, a histogram of R component, G component, and B component is created for each region, and the transform synthesis coefficient is calculated. In the calculating step, a distance from the skin color is calculated.

  According to a third aspect of the present invention, there is provided a video conversion program, a process for storing an input video in a storage unit, a process for reading the video, dividing it into a plurality of areas and storing it in a storage unit, and a desired color component A process for creating a histogram for each region; a process for specifying a target distribution range for each histogram; and a conversion for converting the video so that each histogram is distributed in the target distribution range. A process of creating a parameter, a process of creating a color conversion function using the conversion parameter, a process of creating a color conversion function using the conversion parameter, and applying the color conversion function to the video A conversion adjustment process, a conversion calculation coefficient for each pixel in the input video, a conversion adjustment coefficient from the input video and the converted video. In the process of adjusting the intensity of the conversion of the video according to the magnitude of the conversion synthesis coefficient, in the process of dividing the video into a plurality of areas, to detect the face area, In the process of creating the histogram, a histogram of R component, G component, and B component is created for each region, and in the process of calculating the conversion synthesis coefficient, the distance from the skin color is calculated.

  According to the present invention, it is possible to convert colors while suppressing excessive contrast conversion at the time of color conversion of an image. In particular, even if only skin color is converted, the color of hair that is dyed brown or the like The balance can be adjusted without any loss.

It is a block diagram which shows the whole structure of the video converter in one Embodiment. It is a block diagram which shows the structure of the area division part 22 for detecting face area information and black eye area information, and dividing an input image | video into a face area and a black eye area. It is a figure which shows an example of the process at the time of detecting a human face in the area | region division part. It is a figure which shows an example of RGB distribution information. It is a figure which shows an example of the probability density function (histogram function) showing the probability that the color value of the pixel p in a certain color component will be x. It is a figure which shows the cumulative distribution function about the probability density function shown in FIG. It is a figure which shows an example of the control value calculation function for calculating a control value from the accumulation ratio value A. It is a figure which shows what added the control value and the target value to each RGB distribution record of FIG. It is a figure which shows an example of the brightness | luminance histogram of an input image | video. It is a figure which shows the luminance histogram of the image | video after applying a color conversion function to the RGB data of the input image | video used for FIG. FIG. 5 is a diagram showing information obtained by adding R target value, G target value, and B target value information to each RGB distribution record of FIG. 4. It is a figure which shows an example of the look-up table for each color conversion of R component, G component, and B component. It is a block diagram which shows the structure of the image | video conversion part 27 which converts an input image | video, adjusting a color using conversion LUT. It is a figure which shows an example in the case of suppressing the change of the color of a background in the color adjustment part 272. FIG. It is a block diagram which shows the structure of the video conversion part 27 in 2nd Embodiment. It is a figure which shows an example in the case of suppressing the change of the color of a background in the color adjustment part 272 of 2nd Embodiment. It is a figure which shows the target RGB distribution record which set 20 values lower than R value of the target color to R target value. It is a figure which shows the change of the histogram in the process of converting into the color of a face area, maintaining the gradation of a background. It is a block diagram which shows the structure of the conventional contrast stretching apparatus.

[First Embodiment]
FIG. 1 is a block diagram showing an overall configuration of a video conversion apparatus according to the first embodiment. As shown in FIG. 1, the video conversion apparatus includes an input device 1, a central processing control device 2, and a storage device 3. The input device 1 includes a video signal input unit 11, a shutter input unit 12, and a target distribution selection unit 13. The central processing control device 2 includes a frame extraction unit 21, a region division unit 22, an RGB conversion parameter creation unit 23, a target The storage device 3 includes an RGB distribution generation unit 24, an RGB conversion function generation unit 25, a conversion LUT (look-up table) generation unit 26, and a video conversion unit 27. The storage device 3 stores information output by each unit, An RGB distribution storage unit 31 and a conversion LUT storage unit 32 are included.

  The video conversion apparatus may be configured as dedicated hardware, or may be configured using a general-purpose computer so that the processing of each unit is executed by a computer program. The results processed by each unit are stored in the storage unit so as to be readable.

