JP2000261685A - Method of improving color image quality and device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To emphasize both local and global contrasts with simple calculation, without necessitating judgment by a user, and to improve the quality of a color picture by transforming the R, G, and B values of a color picture into a color space according to the color sense of a person, and using the contrast of the whole average with the local average. SOLUTION: A color space transforming part 2 transforms the pixel values of all the pixels of an inputted color image into values on a color space, and a whole average calculating part 3 calculates the whole mean value by averaging each chromaticity value and luminance value of the whole pixels of the color image. A neighborhood average calculating part 5 calculates a neighborhood mean value by averaging each chromaticity and luminance value of neighborhood pixels extracted by a neighborhood pixel extracting part 4. A pixel value correcting part 9 corrects each chromaticity value and the luminance value of the pixels based on the pixel value variable decided by a positive image pixel value variable deciding part 7 and a pixel value variable decided by a negative image pixel value variable deciding part 8, and calculates a pixel correction value by calculating the logarithm of the contrast of each corrected result.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a color image input from an image sensor, a digital camera, or the like.
A color image quality improvement method that enhances local contrast and global contrast at the same time while enhancing subtle changes in luminance and chromaticity values in bright and dark areas, and outputs the image in accordance with the output range. And an apparatus.

[0002]

2. Description of the Related Art In general, in the field of visual technology involving color image processing or image display, higher definition color images are required. There is a need for a color image quality improvement method of converting a color image into a highly accurate color image.

As this type of color image quality improvement method, there are an exponential conversion method represented by gamma correction and a logarithmic conversion method. When an extreme enhancement of an image is desired, there is a method of contrast enhancement of pixel values of the image by histogram flattening.

Here, in the method based on logarithmic conversion, FIG.
As shown in (1), the pixel value I of the input image is converted to a value that increases the whole. As shown in the figure, since the smaller the pixel value has the larger increment, it can be understood that the contrast of the dark portion is mainly emphasized.

Further, in the method using exponential conversion, FIG.
As shown in (1), when the exponent is smaller than 1, a conversion curve similar to logarithmic conversion is drawn, and conversion is performed to increase the pixel value as a whole. Therefore, the contrast of the dark part is enhanced. on the other hand,
When the exponent is larger than 1, the pixel value is converted to a value that reduces the pixel value as a whole, as shown in FIG. Therefore, the contrast of the bright part is enhanced.

[0006] The histogram flattening method is a technique of appropriately dispersing a cluster of pixel value distributions around and performing contrast enhancement in accordance with the distribution, and is actually an enhancement method among various image enhancement methods. It is a method with a remarkable effect.

[0007] In addition to these, although a black-and-white image is a target instead of a color image, an image enhancement method that is excellent in resolution improvement is to reduce the contrast to some extent in the middle part, which originally has good visibility, and to reduce the contrast in a bright part and a dark part. In Japanese Patent Publication No. Hei 8-8, an image quality improving method and apparatus for enhancing contrast is adopted.
No. 21079 discloses this.

[0008]

However, the above-described color image quality improving method has the following problems.

In the method using logarithmic conversion, the contrast in a section having a small pixel value is emphasized, but there is a problem that the contrast in a section having a large pixel value is suppressed.

In the method using the exponential conversion, it takes time and effort for the user to determine the index by judging each image. When the index is smaller than 1, the contrast of the section having a small pixel value is emphasized, but the section having a large pixel value is emphasized. There is a problem that the contrast is suppressed. When the index is larger than 1, the contrast in the section where the pixel value is large is emphasized,
There is a problem that the contrast in a section having a small pixel value is suppressed.

In the method using histogram flattening, while the effect of enhancing global contrast is remarkable, there is a problem that a relatively small local contrast indicating a change or the like within the region is suppressed.

In the method described in Japanese Patent Publication No. 8-21079, a sharp image is locally obtained due to a high local contrast enhancement effect, while the global contrast tends to be reduced. . Also, since the method is for a black and white image, it has nothing to do with improving the image quality of a color image.

[0013] In summary, the above-described decrease in contrast reduces the practical use of the image data even for high-resolution image data, so that it is difficult to realize a high-definition color image. Cause.

In addition, the above-described image quality improvement methods are as follows.
In the case of a black-and-white image, although the image quality is improved to some extent, it cannot be said that it is still sufficient from the viewpoint of realizing a higher-definition color image.

The present invention has been made in view of the above circumstances, and enhances both local and global contrasts by simple calculations without requiring user judgment to improve the quality of a color image. It is an object of the present invention to provide a color image quality improving method and apparatus which can be performed.

Another object of the present invention is to enhance the substantial use of image data by enhancing local and global contrasts in accordance with human color perception, thereby achieving high-definition color images. An object of the present invention is to provide a color image quality improving method and apparatus which can be realized.

[0017]

A first gist of the present invention will be described. The technique disclosed in Japanese Patent Publication No. 8-21079 uses an inverse S-shaped conversion to linearly correct an S-shaped response characteristic in human vision.
As described above, while the local contrast is enhanced, the global contrast tends to be reduced.

In order to improve this tendency, the present invention
A method of adjusting the chromaticity and a method of adjusting the chromaticity are provided respectively. Shine
To adjust the brightness, the brightness of the entire image (average of the overall brightness)
Value l _mean) To the local brightness (the brightness
Average l_ave) To determine the coefficient corresponding to the overall brightness
Local brightness by emphasizing the difference between
Improve global contrast while maintaining contrast
Is being planned. In addition, this brightness adjustment
Using space CIE-LAB makes luminance independent of chromaticity
It is preferable from the viewpoint that the adjustment can be performed with high accuracy.

On the other hand, as for chromaticity adjustment, the present invention
ー Based on human color vision properties
Also, regarding color vision, the overall hue (average of overall chromaticity)
Value rg _mean, Yb_mean) And local tint (color near pixel)
Average value of degree rg_ave, Yb_ave)
Global contrast while maintaining local contrast.
Improving trust. Note that this chromaticity adjustment
Use color space based on human color vision mechanism
Is preferred from the viewpoint of natural colors.

Here, for example, FIG. 11 is a schematic diagram showing a color vision mechanism in a human outer knee-like body (De Valois, RLa).
nd De Valois, KK: A multi-stage color model, V
ision Research, Vol.33, No.8, pp.1053-1065, 1993.
). When the eye receives light, the L, M, and S cones in the retina output signals when the eye receives light, and the signals are added in various combinations in the outer knee to the cerebrum. The output result indicates that a human produces color vision such as red, yellow, green, and blue including brightness and darkness.

In FIG. 11, L indicates an L cone that senses a long wavelength, M indicates an M cone that senses a medium wavelength, and S
Indicates an S cone that senses a short wavelength. The symbol “o” indicates an ON center shape of ganglion cells in which the center responds positively (excitation) and the periphery responds negatively (suppressed). Similarly, the sign "-" indicates an OFF center shape of a ganglion cell that produces a reaction having a polarity opposite to that of the ON center shape.

Thus, for example, the M of the OFF center type
The cone is shown as "-Mo". For example, "L
The sum of “o”, “Mo” and “So” indicates that “brightness” is perceived, and the sum of “Lo”, “−Mo” and “So” indicates that “red” is perceived. .

The present invention converts the R, G, and B values of a color image into a color space represented by a chromaticity diagram so as to match human color perception based on such a color vision mechanism. By using the ratio to the local average, the chromaticity contrast is improved to realize a high-definition color image.

Further, the present invention considers that the receptive field in the human side suppression mechanism is substantially concentric, and determines the local brightness and chromaticity of the image as the average value l _ave of the brightness near the pixel and the color. As shown by the average values rg _ave and yb _{ave of the} degrees, the local contrast itself is also improved.

Next, the second gist of the present invention will be described. In the first framework, in order to further enhance the global contrast, the overall brightness and chromaticity of the image are converted into the average luminance value l _mean and the chromaticity by considering the light-dark adaptation mechanism and the chromatic adaptation mechanism. The average value rg _ave , yb _ave is represented by a correction value kl obtained by inverse S-shape conversion.
(L _mean ), krg (rg _mean ), and kyb (yb _mean ) are represented by intermediate values Vl _mid , Vrg _mid , Vyb of the output range of the display system.
By performing a linear transformation to match the output range before and after the average luminance and chromaticity, corresponding to _mid
It emphasizes global contrast.