  When the video signal is input, the video signal input unit 11 outputs the video signal to the frame extraction unit 21 and the video conversion unit 27. Here, the video output from the video signal input unit 11 is referred to as an input video.

  The input video handled in the present embodiment is a moving image composed of a plurality of frames, and each frame is a still image composed of a plurality of pixels. When the width of the input video is represented by width and the height is represented by height, each frame is composed of width × height pixels. Each pixel is data representing a color in the RGB color model, and is composed of an x coordinate value, a y coordinate value, an R value, a G value, and a B value in the frame. The x coordinate value is an integer value of 1 or more and width or less. Yes, the y-coordinate value is an integer value between 1 and height, and the R value, G value, and B value are integer values between 0 and 255, respectively.

  The target distribution selection unit 13 outputs the target RGB distribution name selected by the user to the target RGB distribution creation unit 24.

  The target RGB distribution creation unit 24 receives the target RGB distribution name and the RGB distribution information group, creates the target RGB distribution information, and outputs the created target RGB distribution information to the RGB conversion parameter creation unit 23. Note that the RGB distribution information group is input by reading from the RGB distribution storage unit 31.

  When a shutter operation is performed by the user, the shutter input unit 12 outputs a shutter signal that tells the frame extraction unit 21 of the shutter operation.

  The frame extraction unit 21 receives a video signal and a shutter signal and outputs a still image. The still image output from the frame extraction unit 21 is a one-frame still image extracted from the input video.

  The area dividing unit 22 divides the input image into a plurality of object areas.

  The RGB conversion parameter creation unit 23 creates a histogram of each region in the R component, the G component, and the B component, receives the target value output from the target RGB distribution creation unit 24 as an input, and creates an RGB conversion parameter.

  The RGB conversion function creating unit 25 receives the RGB conversion parameters, and the R conversion function and the G conversion function corresponding to each histogram so that the histograms of the R component, the G component, and the B component are in a distribution range that matches the RGB conversion parameters. And B conversion function.

The region dividing unit 22 receives a still image and outputs a region information group. The area information group includes a plurality of area information, and each area information includes an area name R and an area set Region _R. If the x coordinate value of pixel p is xp and the y coordinate value is yp, the region set Region _R is
It is expressed. The area dividing unit 22 may detect the entire area information, the face area information, and the black eye area information when the subject of the still image is a person. Alternatively, when the subject of the still image is an object, the entire region information, the subject region information, and the background region information may be detected. Hereinafter, an example in which the entire area information, the face area information, and the black eye area information are detected when the subject is a person will be described.

Here, the whole area information is composed of the area name WHOLE and the whole area set Region _WHOLE , the face area set is composed of the area name FACE and the face area set Region _FACE , and the black eye area set is the area name EYE and the black eye area set Region _EYE. It shall be comprised by. The whole region set Region _WHOLE is a set including all pixels included in the input still image.

  FIG. 2 is a block diagram showing a configuration of the area dividing unit 22 for detecting face area information and black eye area information and dividing an input image into a face area and a black eye area. The region dividing unit 22 includes an image scale converting unit 221, a detection target region extracting unit 222, a spatial frequency analyzing unit 223, a face candidate determining unit 224, a face candidate selecting unit 225, a skin region extracting unit 226, And a black eye region extraction unit 227.

  Selection of face candidates in the region dividing unit 22 uses Haar-type feature values, and uses an AdaBoost learning method and a method using a cascade structure detector (see, for example, JP-A-2006-293720).

  The detection target region cutout unit 222 cuts out a region having a predetermined size from the scale-converted image. However, the shape of the region handled here may be an arbitrary shape.

  The spatial frequency analysis unit 223 performs spatial frequency filtering on the cut out region. For example, an AdaBoost learning method and a Haar type feature amount detection method having a cascade detector structure may be used.

  The face candidate determination unit 224 determines whether the region can be a face region candidate based on the output value of the spatial frequency analysis unit 223.

  The face candidate selection unit 225 selects and outputs a face area candidate having the maximum area among face area candidates determined to be face candidates.

The skin area extraction unit 226 receives a face area candidate and outputs a face area set Region _FACE . The face area set Region _FACE is composed of pixels included in face area candidates whose pixel color is skin color. In other words, the Region _FACE has xp as the x coordinate value of pixel p, yp as the y coordinate value, rp as the R value, gp as the G value, and bp as the B value.
It is expressed. Note that the RGB values handled as skin color are set in advance. For example, in the YUV color model, a method is used in which the Cb value is -40 or more and -20 or less and the Cb value is 15 or more and 35 or less as skin color.

The black eye region extraction unit 227 outputs a black eye region set Region _EYE with the face region candidate as an input. The black eye region set Region _EYE is composed of pixels with low luminance values among the pixels included in the face region candidate. More specifically, among the pixels included in the face area candidate, a 1% pixel group is set as a black eye area set in order of increasing luminance value. FIG. 3 is a diagram illustrating an example of processing when the area dividing unit 22 detects a human face.

  The target RGB distribution creation unit 24 receives the target RGB distribution name and the RGB distribution information group and outputs target RGB distribution information. The RGB distribution information group is read from the RGB distribution storage unit 31. The RGB distribution information group includes a plurality of RGB distribution information, and each RGB distribution information has an RGB distribution name and includes a plurality of RGB distribution records. FIG. 4 is a diagram illustrating an example of RGB distribution information. Each RGB distribution record is composed of a record number, a region name, a cumulative ratio value, and a target color. Here, the cumulative ratio value is a decimal value between 0 and 1.

  The color components handled in this embodiment will be described. Luminance (luminance: Y component in the YCbCr color model), Cb (Cb component in the YCbCr color model), Cr (Cr component in the YCbCr color model), black (in the CMYK color model) K component), red (R component in RGB color model), green (G component in RGB color model), blue (B component in RGB color model), hue (H component in HSV color model), saturation (HSV color model) S component) and brightness (V component in the HSV color model), and the minimum value is 0 and the maximum value is 255 in all color components.

  The RGB conversion parameter creation unit 23 receives a still image, a region information group, and target RGB distribution information, and creates and outputs an RGB conversion parameter. The RGB conversion parameter is created by the following procedure.

First, a color value calculation function that returns the color value of the pixel p in the color component Component
And Where g (Component, p, x)
If the pixel p included in the region set Region _R in the pixel set P is extracted, the probability that the color value of the pixel p in the color component Component is x, that is, color (p, Component) = x. Probability density function (histogram function) frequency (P, Region _R , Component, x)
It is defined as

For example, region information of whole image region information (region name WHOLE, region set Region _WHOLE ), face region information (region name FACE, region set Region _FACE ), and black eye region information (region name EYE, region set Region _EYE ) If the set of pixels representing an image is Py and the luminance color component is Y, the luminance histogram of the entire area is expressed as frequency (Py, Region _WHOLE , Y, x). In addition, the luminance histogram of the face region is represented as frequency (P, Region _FACE , Y, x), and the luminance histogram of the black eye region is represented as frequency (P, Region _EYE , Y, x).

The cumulative distribution function cumulative (P, Region _R , Component, x) derived from the probability density function frequency (P, Region _R , Component, x) representing the histogram of the color component Component is
It is expressed.

If the probability density function is represented by the curve of FIG. 5, the cumulative distribution function is a curve represented as shown in FIG. Furthermore, the inverse function of the cumulative distribution function with respect to x is
It is expressed. Here, y is a value satisfying the relationship y = cumulative (P, Region _R , Component, x).

Here, the control value calculation function control (P, Region _R , Component, A) is defined as follows using the cumulative distribution function cumulative.

However, here, the set of pixels included in the input still image is P, the region name in the target RGB distribution record is R, the region set corresponding to the region name R created by dividing the still image into regions is Region _R , The color component is Component, and the cumulative ratio value is A.

By applying this control value calculation function to the cumulative ratio value A, as shown in FIG.
Thus, the control value C corresponding to the cumulative ratio value A is calculated.