Now, based on the gist of the present invention described above,
Specifically, the following solution is realized. The invention corresponding to claim 1 converts pixel values r, g, b in each pixel of a color image into chromaticity values a *, b * and a luminance value l * on a uniform color space CIE-LAB, and The average value l * _mean of the luminance values l * of the pixels is obtained, and for each of the pixels, the luminance values l * 1 to l * of the i pixels in the vicinity of the pixel
* The average value l * _ave of i is obtained, and the average value l * of the whole is obtained.
_Find each coefficient corresponding to the ratio between _mean and the average value l * _ave of the neighborhood, multiply each coefficient by the relative value of the average value l * _ave of the neighborhood in the positive image to obtain a positive pixel variable δl * _{While obtaining the posi} , the respective coefficients are multiplied by the relative value of the average value l * _ave of the neighborhood in the negative image to obtain a negative pixel variable δl * _nega , and the positive pixel variable δl * _posi is obtained by the positive image in the positive image. This relative value is corrected in addition to the relative value of the luminance value l * of the pixel, and the relative value is corrected by adding the negative pixel variable δl * _nega to the relative value of the luminance value l * of the pixel in the negative image. Then, the luminance value l * of the pixel is corrected by the logarithm of the ratio of the corrected relative values, and the correction result kl * (l *
) Is normalized within the output range, and the normalized result Vl * (l *) and each of the chromaticity values a * and b * are inversely converted from the uniform color space CIE-LAB to the rgb color space. This is a color image quality improvement method for constructing and outputting the entire image based on the inverse conversion result.

According to a second aspect of the present invention, in the color image quality improving method according to the first aspect, the normalization includes the correction result k of the overall average value l * _mean.
This is a color image quality improvement method executed so that l * (l * _mean ) corresponds to the intermediate value Vl * _mid of the output range.

Further, according to the present invention, there is provided an image input means for inputting a color image, and a pixel value r of all pixels of the color image input by the image input means.
g and b are defined as chromaticity values a * and 上 の on the uniform color space CIE-LAB.
a color space conversion means for converting into b * and a luminance value l *, and an overall average calculation for calculating an overall average value l * _mean by averaging the luminance values l * of all the pixels obtained by the color space conversion means Means, and for each pixel obtained by the color space conversion means, each luminance value l * 1 to l * of i pixels in the vicinity of the pixel.
a neighboring image extracting means for extracting i, and respective luminance values l * 1 to l * i of the neighboring pixels extracted by the neighboring image extracting means
And neighborhood average calculating means for calculating the neighborhood average value l * _ave the average, the overall total is calculated by the average calculating means averages l * _{m ean} and the neighboring average value calculated by the neighboring average calculation means l * coefficient αl * _posi , α corresponding to the ratio to _ave
a coefficient calculating means for calculating l * _nega , and a pixel of a positive image based on a nearby average value l * _{ave of the} pixel obtained by the nearby average calculating means and a coefficient αl * _posi calculated by the coefficient calculating means. A positive image pixel value variable determining means for determining the value variable δl * _posi , and the coefficient αl * _nega calculated by the coefficient calculating means and the neighborhood average value l * _{ave of the} pixel obtained by the neighborhood average calculating means. A negative image pixel value variable determining means for determining the pixel value variable δl * _nega of the negative image; a pixel value variable δl * _posi determined by the positive image pixel value variable determining means; and the negative image pixel value variable determining means. The luminance value l of the pixel is corrected based on the pixel value variable δl * _nega determined by, and the logarithm of the ratio of the correction result is obtained to obtain a pixel correction value kl *
A pixel value correction means for calculating (l *), and a maximum value kl * _max and a minimum value kl * _min among all the pixel correction values kl * (l *) calculated by the pixel value correction means. Correction value maximum / minimum value calculation means, and the maximum value kl * _max and the minimum value kl * calculated by the correction value maximum / minimum value calculation means
Pixel value normalization for calculating a normalized conversion value Vl * (l *) by normalizing the pixel correction value kl * (l *) calculated by the pixel value correction means within the output range based on _min . Means for converting the normalized converted value Vl * (l *) of each pixel calculated by the pixel value normalizing means and the chromaticity values a * and b * obtained by the color space converting means into the uniform color space. CIE
A color image quality improving apparatus comprising: an image output unit that performs an inverse conversion from LAB to an rgb color space and forms and outputs an entire image based on the obtained inverse conversion result.

According to a fourth aspect of the present invention, there is provided the color image quality improving apparatus according to the third aspect, wherein the pixel value normalizing means and the image output means are replaced by the overall average calculating means. The average value l * _mean and the coefficient αl * _posi , α calculated by the coefficient calculating means.
Based on l * _nega , the correction value k of the overall average l * _mean
means for calculating an overall average correction value for calculating l * (l * _mean );
The pixel correction value kl * (l *) of each pixel calculated by the pixel value correction means is compared with the correction value kl * (l * _mean ) calculated by the overall average correction value calculation means, and the pixel correction is performed. The correction value kl * (l) corresponding to the value kl * (l *)
* _mean ) is large, the pixel correction value kl * (l *) is linearly converted to an output range section smaller than the intermediate value Vl * _mid within the corresponding output range, and the pixel correction value kl * (l *) is obtained.
The correction value corresponding to kl * when (l * _{m ean)} are equal converts the intermediate value Vl * _mid within the output range range corresponding pixel correction value kl * (l *), the pixel correction value kl * (L
When the corresponding correction value kl * (l * _mean ) is smaller than *), the pixel correction value kl * (l *) is linearly converted to an output range section larger than the intermediate value Vl * _mid in the corresponding output range. By doing so, each pixel correction value kl * (l *) is normalized within the range of the output range, and the normalized conversion value Vl * (l *)
), And a normalized conversion value Vl * of each pixel calculated by the pixel correction value conversion means.
(L *) and the chromaticity values a * and b * obtained by the color space conversion means are converted from the uniform color space CIE-LAB to rgb.
A color image quality improving apparatus comprising: a normalized image output unit that performs inverse conversion to a color space and forms and outputs the entire image based on the obtained inverse conversion result.

Further, according to the invention corresponding to claim 5, the color
-The pixel values r, g, b for each pixel of the image
Each chromaticity value rg, yb and luminance value l are converted to
Average value r of each of degree value rg, yb and luminance value l
g_mean, Yb_mean, L_meanAnd calculate each of the color images.
For each pixel, each chromaticity of i pixels in the vicinity of the pixel
Values rg1 to rgi, yb1 to ybi, and luminance values 11 to
average value rg of each of li_ave, Yb_ave, L_aveSeeking
The average value rg of the whole_mean, Yb_mean, L_meanAnd before
Average value rg of the neighborhood_ave, Yb_ave, L_aveTo the ratio
Find each corresponding coefficient and set each coefficient as positive
Average value rg of the neighborhood in the image_ave, Yb_ave, L_ave
Multiply the relative value of each by the positive pixel variable δrg_posi,
δyb_posi, Δl_{pos i} And calculate each of the coefficients
Average value rg of the neighborhood in a negative image _ave, Yb
_ave, L_aveNegative pixel change by multiplying each relative value of
Quantity δrg_nega, Δyb_nega, Δl_nega The positive
Pixel variable δrg_posi, Δyb_posi, Δl_{p osi} The positive
Each chromaticity value rg, yb and luminance value l of the pixel in the image
Correcting these relative values in addition to the relative values of
The negative pixel variable δrg _nega, Δyb_nega, Δl
_nega Are the respective chromaticity values rg of the pixel in the negative image,
yb and the relative value of the luminance value l,
Modify the values and log the ratio of these modified relative values.
Each chromaticity value rg, yb and luminance value l of the pixel are complemented, respectively.
Correct these correction results krg (rg), kyb (yb),
color that normalizes kl (l) within the output range
-Image quality improvement method.

[0031] The invention corresponding to claim 6 is the present invention.
5. The method for improving image quality of a color image corresponding to 5.
As the normalization, the entire average value rg_mean, Y
b_mean, L _meanCorrection results krg (rg), kyb
(Yb_mean), Kl (l_mean) Is the middle of the output range
Value Vrg_mid, Vyb_mid, Vl_mid To respond to each
This is a color image quality improvement method to be executed.