Further, the control value sequence _{[C k] k = 1,2,} ..., a _n (although C1 ≦ C2 ≦ ... ≦ Cn) , the target value sequence _{[T k] k = 1,2,} ..., with _n, The color conversion function convert ([C _k ] _{k = 1, 2,} ... _{, N} , [T _k ] _{k = 1, 2,} ... _{, N} , x) for converting the color value x
It is defined as

  Here, for example, a luminance histogram of a still image is distributed as shown in FIG. 9, and control values [C1, C2, C3, C4] and target values [T1, T2, T3, T4] are determined as shown in the figure. FIG. 10 shows a histogram after the luminance value is converted by applying the color conversion function convert ([C1, C2, C3, C4], [T1, T2, T3, T4], x). become. As shown in the figure, the color conversion function converts the histogram so that the control value matches the target value.

  Here, as shown in FIG. 11, the R value of the target color is set as the R target value, the G value as the G target value, and the B value as the B target value for each target RGB distribution record.

Further, assuming that a set of pixels of a still image is P, an area name of a certain target RGB distribution record Record is AREA, and an accumulation ratio value is A, an R control value C _R is calculated using a control value calculation function control.
R control value sequence {C _{R, k} } _{k = 1, 2,} ... _{, N} (where C _{R, 1} ≦ C _{R, 2} ≦ ... ≦) by calculating control values for all target RGB distribution records. C _{R, n} ) can be calculated.

Similarly, G control value C _G of Record is
And the G control value sequence {C _{G, k} } _{k = 1, 2,} ... _{, N} can be calculated.

Also, the B control value C _B of Record is
And B control value sequence {C _{B, k} } _{k = 1, 2,} ... _{, N} can be calculated.

  The RGB conversion function creation unit 25 receives RGB conversion parameters and outputs an R conversion function, a G conversion function, and a B conversion function.

Here, the R control value sequence {C _{R, k} } _{k = 1, 2,} ... _{, N} (where C _{R, 1} ≦ C _{R, 2} ≦ ... ≦ C _{R, n} ) and the R target value sequence {T _{R , k} } _{k = 1, 2,} … _{, n} , an R conversion function convert _R (x) that converts R values
Is defined as, G control value column _{{C G, k} k =} 1,2, ..., and _n, G target value column _{{T G, k} k =} 1,2, ..., the G value by using the _n G conversion function convert _G (x)
B control value column _{{C B, k} k =} 1,2, ..., B to perform and _n, B target value column _{{T B, k} k =} 1,2, ..., the conversion of B values with _n Convert function convert _B (x)
It is defined as

The RGB conversion function creation unit 25 outputs the R conversion function convert _R (x), the G conversion function convert _G (x), and the B conversion function convert _B (x).

  The conversion LUT creation unit 26 receives the R conversion synthesis function, the G conversion synthesis function, and the B conversion synthesis function, and outputs an R conversion LUT, a G conversion LUT, and a B conversion LUT. Specifically, an LUT as shown in FIG. 12 corresponding to values obtained by applying the R conversion function, the G conversion function, and the B conversion function to all R values, G values, and B values that can be taken as values, respectively. It is created and stored in the conversion LUT storage unit 32.

  The video conversion unit 27 receives the input video and the color conversion LUT and outputs a converted video.

  FIG. 13 is a block diagram illustrating the configuration of the video conversion unit 27 that converts the input video while adjusting the color using the conversion LUT. The video conversion unit 27 includes an LUT reference unit 271 and a color adjustment unit 272, converts all the pixels of the input video, and outputs the converted video.

The LUT reference unit 271 receives an input pixel value (r, g, b) as an input, and uses the LUT conversion pixel value (r ′, g ′, b ′) when the conversion LUT is used.
Asking. However, LUTr is an R conversion LUT, LUGg is a G conversion LUT, and LUTb is a B conversion LUT.

  The color adjustment unit 272 receives the input pixel value (r, g, b) and the LUT conversion pixel value (r ′, g ′, b ′), and, for example, a color adjustment pixel value (in which a change in the color of the background is suppressed) r ", g", b ") is calculated.

  For example, a procedure in the case where the LUT conversion of the skin color region is strong and the LUT conversion of the region other than the skin color is weak will be described below.