Further, according to the invention corresponding to claim 7, the color
Image input means for inputting an image, and the image input means
Pixel values r of all pixels of the color image input by the means,
g and b are converted to chromaticity values rg and yb and a luminance value l in the color space.
Color space conversion means for converting, and the color space conversion means
Chromaticity values rg, yb and luminosity of all pixels of the obtained color image
Average each of the degree values l to obtain the overall average value rg_mean, Y
b_mean, L_meanMeans for calculating the average, and the color
For each pixel of the color image obtained by the space conversion means,
Each chromaticity value rg, yb and luminance value l of the pixel and its vicinity
Chromaticity values rg1 to rgi, yb of i pixels in the vicinity
Neighboring images for extracting 1-ybi and luminance values 11-li
Extraction means, and the neighborhood extracted by the neighborhood image extraction means.
Chromaticity values rg1 to rgi, yb1 to ybi of neighboring pixels
Average the respective luminance values l1 to li to obtain a neighborhood average value r
g_ave, Yb_ave, L_aveMeans for calculating the neighborhood average
And the overall average value calculated by the overall average calculation means
rg_mean, Yb_mean, L_meanAnd the neighborhood average calculation means
Average value rg calculated from_ave, Yb_ave, L_ave
Coefficient αrg corresponding to the ratio_posi, Αyb_posi, Αl_posi ,
αrg_nega, Αyb_nega, Αl_nega Coefficient calculating means for calculating
And a neighborhood average value calculated by the neighborhood average calculation means
rg_ave, Yb_ave, L_aveAnd the coefficient calculation means
Issued coefficient αrg_posi, Αyb_posi, Αl_posi Based on
The pixel value variable δrg of the positive image_posi, Δyb_posi,
δl_posi Positive image pixel value variable determining means for determining
And a neighborhood average value calculated by the neighborhood average calculation means
rg_ave, Yb_ave, L_aveAnd the coefficient calculation means
Issued coefficient αrg_nega, Αyb_nega, Αl_nega Based on
The pixel value variance δrg of the negative image_nega, Δyb_nega,
δl_nega Negative image pixel value variable determining means for determining
Determined by the positive image pixel value variable determining means.
Pixel value variable δrg_posi, Δyb_posi, Δl_posi And before
Determined by the negative image pixel value variable determining means
Pixel value variation δrg_nega, Δyb_nega, Δl_nega Based on each of
And the chromaticity values rg and yb and the luminance value l of the pixel, respectively.
Pixel correction by calculating the logarithm of the ratio of each correction result
Calculate the values krg (rg), kyb (yb), and kl (l)
Pixel value correction means, calculated by the pixel correction means
All pixel correction values krg (rg), kyb (yb), KL
(L) the maximum value krg_max, Kyb_max, Kl_{m ax} When
Minimum value krg_min, Kyb_min, Kl_min And correction to calculate
Value maximum / minimum value calculating means, and the correction value maximum / minimum value calculating means.
Maximum value krg calculated by step_max, Kyb _max, Kl_max
 And minimum value krg_min, Kyb_min, Kl_min Based on
The pixel correction value k calculated by the pixel value correction means.
rg (rg), kyb (yb), kl (l) are output range
Normalized conversion values Vrg (rg), Vyb
(Yb), a pixel value normalizing means for calculating Vl (l);
Normalization of each pixel calculated by the pixel value normalization means
Based on the conversion values Vrg (rg), Vyb (yb), Vl (l)
And image output means for composing and outputting the entire image.
This is a color image quality improvement device.

[0033] The invention corresponding to claim 8 is the present invention.
7. In the color image quality improvement device corresponding to
In place of the prime value normalizing means and the image output means,
Overall average value rg calculated by body average calculation means_mean,
yb_mean, L_meanAnd calculated by the coefficient calculating means.
Coefficient αrg_posi, Αyb_posi, Αl_posi , Αrg_nega, Αyb
_nega, Αl_nega Based on the total average value rg_mean,
yb_mean, L_meanCorrection value krg (rg_mean), Kyb (y
b_mean), Kl (l_mean) To calculate the overall average correction value
Output means and each pixel calculated by the pixel value correction means
Pixel correction values krg (rg), kyb (yb), kl (l)
And the correction value calculated by the overall average correction value calculation means.
krg (rg_mean), Kyb (yb_mean), Kl (l_mean)
And the pixel correction values krg (rg) and kyb
(Yb), the correction value krg corresponding to kl (l)
(Rg_mean), Kyb (yb_mean), Kl (l_mean) Is large
, The pixel correction values krg (rg), kyb (yb), k
l (l) is the intermediate value Vr within the corresponding output range
g_mid, Vyb_mid, Vl_mid Smaller output range section
To the pixel correction values krg (rg), kyb (y
b), kl (l) and the corresponding correction value krg (r)
g_mean), Kyb (yb_mean), Kl (l_meanEquals)
Pixel correction values krg (rg), kyb (yb), k
l (l) is the intermediate value Vr within the corresponding output range
g_mid, Vyb_mid, Vl_mid To the pixel correction value
Corresponds better than krg (rg), kyb (yb), kl (l)
The correction value krg (rg_mean), Kyb (yb_mean), K
l (l_mean) Is small, the pixel correction value krg (rg),
kyb (yb), KL (l) within the corresponding output range
Intermediate value of Vrg_mid, Vyb _mid, Vl_mid Greater output
By linearly converting to a range section, each pixel correction value k
rg (rg), kyb (yb), kl (l) are output range
Normalized conversion values Vrg (rg), Vyb
(Yb), pixel correction value conversion means for calculating Vl (l)
And each pixel calculated by the pixel correction value conversion means.
Normalized conversion values Vrg (rg), Vyb (yb), Vl (l)
The normalized image output that composes and outputs the entire image based on
And a color image quality improving device provided with a power unit.

(Operation) Accordingly, the present invention is characterized in that:
The corresponding invention takes the above measures,
Claims 1 and 3 relate to luminance (l *), and claims 5 and 7 refer to color
Degrees (rg, yb and luminance value 1) are input respectively.
The distribution characteristics of the chromaticity values rg, yb and the luminance values l, l * of the image
The overall average of the input image without any assumptions
Value rg _mean, Yb_mean, L_mean, L *_mean Seeking
, The local chromaticity values rg1 to rgi, yb1 to yb
Adapts to the distribution of i and luminance values l1-li, l * 1-l * i
The degree of contrast enhancement in the vicinity of each pixel.
Since it can be adjusted only by processing, the judgment of the user
Improve color image quality with simple calculations
Can be made.

Further, the local and global contrasts are enhanced in accordance with the color vision of a human, so that the substantial utilization of the image data can be increased and a high-definition color image can be realized.

The inventions corresponding to the second, fourth, sixth, and eighth aspects have the same effects as the first, third, fifth, and seventh aspects, and the second and fourth aspects relate to luminance (l *). The terms 6 and 8 are chromaticity (rg, yb and luminance value 1), respectively, and are correction values krg of the average values rg _mean , yb _mean , l _mean , and l * _mean respectively.
(Rg _mean ), kyb (yb _mean ), kl (l _mean ), k
l * (l * _mean ) is the intermediate value Vrg with good sensitivity of output range
Since normalization is performed in association with _mid , Vyb _mid , Vl _mid , and Vl * _mid , local contrast and global contrast can be enhanced simultaneously.

[0037]

Embodiments of the present invention will be described below with reference to the drawings.

(First Embodiment) FIG. 1 shows a first embodiment of the present invention.
FIG. 3 is a block diagram illustrating a configuration of a color image quality improvement device according to the embodiment. This image quality improvement device includes an image input unit 1, a color space conversion unit 2, an overall average calculation unit 3, a neighborhood image extraction unit 4, a neighborhood average calculation unit 5, a coefficient calculation unit 6, a positive image pixel value variable determination unit 7, a negative Image pixel value variation determination unit 8, pixel value correction unit 9, pixel value correction control unit 10, pixel correction value storage unit 11, pixel correction value maximum / minimum value calculation unit 12, pixel value normalization unit 13, pixel value normalization control A storage unit 15, a pixel value normalized conversion value storage unit 15, and an image output unit 16. Further, the data obtained by the components at the preceding stage can be used by the components at the subsequent stage, and the same applies to the following embodiments.

Here, the image input unit 1 receives a color image and outputs a pixel value r corresponding to RGB for each pixel.
(Red), g (green), and b (blue), which can be read from the color space conversion unit 2 or the like.
For example, an image sensor, a monitor camera, a digital camera, or the like can be appropriately used. The image input unit may simply be a storage device such as a hard disk in which a color image is stored.

The color space conversion unit 2 converts the pixel values r, g, b of all the pixels of the color image input by the image input unit 1 into values in the color space, and converts the conversion result into an overall average calculation unit 3 It has a function to give to.

Specifically, the color space conversion unit 2 converts the pixel values r, g, b of all the pixels into the uniform color space CIE
A function of converting each of the chromaticity values a * and b * on the LAB into a luminance value l * and giving the luminance value l * to the overall average calculation unit 3; It should be noted that the luminance value l * is converted by the color space conversion unit 2 to the overall average calculation units 3 to 3 from the last stage to the inverse conversion to the rgb color space by the image output unit 16 at the last stage.
During the image output unit 16, arithmetic processing is performed in the same manner as the luminance value 1 of the chromaticity adjustment described below. Therefore, for the sake of simplicity, the calculation processing of the luminance value l * will be described using the calculation processing of the luminance value 1 of the chromaticity adjustment between the overall average calculation unit 3 and the image output unit 16.