(Example)
FIG. 14 is a diagram illustrating an example of a case where the color adjustment unit 272 suppresses a change in background color. In FIG. 14, first, an LUT converted video of an input video is created, and luminance information and color information of the LUT converted video are calculated. Also, the skin color information and the color information of the input video are calculated, and the color information of the LUT converted video and the color information of the color information of the input video are combined based on the skin color information.

  In the synthesis of the above-described color information, the region with strong skin chromaticity uses the color information of the LUT conversion image, and the region with low skin chromaticity uses the color information of the input image, thereby strongly performing LUT conversion of the skin color region It becomes possible to weaken LUT conversion other than skin color.

  By adjusting the luminance information of the LUT-converted video created by the above procedure and the synthesized color information, the color can be adjusted. Hereinafter, a specific color adjustment procedure will be described.

First, the color value calculation function colorBalance (r, g, b) for calculating the color value (colorR, colorG, colorB) of a certain pixel value (r, g, b)
And

Here, if the assumed skin color RGB values are (skinR, skinG, skinB), the skin color tone values (skinColorR, skinColorG, skinColorB) are
Is calculated by

Next, a function skinDistance (r, g, b) for calculating a distance from a skin color of a certain pixel value is defined as follows, with the width of the skin color tint value being (skinRangeR, skinRangeG, skinRangeB).
Note that (skinR, skinG, skinB) uses values such as (200, 140, 130), and (skinRangeR, skinRangeG, skinRangeB) uses values such as (15, 5, 5).

Next, the function normalizeSkinDistance (r, g, b), which performs the normalization of the distance obtained by skinDistance (r, g, b), using the minimum distance minDist and the maximum distance maxDist
And

Furthermore, the LUT conversion synthesis coefficient calculation function alphaLUT (r, g, b) is used with the minimum synthesis coefficient minAlpha and the maximum synthesis coefficient maxAlpha.
And

Next, the luminance synthesis coefficient alphaL and the tint synthesis coefficient alphaColor are defined. For example, in the case where the luminance is converted according to the LUT conversion and the color is converted around the skin color region, for example.
And

From the above definition, the synthetic luminance synthL is
And calculate. here
Is an RGB value that changes only the luminance value of the input pixel, and the composite color value (synthR, synthG, synthB) of the input image is
The color adjustment pixel value (r ", g", b ")
And output from the color adjustment unit 272.

When always alphaL = 1.0 and alphaColor = 0.0, the color adjustment unit 272 performs only luminance conversion by LUT conversion. In this case, the speed can be increased by the following procedure. First, the fast conversion table FCT (Fast Conversion Table)
Create as. Here, FCTr is R conversion FCT, FCTg is G conversion FCT, and FCTb is B conversion FCT.

At this time, the function for calculating the luminance difference of (LUTr [r], LUTg [g], LUTb [b]) with respect to (r, g, b), the luminance difference calculating function yDiff (r, g, b) is
The color adjustment pixel value (r ", g", b ") is
Thus, it can be calculated at high speed only by referring to the table three times and adding five times.

  By performing pixel conversion according to the above procedure on all the pixels of the input video, it is possible to output a converted video with adjusted color.

  As described above, according to the first embodiment, it is possible to convert colors while suppressing excessive contrast conversion when converting the color of an image. In particular, even if only the skin color is converted, the color balance with the hair dyed in brown is not lost, and it can be adjusted as desired.

[Second Embodiment]
Next, a second embodiment of the present invention will be described. Here, the description will focus on differences from the first embodiment.

  FIG. 15 is a block diagram illustrating a configuration of the video conversion unit 27 in the second embodiment.

  The video conversion unit 27 includes an LUT reference unit 271, a color adjustment unit 272, and an RGB value processing unit 273, converts all the pixels of the input video, and outputs a converted video.

The LUT reference unit 271 receives an input pixel value (r, g, b) as an input, and determines an LUT conversion pixel value when the conversion LUT is used.
Asking. However, LUTr is an R conversion LUT, LUGg is a G conversion LUT, and LUTb is a B conversion LUT.