On the other hand, in the case of chromaticity adjustment, the color space conversion unit 2 converts the pixel values r, g, b of all pixels into respective chromaticity values rg, yb of a three-dimensional color space having R, G, and B axes. And the luminance value l
And provides the obtained chromaticity values rg, yb and luminance value l to the overall average calculation unit 3. here,
The chromaticity value rg indicates a value on an RG plane indicated by the R axis-G axis of the three-dimensional color space, and the chromaticity value yb indicates a yb plane indicated by the G axis-B axis of the three-dimensional color space. The above values are shown. In other words, each chromaticity value rg, yb is represented on the chromaticity diagram by a broadband (rg) axis and a narrowband (y-) axis of a perceived color space.
b) Each value of the axis is shown.

The overall average calculating section 3 uses the color space converting section 2
Each chromaticity value rg, yb of all pixels of the obtained color image and
The luminance values 1 are each averaged to obtain an overall average value rg_mean, Y
b_mean, L _meanIs calculated, and the overall average value rg_mean, Y
b_mean, L_meanIs provided to the coefficient calculation unit 6
You.

The neighborhood image extraction unit 4 calculates, for each pixel of the color image obtained by the color space conversion unit 2, each chromaticity value rg, yb and luminance value l of the pixel, and i Chromaticity values rg1 to rgi, yb1 to ybi of the pixel
And a neighborhood average calculation unit 5 that extracts brightness values 11 to li
It has a function to give to.

The neighborhood average calculation section 5 calculates the chromaticity values rg1 to rgi, y of the neighborhood pixels extracted by the neighborhood image extraction section 4.
b1~ybi and luminance values l1~li respectively averaged near average value rg _ave, yb _ave, calculates l _ave, these neighborhood average rg _ave, yb _{av e,} coefficient calculating section 6 l _ave, positive images It has a function to be provided to the pixel value variation determining unit 7 and the negative image pixel value variation determining unit 8.

The coefficient calculator 6 calculates the overall average values rg _mean , yb _mean , l _mean received from the overall average calculator 3 and the neighborhood average values rg _ave , yb _ave , l received from the neighborhood average calculator 5.
The coefficient αrg _{posi of the} positive image corresponding to the ratio to _ave ,
αyb _posi , αl _posi and negative image coefficient αr
g _nega , αyb _nega , αl _nega , the function of giving the coefficients αrg _posi , αyb _posi , αl _posi to the positive image pixel value variable determining unit 7, and the functions αrg _nega , αyb
_nega and αl _nega are provided to the negative image pixel value variable determining unit 8.

The positive image pixel value variable determining unit 7
The neighborhood average value rg calculated by the neighborhood average calculation unit 5_ave,
yb_ave, L_aveAnd the coefficient calculated by the coefficient calculator 6
Number αrg_posi, Αyb_posi, Αl_posi And based on
Pixel value variable δrg of the live image_{po si}, Δyb_posi, Δl_posi To
This pixel value variable δrg_posi, Δyb
_posi, Δl_posi To the pixel value correction unit 9
I have.

The negative image pixel value variable determining unit 8 calculates the neighborhood average value rg _ave ,
yb _ave, and l _ave, coefficient Arufarg _nega calculated by the coefficient calculation section 6, αyb _nega, based on the .alpha.l _nega, pixel value variable [Delta] rg _{ne ga} negative image,? YB _nega, is intended to determine the .DELTA.l _nega , These pixel value variables δrg _nega , δyb
_nega and δl _nega are provided to the pixel value correction unit 9.

The pixel value correction unit 9 determines the pixel value variation δrg _posi , δ determined by the positive image pixel value variation determination unit 7.
yb _posi , δl _posi and the pixel value variables δrg _nega , δyb _nega , δ determined by the negative image pixel value variable determination unit 8
l _nega each chromaticity value rg,
yb and the luminance value 1 are corrected, and the logarithm of the ratio of each correction result is obtained to obtain pixel correction values krg (rg), kyb (yb), k
l (l), and this pixel correction value krg
(Rg), kyb (yb), and kl (l) are provided to the pixel value correction control unit 10.

The pixel value correction control section 10 includes a pixel value correction section 9.
Pixel correction values krg (rg), kyb (yb), k
l (l) to the pixel correction value storage unit 11 and pixel correction values krg (rg) and kyb (y) for all pixels
b) and a function of controlling the neighboring image extracting unit 4 to the pixel value correcting unit 9 to calculate kl (l).

The pixel correction value storage unit 11 stores the pixel correction values krg (rg) and kyb (y) received from the pixel value correction control unit 10.
b) and kl (l) are stored in a memory or the like, and can be read from the pixel correction value maximum / minimum value calculation unit 12 and the pixel value normalization means.

The pixel correction value maximum / minimum value calculation unit 12 calculates all the pixel correction values krg stored in the pixel correction value storage unit 11.
(Rg), kyb (yb), kl (l), the maximum value k
_{rg ma x, kyb max, kl} max and the minimum value krg _min, kyb
_It has a function of calculating _min and kl _min and providing the calculated _min and kl _min to the pixel value normalizing unit 13.

The pixel value normalizing section 13 receives the maximum values krg _max , kyb _max ,
kl _max and minimum value krg _min, kyb _min, based on the kl _min, pixel correction value of each pixel calculated by the pixel value correcting unit 9 krg (rg), kyb ( yb), the output range of kl (l) Normalized within the range, and the respective normalized conversion values Vrg (r
g), Vyb (yb) and Vl (l) are calculated, and these normalized conversion values Vrg (rg), Vyb (yb) and Vl (l) are provided to the pixel value normalization control unit 14.

The pixel value normalization control unit 14 receives the normalized conversion values Vrg (rg) and Vyb (y) received from the pixel value normalization unit 13.
b) and Vl (l) to the pixel value normalized conversion value storage unit 15, and a normalized conversion value Vrg (r) for all pixels.
g), Vyb (yb), and Vl (l).

The pixel value normalized conversion value storage unit 15 stores each normalized conversion value Vrg (r) received from the pixel value normalization control unit 14.
g), Vyb (yb), and Vl (l) are stored in a memory or the like, and can be read from the image output unit 16.

In the case of chromaticity adjustment, the image output unit 16 normalizes the converted values Vrg (rg), Vyb (yb), and Vl (l) of each pixel stored in the pixel value normalized converted value storage unit 15. Has the function of composing and outputting the entire image based on the

In the case of brightness adjustment, the image output unit 16
Is obtained by converting the normalized conversion value Vl * (l *) of each pixel stored in the pixel value normalized conversion value storage unit 15 and the chromaticity values a * and b * obtained by the color space conversion unit 2 into uniform colors. Space CIE-L
It has a function of performing an inverse conversion from AB to the rgb color space and forming and outputting the entire image based on the obtained inverse conversion result.

Next, a method for improving the image quality of a color image by the apparatus for improving the image quality of a color image configured as described above will be described with reference to the flowchart of FIG.

First, a color image is input by the image input unit 1 such as an image sensor, a monitor camera, or a digital camera. At this time, the lower limit of the domain that each pixel value of the input image can take is represented by r _L , g _L , b _L , and the upper limit is represented by r _U , g _U , b _U. Note that these suffixes “L” and “U” represent the lower limit and the upper limit for the converted chromaticity value and luminance value 1 in the same manner.

In the case of chromaticity adjustment, the color space converter 2 adjusts the pixel values r, of all the pixels of the color image input by the image input unit.
g and b are converted into chromaticity values rg and yb and luminance values 1 for each pixel as shown in the following equations (1) to (3) (ST
1) The obtained chromaticity values rg, yb and luminance value 1 are given to the overall average calculation unit 3.

[0061]

(Equation 1)

Further, the color space converter 2 adjusts
The pixel values r, g, and b of all pixels of the color image input by the image input unit are converted into chromaticity values a in the uniform color space CIE-LAB.
*, B * and luminance value l *, and the obtained luminance value l *
Is given to the overall average calculator 3. However, since the operation of the subsequent luminance adjustment is the same as the calculation processing of the luminance value 1 in the chromaticity adjustment as described above, the overall average calculation unit 3 to the image output unit 1
Until the middle of step 6, the description will be made including the calculation processing of the luminance value 1 in the chromaticity adjustment. That is, the operation of the brightness adjustment is performed by the following equations (6), (9), (12), (15),
The luminance value l and the value l based on it in the expressions (18), (21), (24), (27) and (30), etc.
_mean , l1-l8, l _ave , αl _posi , αl _nega ,
l _L , l _U , βl _posi , βl _nega , δl _posi , δl
_nega , kl (l), kl _max , kl _min , Vl _max ,
Vl _min and Vl (l) are represented by the luminance value l * and its values l * _mean , l * 1 to l * 8, l * _ave , αl * _posi ,
αl * _nega , l * _L , l * _U , βl * _posi , βl * _nega , δl *
_posi , δl * _nega , kl * (l *), kl * _max , kl * _min ,
Vl * _max , Vl * _min and Vl * (l *).