The RGB value processing unit 273 adds, for example, liftR, liftG, and liftB to the R value, G value, and B value, respectively, and converts the converted pixel value (r ′, g ′, b ′).
Ask for. Note that liftR, liftG, and liftB are predetermined values and are stored in the video conversion device. liftR, liftG, and liftB may each be a negative value. Further, at least one of liftR, liftG, and liftB may be 0 unless the other is 0.

  The color adjustment unit 272 receives the input pixel value (r, g, b) and the converted pixel value (r ′, g ′, b ′) and, for example, the color adjustment pixel value (r that suppresses the change in the color of the background) ", g", b ") is calculated.

  For example, the procedure in the case where the conversion of the skin color region is strong and the conversion of the region other than the skin color is weak will be described below.

(Example)
FIG. 16 is a diagram illustrating an example of a case where a change in background color is suppressed by the color adjustment unit 272 according to the second embodiment. In FIG. 16, first, an LUT converted image of the input image is created, the RGB values of the LUT converted image are processed, and luminance information and color information of the processed RGB values are calculated. Also, the skin color information and the color information of the input video are calculated, and the color information of the processed RGB value and the color information of the color information of the input video are combined based on the skin color information.

  In the synthesis of the above-described color information, the skin color region color conversion is performed by using the processed RGB value color information for regions with high skin chromaticity and the color information of the input video for regions with low skin chromaticity. It is possible to weaken the color conversion other than the skin color.

  By adjusting the luminance information of the processed RGB value created by the above procedure and the synthesized color information, the color can be adjusted.

  Further, in the specific tint adjustment procedure described in the first embodiment, the tint adjustment pixel value (r ", g", b "calculated by Expression 27 and output from the tint adjusting unit 272 is used. In the second embodiment, the color represented by) is changed in color by the LUT conversion and RGB value processing for the skin color area, and no change in color is performed for the background area.

  If the degree of histogram conversion is large, the background color gradation may be lost due to LUT conversion. In such a case, it is possible to perform color conversion while maintaining the gradation of the background color by weakening the degree of color conversion by LUT conversion and performing color adjustment using RGB value processing instead. It becomes.

  Specifically, when setting the R target value, G target value, and B target value for each target RGB distribution record, the R value of the target color is the R target value, the G value is the G target value, and the B value. Is set as the B target value as it is, as shown in FIG. 17, for some target RGB distribution records, for example, a value 20 lower than the R value of the target color is set as the R target value.

  Further, when such a setting is made, the RGB value processing unit 273 performs, for example, video conversion by adding 20 to the R value with liftR = 20, liftG = 0, liftB = 0 (STEP 1 in FIG. 18). ).

  Then, histogram conversion is performed using the target RGB distribution record of FIG. 17 (STEP 2 of FIG. 18). Thereby, the color of the background color can be adjusted.

  As a result, it is possible to convert the color of the face area to the target color while maintaining the gradation of the background area.

  By performing pixel conversion according to the above procedure on all the pixels of the input video, it is possible to output a converted video with adjusted color.

DESCRIPTION OF SYMBOLS 1 ... Input device 2 ... Central processing control device 3 ... Memory | storage device 11 ... Video signal input part 12 ... Shutter input part 13 ... Target distribution selection part 21 ... Frame extraction part 22 ... Area division part 23 ... RGB conversion parameter creation part 24 ... Target RGB distribution creation unit 25 ... RGB conversion function creation unit 26 ... conversion LUT creation unit 27 ... video conversion unit 31 ... RGB distribution storage unit 32 ... conversion LUT storage unit 42 ... distribution calculation unit 43 ... stretching unit 221 ... image scale conversion Unit 222 ... detection target region extraction unit 223 ... spatial frequency analysis unit 224 ... face candidate determination unit 225 ... face candidate selection unit 226 ... skin region extraction unit 227 ... black eye region extraction unit 271 ... LUT reference unit 272 ... color adjustment unit 273 ... RGB value processing section

Claims (11)