The overall average calculating section 3 individually adds each of the chromaticity values rg, yb and the luminance value l of all the pixels of the color image obtained by the color space converting section 2 and outputs the addition result to each of the pixels. By dividing by numbers, the overall average values rg _mean , yb _mean , and l _mean are calculated (ST2), and the overall average values rg _mean , y
b _mean and l _mean are given to the coefficient calculator 6.

The neighborhood image extraction unit 4 receives the conversion result from the color space conversion unit 2, extracts a given or appropriately determined number of neighborhood regions, and sets it as a neighborhood image (S
T3).

The simplest example of the neighborhood is the pixel 8 of interest.
In the neighborhood, as shown in FIG.
g, and eight chromaticity values in the vicinity of the pixel of interest are represented by one of rg1 to rg8.
FIG. 4 is a schematic diagram showing the relationship between the entire image, the pixel of interest, and its neighboring pixels. Note that the chromaticity value yb and the luminance value l
, The chromaticity values yb1 to yb respectively as in FIG.
8, which are represented as luminance values 11 to 18 and have the same relationship as FIG. Hereinafter, an example in which the number of neighboring pixels is eight will be described.

The neighborhood average calculation section 5 receives the neighborhood image from the neighborhood image extraction section 4 and averages the pixel values of the neighborhood image as shown in the following equations (4) to (6) to obtain a neighborhood average value rg
_ave , yb _ave and l _ave are calculated (ST4).

[0067]

(Equation 2)

[0068] coefficient calculation section 6, the overall average calculator 3 average than overall value rg _mean, yb _mean, l receives _{me an,,} near the average value from the neighboring average calculating unit 5 rg _ave, yb _ave, l
_ave is received, and the ratio is calculated as shown in the following equations (7) to (12) (ST5), and the positive pixel value correction coefficients αrg _posi , αyb _posi , αl _posi and the negative pixel value correction coefficient are calculated. αrg _nega , αyb _nega and αl _nega .

[0069]

(Equation 3)

The positive image pixel value variable determining unit 7 determines
The neighborhood average value rg calculated by the neighborhood average calculation unit 5_ave,
yb_ave, L_aveAnd the coefficient calculated by the coefficient calculator 6
Number αrg_posi, Αyb_posi, Αl_posi And based on the following
As shown in Expressions (13) to (15), the pixels of the converted image
Value lower limit rg_L, Yb_L, L_LIs calculated by calculating the difference between
Raw value correction coefficient αrg_posi, Αyb_posi, Αl_posi Multiplied by
A proper constant βrg_posi, Βyb _posi, Βl_posi And add
Live image pixel value variable δrg_posi, Δyb_posi, Δl_{pos i} Toss
(ST6).

Δrg _posi = αrg _posi (rg _ave− rg _L ) + βrg _posi (13) δyb _posi = αyb _posi (yb _ave− yb _L ) + βyb _posi (14) δl _posi = αl _posi (l _ave− l _L) ) + Βl _posi (15) The negative image pixel value variable determination unit 8 calculates the neighborhood average values rg _ave , yb _ave , l calculated by the neighborhood average calculation unit 5.
_ave and the coefficient αrg _nega calculated by the coefficient calculation unit 6,
Based on αyb _nega and αl _nega , the following equation (16)
As shown in equation (18), the pixel value upper limit rg of the converted image
_U, yb _U, pixel values obtained by calculating the difference between the l _U correction coefficient αrg _nega, αyb _nega, appropriate constant multiplied by the .alpha.l _nega beta
rg _nega, βyb _nega, βl _nega added, negative image pixel value variable [Delta] rg _nega,? YB _nega, and δl _{n ega} (ST
7).

[0072] _{δrg nega = αrg nega (rg U} -rg ave) + βrg nega ... (16) δyb nega = αyb nega (yb U -yb ave) + βyb nega ... (17) δl nega = αl nega (l U -l ave ) + Βl _nega (18) The pixel value correction section 9 is a positive image pixel value variable determination section 7
More positive image pixel value variables δrg _posi , δyb _posi , δ
l _Posi received, negative image pixel value variable determination unit 8
More negative image pixel value variables δrg _nega , δyb _nega , δ
receive l _nega, as shown in the following (19) to (21), the chromaticity value rg of the pixel of interest, yb and luminance value pixel values lower from l values rg _L, yb _L, minus l _L The values are positive image pixel value variables δrg _posi , δyb _posi , δl
_Posi is added to obtain a positive image pixel value correction value, and a negative image pixel variable δrg is obtained by subtracting the chromaticity values rg, yb and the luminance value l of the pixel of interest from the pixel value upper limit values rg _U , yb _U , and L _{U. nega} , δyb _nega , δl _nega are added to obtain a negative image pixel value correction value, a positive image pixel value correction value is divided by a negative image pixel value correction value to obtain a ratio, and a logarithm (natural logarithm) of the ratio is calculated. Then, pixel correction values krg (rg), kyb (yb), and kl (l) are set (ST8).

[0073]

(Equation 4)

The pixel value correction control unit 10 controls the pixel correction values krg (rg), kyb (yb), kyb (yb) for all the pixels of the input image.
The pixel value correction unit 9 and the like are controlled so as to calculate kl (l).

The pixel correction value storage unit 11 stores the pixel correction values krg (r) for all the pixels from the pixel value correction control unit 10 for all the pixels.
g), kyb (yb), and kl (l) are received and stored in a memory or the like.

The pixel correction value maximum / minimum value calculation unit 12 stores the pixel correction values krg (r
g), kyb (yb), and kl (l) are read, and the next (2)
2) As shown in equations (27), their maximum value krg
_max , kyb _max , kl _max , minimum value krg _min , ky
Calculate b _min and kl _min .

Krg _max = max (krg (rg ₀ ), krg (rg ₁ ),..., Krg (rg _n )) (22) kyb _max = max (kyb (yb ₀ ), kyb (yb ₁ ),. , Kyb (yb _n )) (23) kl _max = max (kl (l ₀ ), kl (l ₁ ),..., Kl (l _n )) (24) krg _min = min (krg (rg ₀ )) , Krg (rg ₁ ),..., Krg (rg _n )) (25) kyb _min = min (kyb (yb ₀ ), kyb (yb ₁ ),..., Kyb (yb _n )) (26) kl _min = Min (kl (l ₀ ), kl (l ₁ ),..., Kl (l _n )) (27) where n is the total number of pixels.

The pixel value normalizing section 13 reads the pixel correction value from the pixel correction value storage section 11 and outputs the maximum value V of the output range.
rg _max , Vyb _max , Vl _max , minimum values Vrg _min , Vyb
_min, relative Vl _min, pixel correction value krg (rg), k
yb (yb) and kl (l) are normalized and transformed as shown in the following equations (28) to (30) (ST9).

[0079]

(Equation 5)

The pixel value conversion control unit 13 calculates pixel value conversion values Vrg (rg), Vyb (yb), and Vl (l) for all pixels of the input image by the pixel value normalization unit 13.

The pixel value normalized conversion value storage unit 15 receives and stores the pixel value conversion values Vrg (rg), Vyb (yb), Vl (l) of all pixels from the pixel value normalization control unit 14 for all pixels. I do.

In the case of chromaticity adjustment, the image output unit 16 stores the pixel value conversion value V of each pixel from the pixel value normalization conversion value storage unit 15.
rg (rg), Vyb (yb), and Vl (l) are received, and the whole image is constructed and output.

In the case of brightness adjustment, the image output unit 16
Is obtained by converting the normalized conversion value Vl * (l *) of each pixel stored in the pixel value normalized conversion value storage unit 15 and the chromaticity values a * and b * obtained by the color space conversion unit 2 into uniform colors. Space CIE-L
Inverse conversion is performed from AB to rgb color space, and the entire image is constructed and output based on the obtained inverse conversion result.

(Evaluation) Next, as to how the image quality of a color image is improved by such an image quality improvement method, luminance adjustment (method 1 in FIG. 5 and FIG. 6) and chromaticity adjustment (FIG. This will be described in the order of the method 2) in FIG.

FIG. 5A is a diagram showing an original image of a building having a standing image on the front surface before image quality improvement. Because the original image is lit in a building during twilight,
The dark part is completely dark, and the light part looks very bright.

On the other hand, the image after the image quality improvement according to the present embodiment is different from the original image due to the adjustment of the luminance value l *, as shown in FIG. And the fine irregularities of the standing image can be clearly seen, and the visibility of the dark part is improved.

FIG. 5 (c) is a view showing an original image before image quality improvement in which a tower having a fountain square in the front is imaged. This original image is imaged during fine daylight, with the central tower and surrounding green parts being dark, and the water fountain splashing light to reflect light.

On the other hand, the image after the image quality improvement according to the present embodiment differs from the original image by adjusting the luminance value l *, as shown in FIG. It becomes clear, and even the fine lines of the fountain can be clearly seen. That is, by the luminance adjustment of the present embodiment, the image quality is improved in FIG. 5D into a bright image in which details can be clearly seen while maintaining a natural feeling.

Next, instead of the example of the cool color system shown in FIG.
An example of a warm color system will be described with reference to FIG.

FIG. 6A is a diagram showing an original image of a fundus photograph, which is a kind of medical image, before image quality improvement. In the original image, although thick blood vessels are clearly visible, there are portions where the small blood vessels are slightly difficult to visually recognize.

On the other hand, the image after the image quality improvement according to the present embodiment is different from the original image by adjusting the luminance value l *, as shown in FIG. The bright part in the center is clearer.

FIG. 6 (c) is a diagram showing an original image before image quality improvement in which a scene of a town having two persons in front is captured. In the original image, the subject (person) on the front surface is dark and hard to see as compared with the highlight portion behind it.

On the other hand, the image after the image quality improvement according to the present embodiment is different from the original image by the adjustment of the luminance value l *, as shown in FIG. Can be recognized, the visibility of the dark area is improved, and the structure behind the bright area is improved so as to be clearer.

Next, instead of adjusting the luminance value l * using the uniform color space CIE-LAB, the chromaticity values rg, yb and the luminance value l * are adjusted.
The effect when the chromaticity is adjusted will be described. While the brightness adjustment of FIGS. 5 and 6 described above improved the image quality without changing the tint as much as possible, the adjustment of the chromaticity improved the image quality by positively changing the tint. I have. For this reason, the chromaticity adjustment method is simply difficult to compare with the luminance adjustment method, and is compared with the well-known processing result by histogram flattening, which is most often used as a color enhancement method and is said to be powerful. explain.

FIG. 7A is a diagram showing an original image of a large Buddha image taken outside before image quality improvement. The original image has a low-resolution image with a hazy and blurred feel as a whole.

FIG. 7 (b) shows an image after the image quality has been improved by the well-known histogram flattening process. Each color becomes excessively clear and strong, and a false image is formed around the Great Buddha image or in the sky. The contour is clearly visible. This false contour is a phenomenon that occurs due to excessive emphasis that completely deviates from the atmosphere of the original image.

On the other hand, as shown in FIG. 7 (c), the image after the image quality improvement according to the present embodiment is different from the well-known histogram emphasizing method by adjusting the chromaticity values rg and yb to change the atmosphere of the original image. Each color can be surely emphasized while maintaining the same, and no false contour is generated anywhere.

Next, FIG. 8A is a diagram showing an original image before the image quality improvement, in which the road conditions at night in a city are imaged. This original image has a red cast due to the influence of street lights and car tail lamps.

FIG. 8B shows an image after the image quality has been improved by the well-known histogram flattening process. The reddishness of the red cast is further enhanced, and the entire image is excessively enhanced. Let me.

On the other hand, in the image after the image quality improvement according to the present embodiment, as shown in FIG. 8C, the reddishness of the red cast is removed, and the visibility is clearly improved.

The above two methods of luminance adjustment and chromaticity adjustment have different characteristics, and both can naturally improve the image quality while maintaining the atmosphere of the original image. Such effects are difficult to achieve with conventional purely engineering methods, but the method of the present invention incorporates human physiological visual mechanisms to enhance colors in a natural way for humans. ,Obtainable. In both methods, local and global contrast enhancement is performed in accordance with human color vision, and substantial utilization of image data is improved, and a high-definition color image is realized.

As described above, according to the first embodiment, the chromaticity (rg, yb and luminance value l) adjustment processing and the luminance (l *) adjustment processing are performed on the chromaticity of the input image, respectively. Without assuming in advance the distribution characteristics of the values rg, yb and the luminance values l, l *, the overall average value r of the input image
By calculating g _mean , yb _mean , l _mean , and l * _mean , it is adapted to the distribution of local chromaticity values rg1 to rgi, yb1 to ybi and luminance values l1 to li, l * 1 to l * i. Since the degree of enhancing the contrast can be adjusted only by the processing in the vicinity of each pixel, the image quality of the color image can be improved by a simple calculation without requiring the user's judgment.

Further, by emphasizing local and global contrasts in accordance with human color vision, the degree of practical use of image data can be increased and a high-definition color image can be realized.

Specifically, a sharp image with improved contrast near each pixel can be obtained from the input original image. For example, it is possible to correct the luminance signal from the image sensor with high precision and high resolution, and to improve the image quality of image data with low contrast.

Further, by adjusting the chromaticity according to the chromaticity values rg and yb, it is possible to realize a natural color tone suitable for human color perception.

Further, in this embodiment, since the color space for converting the pixel value is changed between the luminance adjustment and the chromaticity adjustment, it is possible to achieve more optimal improvement of the image quality according to the contents of the adjustment. For example, in the brightness adjustment, the pixel value rgb is
Uniform color space CI that is precise and highly independent of chromaticity and luminance
The conversion into E-LAB, where the inverse S-shaped method is applied to the luminance axis, can significantly improve the image quality with respect to the luminance of the color image.

In the chromaticity adjustment, the pixel value rgb
Is converted to a color space based on the human color vision mechanism, and the inverse S-shape method is applied to the (rg) axis and the (yb) axis. Can be improved.

(Second Embodiment) FIG. 9 shows a second embodiment of the present invention.
2 is a block diagram illustrating a configuration of a color image quality improving apparatus according to an embodiment of the present invention. The same parts as those in FIG. 1 are denoted by the same reference numerals, detailed description thereof will be omitted, and only different parts will be described.

That is, in the present embodiment, in addition to the processing of the first embodiment, the average luminance is associated with the luminance center of the display system, and the average chromaticity is associated with the chromaticity center of the display system. The normalization is intended to improve the contrast of the entire image. Specifically, as shown in FIG. 9, instead of the pixel value normalizing unit 13 to the image output unit 16, a whole average correction value calculation unit 20, a pixel correction value conversion unit 21, a pixel correction value conversion control unit 22, A correction value conversion value storage unit 23 and a normalized image output unit 24 are provided.

As described above, the luminance value l * of the luminance adjustment is used.
Is the same as the calculation process of the luminance value 1 in the chromaticity adjustment
Therefore, the process from the overall average calculation unit 3 to the normalized image output unit 24
In the middle, it is included in the calculation process of the luminance value 1 in the chromaticity adjustment.
I will tell you. In other words, the operation of the brightness adjustment
In addition to the equations, the equations (33), (38) and (3)
9) Luminance value l in equation and the like and value based on it
l_mean, Αl_posi , Αl_nega , L_L, L_U, Δl_posi ,
δl_nega , Kl (l), kl_max , Kl_min , Vl_max,
Vl_min And Vl (l) are calculated based on the luminance value l * and the luminance value l *.
Value l *_mean , Αl *_{po si}, Αl *_nega, Δl *_posi, Δl *
_nega, Kl * (l *), kl *_max, Kl *_min, Vl * _max,
Vl *_minAnd Vl * (l *).

Here, the overall average correction value calculating section 20 calculates the overall average values rg _mean ,
yb _mean , l _mean and coefficients αrg _posi , αyb _posi , αl _posi , αrg _nega , αy calculated by the coefficient calculator 6
On the basis of b _nega and αl _nega , the overall average value RG _mean , y
b _mean , l _mean correction values krg (rg _mean ), kyb (yb
_mean ), kl (l _mean ) are calculated, and the correction value krg (r
g _mean ), kyb (yb _mean ), and kl (l _mean ) are provided to the pixel correction value converter 21.

The pixel correction value conversion unit 21
Pixel correction value krg (rg), k of each pixel calculated by
yb (yb), kl (l) and the correction values krg (rg _mean ), kyb (yb) calculated by the overall average correction value calculator 20
_It has a conversion function of comparing magnitude relations with _mean ) and kl (l _mean ) and performing a conversion according to the comparison result.

This conversion function takes the chromaticity value rg as an example, and when krg (rg) <krg (rg _mean ), the pixel correction value krg (rg) is set to the intermediate value Vrg _mid in the output range.
Linearly converted to a smaller output range section, krg (rg)
= Krg (rg _mean ), the pixel correction value krg (rg)
Is linearly converted to an intermediate value Vrg _mid within the output range, and k
pixel correction value k at the time of rg (rg)> krg (rg mean)
rg (rg) is linearly converted to an output range section larger than the intermediate value Vrg _mid in the output range range (kyb (y
b _mean ) and kl (l _mean ) are similarly _subjected to comparison and linear conversion).

The pixel value correction value conversion unit 21 uses this conversion function to convert each pixel correction value krg (rg), kyb (yb), k
l (l) is normalized within the range of the output range to calculate normalized conversion values Vrg (rg), Vyb (yb), Vl (l), and the normalized conversion values Vrg (rg), Vyb ( y
b) and Vl (l) to the pixel correction value conversion control unit 22.