Storage means for storing the input video;
Area dividing means for reading out the video, dividing it into a plurality of areas and storing it in a storage means;
A histogram creating means for creating a histogram of each region in a desired color component;
Target range specifying means for specifying a target distribution range for each histogram;
Conversion parameter creating means for creating a conversion parameter for converting the video so that each histogram is distributed in the target distribution range;
Color conversion function creating means for creating a color conversion function using the conversion parameter;
Video conversion means for converting the video by applying the color conversion function to the video;
Conversion synthesis coefficient calculation means for calculating a conversion synthesis coefficient of each pixel in the input video;
Conversion adjustment video creating means for adjusting the strength of the video conversion according to the magnitude of the conversion synthesis coefficient when creating a conversion adjustment video from the input video and the converted video,
The area dividing means detects a face area,
The histogram creating means creates a histogram of R component, G component, and B component for each region,
The image conversion apparatus, wherein the conversion synthesis coefficient calculation means calculates a distance from the skin color. The video according to claim 1, wherein the conversion adjustment video creation means adjusts the strength of the color conversion while not adjusting the strength of the luminance conversion by using the luminance of the video after the conversion. Conversion device. The conversion-adjusted image creation means creates a high-speed conversion table by multiplying the difference between the R value, G value, and B value of the input image pixel and the converted image pixel by the conversion synthesis coefficient. The video conversion apparatus according to claim 1, wherein a difference between a luminance of a pixel of the input video and a luminance of a pixel of the converted video is calculated using the high-speed conversion table. The conversion adjustment video creation means includes:
Means for converting the input video using the conversion synthesis coefficient;
Means for adding a predetermined value to at least one of the R value, G value, and B value of the pixel of the converted image;
2. The video conversion according to claim 1, further comprising means for creating a conversion adjustment video from the video based on the R value, the G value, and the B value added with the predetermined value and the input video. apparatus.   The target specified by the target range specifying means in correspondence with adding the predetermined value to at least one of the R value, G value, and B value of the pixel of the image after conversion using the conversion synthesis coefficient. 5. The video conversion apparatus according to claim 4, wherein the distribution range of the video is changed. A video conversion method performed by a video conversion device,
Storing the input video in a storage means;
Reading the video, dividing it into a plurality of areas and storing it in a storage means;
Creating a histogram of each region in a desired color component;
Designating a target distribution range for each of the histograms;
Creating a conversion parameter for converting the video so that each histogram is distributed in the target distribution range;
Creating a color conversion function using the conversion parameters;
Creating a color conversion function using the conversion parameters;
Applying the color conversion function to the video to convert the video;
Calculating a transform synthesis coefficient for each pixel in the input video;
Adjusting the conversion strength of the video according to the magnitude of the conversion synthesis coefficient when creating a conversion adjustment video from the input video and the converted video,
In the step of dividing the video into a plurality of areas, a face area is detected,
In the step of creating the histogram, a histogram of R component, G component, and B component is created for each region,
In the step of calculating the conversion synthesis coefficient, a distance from the skin color is calculated. The step of creating the conversion adjustment video does not adjust the intensity of luminance conversion by using the luminance of the video after conversion, but adjusts the intensity of color conversion. Video conversion method. In the step of creating the conversion adjustment image, a high-speed conversion table is obtained by multiplying the difference between the R value, the G value, and the B value of the input image pixel and the converted image pixel by the conversion synthesis coefficient. The video conversion method according to claim 6, wherein a difference between the brightness of the input image pixel and the brightness of the pixel of the converted image is calculated using the high-speed conversion table. The step of creating the conversion adjustment video includes:
Converting the input video using the conversion synthesis coefficient;
Adding a predetermined value to at least one of the R value, G value, and B value of the pixel of the converted video;
7. The image conversion according to claim 6, further comprising the step of creating a conversion adjustment image from the image based on the R value, the G value, and the B value added with the predetermined value and the input image. Method.   The target specified by the target range specifying means in correspondence with adding the predetermined value to at least one of the R value, G value, and B value of the pixel of the image after conversion using the conversion synthesis coefficient. The video conversion method according to claim 6, wherein the distribution range of the video is changed.   6. A video conversion program for causing a computer as the video conversion device according to claim 1 to execute processing performed by the video conversion device.