The pixel correction value conversion control unit 22 receives the normalized conversion values Vrg (rg), Vyb received from the pixel correction value conversion unit 21.
(Yb), Vl (l) to the pixel correction value conversion value storage unit 23, and the normalized conversion value Vrg for all pixels
(Rg), Vyb (yb), and Vl (l).

The pixel correction value conversion value storage unit 23 stores each normalized conversion value Vrg (r) received from the pixel correction value conversion control unit 22.
g), Vyb (yb), and Vl (l) are stored in a memory or the like, and can be read from the normalized image output unit 24.

In the case of chromaticity adjustment, the normalized image output unit 24 converts the normalized conversion values Vrg (rg), Vyb (yb), Vl of each pixel stored in the pixel correction value conversion value storage unit 23.
It has the function of composing and outputting the entire image based on (l).

In the case of luminance adjustment, the normalized image output unit 24 outputs the normalized conversion value Vl * (l *) of each pixel stored in the pixel correction value conversion value storage unit 23 and the color space conversion unit 2. Are converted into uniform color space CIE-
It has a function of performing an inverse conversion from the LAB to the rgb color space and constructing and outputting the entire image based on the obtained inverse conversion result.

Next, a color image quality improving method by the color image quality improving apparatus thus configured will be described with reference to FIG.
0 will be described.

It is now assumed that the processes from the image input unit 1 to the pixel correction value maximum / minimum value calculation unit 12 have been completed (ST1 to ST8).

The overall average correction value calculating section 20 calculates the overall average value rg _mean , y calculated by the overall average calculating section 3.
Based on b _mean , l _mean, etc.
Using equation (21), the overall average value rg _mean , yb _mean , l
_mean correction values krg (rg _mean ), kyb (yb _mean ), k
Calculate l (l _mean ) and give it to the pixel correction value converter 21.

The pixel correction value converter 21 calculates the correction value krg
(Rg _mean ), kyb (yb _mean ), and kl (l _mean ), and the pixel correction value krg (r
g), kyb (yb), and kl (l) are read. Further, as shown in the following equations (31) to (33), the maximum values Vrg _max , Vyb _max , Vl _max and the minimum value Vr of the output range are obtained.
g _{mi n,} Vyb _min, relative Vl _min, intermediate value Vrg _mid respectively, Vyb _mid, calculates the Vl _mid (ST20).

[0123]

(Equation 6)

Next, the overall average values rg _mean , yb _mean ,
l _mean intermediate values Vrg _mid, Vyb _mid, so as to correspond to Vl _mid, pixel correction value krg (rg), kyb (yb ),
kl (l) is linearly transformed as in the following equations (34) to (39) (ST21).

[0125]

(Equation 7)

[0126]

(Equation 8)

The pixel correction value conversion control unit 22 calculates pixel correction value conversion values Vrg (rg), Vyb for all pixels of the input image.
The pixel correction value converter 21 is controlled so as to calculate (yb) and Vl (l).

The pixel correction value conversion value storage unit 23 receives the pixel correction value conversion values Vrg (rg), Vyb (yb), Vl (l) of all pixels from the pixel correction value conversion control unit 22 for all pixels, and And so on.

In the case of chromaticity adjustment, the normalized image output unit 24 reads the pixel correction value conversion values Vrg (rg), Vyb (yb), Vl (l) of each pixel from the pixel correction value conversion value storage unit 23. , And construct and output the entire image.

In the case of the luminance adjustment, the normalized image output unit 24 outputs the normalized conversion value Vl * (l *) of each pixel stored in the pixel correction value conversion value storage unit 23 and the color space conversion unit 2. Are converted into uniform color space CIE-
Inverse conversion from LAB to rgb color space is performed, and the entire image is constructed and output based on the obtained inverse conversion result.

As described above, according to the second embodiment,
For example, in addition to the effects of the first embodiment, chromaticity adjustment processing and
Regarding the brightness adjustment processing, respectively, the overall average value rg_mean,
yb_{me an}, L_mean, L *_mean Correction value krg (r
g_mean), Kyb (yb_mean), Kl (l _mean), Kl *
(L *_mean ) Is the intermediate value Vr with good sensitivity of the output range.
g_mid, Vyb_mid, Vl_mid , Vl *_midCorresponding to regular
Local contrast and global contrast
Can be emphasized at the same time.

More specifically, from the input original image, the contrast in the vicinity of each pixel is improved, and the contrast of the entire image is also improved, so that a sharp and highly visible image can be obtained. . For example, it is possible to correct the luminance signal from the image sensor with high precision and high resolution, and to improve the image quality of image data with low contrast.

(Other Embodiments) In the first and second embodiments, the size of the neighboring area may be fixed to an appropriate value, or may be appropriately given as a parameter. Regarding the size of the neighboring region, for example, in a substantially concentric region, it is preferable that the radius is 3 to 7 pixels, and the area is 20 to 160 pixels. Note that the actual tendency is that local contrast can be enhanced by taking the neighboring area finely, such as up to an area of about 100 pixels, and the global area can be taken up by taking the neighboring area large, such as the area of more than 100 pixels. Contrast can be emphasized.

In the first and second embodiments, the values of βrg _nega , βyb _nega , and βl _nega may be fixed to appropriate values or may be appropriately provided as parameters.

In the first and second embodiments, the case where only the luminance value l * is adjusted and the case where both the luminance value l and the chromaticity values rg, yb are adjusted have been described.
However, the present invention is not limited to this. Even if only the chromaticity values rg and yb are adjusted, the image quality of the color vision can be improved.

The method described in the above embodiment can be executed by a computer as a program such as a magnetic disk (floppy disk, hard disk, etc.), an optical disk (CD-ROM, DVD, etc.),
It can also be stored in a storage medium such as a semiconductor memory and distributed.

In addition, the present invention can be variously modified and implemented without departing from the gist thereof.

[0138]

As described above, according to the present invention,
It is possible to provide a color image quality improvement method and apparatus capable of enhancing both local and global contrasts and improving the quality of a color image by a simple calculation without requiring a user's judgment.

A method for improving the quality of a color image capable of realizing a high-definition color image by increasing the substantial utilization of image data by enhancing local and global contrasts in accordance with human color vision. Equipment can be provided.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a color image quality improving apparatus according to a first embodiment of the present invention.

FIG. 2 is a flowchart for explaining an operation in the embodiment;

FIG. 3 is a schematic diagram for explaining a pixel of interest and pixels in the vicinity thereof according to the embodiment;

FIG. 4 is a schematic diagram showing the relationship between the entire image, a pixel of interest, and its neighboring pixels according to the embodiment;

FIG. 5 is a diagram showing an image for explaining an effect in the embodiment by photo printing.

FIG. 6 is a view showing an image for explaining an effect in the embodiment by photo printing.

FIG. 7 is a view showing an image for explaining an effect in the embodiment by photo printing;

FIG. 8 is a diagram showing an image for explaining an effect in the embodiment by photo printing.

FIG. 9 is a block diagram showing a configuration of a color image quality improvement device according to a second embodiment of the present invention.

FIG. 10 is a flowchart for explaining the operation in the embodiment;

FIG. 11 is a schematic view showing a human color vision mechanism applied to the present invention.

FIG. 12 is a diagram for explaining a conventional color image quality improvement method by logarithmic conversion or exponential conversion.

FIG. 13 is a view for explaining a conventional color image quality improvement method by exponential conversion.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... Image input part 2 ... Color space conversion part 3 ... Whole average calculation part 4 ... Neighborhood image extraction part 5 ... Neighborhood average calculation part 6 ... Coefficient calculation part 7 ... Positive image pixel value variation determination part 8 ... Negative image pixel value variation Determining unit 9 ... Pixel value correction unit 10 ... Pixel value correction control unit 11 ... Pixel correction value storage unit 12 ... Pixel correction value maximum / minimum value calculation unit 13 ... Pixel value normalization unit 14 ... Pixel value normalization control unit 15 ... Pixel Value normalized conversion value storage unit 16 ... Image output unit 20 ... Overall average correction value calculation means 21 ... Pixel correction value conversion unit 22 ... Pixel correction value conversion control unit 23 ... Pixel correction value conversion value storage unit 24 ... Normalized image output Department

[Procedure amendment]

[Submission date] January 31, 2000 (200.1.3
1)

[Procedure amendment 1]

[Document name to be amended] Statement

[Correction target item name] Claim 1

[Correction method] Change

[Correction contents]

[Procedure amendment 2]

[Document name to be amended] Statement

[Correction target item name] Claim 3

[Correction method] Change

[Correction contents]

[Procedure amendment 3]

[Document name to be amended] Statement

[Correction target item name] Claim 5

[Correction method] Change

[Correction contents]

[Procedure amendment 4]

[Document name to be amended] Statement

[Correction target item name] Claim 7

[Correction method] Change

[Correction contents]

[Procedure amendment 5]

[Document name to be amended] Statement

[Correction target item name] 0026

[Correction method] Change

[Correction contents]

Now, based on the gist of the present invention described above,
Specifically, the following solution is realized. The invention corresponding to claim 1 converts pixel values r, g, b in each pixel of a color image into chromaticity values a *, b * and a luminance value l * on a uniform color space CIE-LAB, and The average value l * _mean of the luminance values l * of the pixels is determined, and for each of the pixels, the average value l * _ave of the luminance values l * 1 to l * i of the i pixels in the vicinity of the pixel is determined, Before the overall _mean l * _mean
Positive pixel value correction coefficient .alpha.l * _posi corresponding serial luminance value l * minus the lower limit l * _L between average value l * _ave of the vicinity to the ratio of the value obtained by subtracting the lower limit l * _L When asked
From the upper limit l * _U of the luminance value l *, the overall average value
Rights l * _mean a value obtained by subtracting from the upper limit l * _U of the neighboring
Negative image corresponding to the ratio to the value obtained by subtracting the average l * _ave
Find the prime value correction coefficient αl * _nega, and calculate the positive pixel value
Wherein the correction coefficient .alpha.l * _posi average value l * _ave of the neighboring
The positive image pixel value variation δl * _posi is obtained by multiplying the value obtained by subtracting the lower limit l * _L, and the negative pixel value correction coefficient αl * _nega is calculated from the upper limit l * _U. The negative image pixel value variation δl * _nega is obtained by multiplying a value obtained by subtracting the average value l * _ave of the neighborhood, and the positive image pixel value variation δl * is obtained.
subtracting the lower limit l * _L a _posi from the luminance value l * for those pixel
In addition to correcting this relative value in addition to the relative value, the negative image pixel value variable δl * _nega is added to the relative value obtained by subtracting the luminance value l * of the pixel from the upper limit l * _U. The relative value is corrected, the luminance value l * of the pixel is corrected by the logarithm of the ratio of the corrected relative value, and the correction result kl * (l
*) Within the range of the output range, and the normalized result Vl * (l *) and each of the chromaticity values a *, b * are inversely converted from the uniform color space CIE-LAB to the rgb color space. This is a color image quality improvement method for constructing and outputting the entire image based on the obtained inverse transformation result.

[Procedure amendment 6]

[Document name to be amended] Statement

[Correction target item name] 0028

[Correction method] Change

[Correction contents]

Further, according to the invention corresponding to claim 3, the color
Image input means for inputting an image, and the image input means
Pixel values r of all pixels of the color image input by the means,
g and b are each chromaticity value a *, in the uniform color space CIE-LAB.
color space conversion means for converting the color into b * and a luminance value l *;
Average each luminance value l * of all pixels obtained by the space conversion means
And the overall average l *_meanAverage calculation means for calculating
And for each pixel obtained by the color space conversion means,
Each luminance value l * 1 to l * i of i pixels in the vicinity of the pixel
Image extracting means for extracting the image, and the adjacent image extracting means
The luminance values l * 1 to l * i of the neighboring pixels extracted by
Averaged neighborhood average l *_aveMeans for calculating the neighborhood average
And the overall average value calculated by the overall average calculation means
l *_mean From the lower limit l * of the luminance value l * _L MinusAnd before
The neighborhood average value l * calculated by the neighborhood average calculation means
_ave From the lower limit l * _L MinusCorresponding to the ratioPo
Exclusive pixel value correctionCoefficient αl *_posi , And
Upper limit l * of the brightness value l * _U Means for calculating the overall average from
Average value l * calculated by _mean Minus the value
Upper limit l * _U From the neighborhood average calculation means.
Neighborhood average l * _ave Negation corresponding to the ratio to the value minus
Active pixel value correction coefficientαl *_negaCoefficient calculating means for calculating
And the neighborhood of the pixel obtained by the neighborhood average calculating means.
Average l *_ave From the lower limit l * _L Subtracted from the previous
RecordCalculated by coefficient calculation meansPositive pixel value correction
Coefficient αl *_posi Multiply byThe pixel value variable δ of the positive image
l *_posiPositive image pixel value variable determining means for determining
And the neighborhood of the pixel obtained by the neighborhood average calculating means.
Average l *_ave Is the upper limit l * _U From the value subtracted from
SaidCalculated by coefficient calculation meansNegative pixel value complement
CorrectCoefficient αl *_nega Multiply byThe pixel value variate of the negative image
δl *_negaNegative image pixel value variable determining means for determining
Determined by the positive image pixel value variable determining means.
Was donePositive imagePixel value variable δl *_posi Of the pixel
From the luminance value l * to the lower limit value l _{L *} Minus the relative value
While correcting this relative value,The negative image pixel
Determined by value variable determination meansNegative imagePixel value
Variate δl *_nega Is the upper limit l * _U From the luminance value l of the pixel
Modify this relative value in addition to the relative value minus *
Corrected relative valueThe logarithm of the ratio of the pixel correction values kl *
*) Pixel value correction means for calculating the pixel value, and the pixel value correction means
Of all pixel correction values kl * (l *) calculated by
And the maximum value kl *_maxAnd the minimum value kl *_minAnd correction to calculate
Value maximum / minimum value calculating means, and the correction value maximum / minimum value calculating means.
Maximum value kl * calculated by step_maxAnd the minimum value kl * _min
Pixel calculated by the pixel value correction means based on
Normalize the correction value kl * (l *) within the output range
Pixel value normalizing means for calculating a normalized conversion value Vl * (l *)
And the positive value of each pixel calculated by the pixel value normalizing means.
Normalized conversion value Vl * (l *) and obtained by the color space conversion means.
The obtained chromaticity values a * and b * are converted into the uniform color space CIE-LA.
B is converted back to rgb color space, and the obtained inverse conversion result is
Image output means for configuring and outputting the entire image based on the
Is a color image quality improvement device including

[Procedure amendment 7]

[Document name to be amended] Statement

[Correction target item name] 0030

[Correction method] Change

[Correction contents]

Further, according to the present invention, the pixel values r, g, b of each pixel of the color image are converted into respective chromaticity values rg, yb and a luminance value 1 in a color space, and the respective chromaticity values are converted. rg, yb and the average value r of each of the luminance values l
g _mean , yb _mean , and l _mean are determined, and for each pixel of the color image, chromaticity values rg1 to rgi, yb1 to ybi, and luminance values l1 to i of i pixels in the vicinity of the pixel are determined.
Each of the average value rg _ave of li, yb _ave, seeking l _ave, the overall average value rg _mean, yb _mean, from l _mean
Each lower limit rg of each chromaticity value rg, yb and luminance value l
_L, yb _L, the average value r of the near and minus l _L
g _ave, yb _ave, said from l _ave the lower limit rg _L, y
Positive pixel value corresponding to the ratio to the value obtained by subtracting b _L and l _L
When the correction coefficients αrg _posi , αyb _posi , and αl _posi are obtained
In each case, the upper limit of each of the chromaticity values rg, yb and the luminance value l
From the values rg _U , yb _U , l _U , the overall mean value rg _mean ,
yb _mean , l _mean minus the upper limit rg _U , y
The average value rg _ave , yb _ave , l of the neighborhood from b _U , l _U
Negative pixel value correction corresponding to the ratio with the value after subtracting _ave
Coefficient αrg _nega, αyb _nega, seeking .alpha.l _nega, said port
Additive pixel value correction coefficients αrg _posi , αyb _posi , αl
Mean values of the neighborhood of the _posi rg _ave, yb _ave, or l _ave
The positive image pixel value variables δrg _posi , δyb _posi , and δ are multiplied by the relative values obtained by subtracting the lower limit values rg _L , yb _L , and l _L from the above.
l _Posi and negative pixel value correction coefficient
αrg _nega , αyb _nega , αl _nega are set to the respective upper limit values rg
_U, yb _U, the mean value rg _ave of the neighborhood from l _U, yb
_ave, negative image pixel value by multiplying the value obtained by subtracting the l _ave <br/> variable δrg _nega, δyb _nega, seeking .DELTA.l _nega, the positive image pixel value variable rg _posi, δyb _posi, the .DELTA.l _posi of those pixel each chromaticity values rg, each bottom yb and luminance value l
Limit value rg _L, yb _L, with in addition to the relative value of each minus l _L fix these relative values, the negative image image <br/> element value variable αrg _nega, αyb _nega, the .alpha.l _nega each Up
Limit value rg _U, yb _U, the chromaticity value rg of the pixel from the l _U,
These relative values are corrected in addition to the relative values obtained by subtracting yb and the luminance value l, and the chromaticity values rg, yb and the luminance value l of the pixel are respectively corrected by the logarithm of the ratio of the corrected relative values, These correction results krg (rg), kyb (y
This is a color image quality improvement method in which b) and kl (l) are respectively normalized within the output range.

[Procedure amendment 8]

[Document name to be amended] Statement

[Correction target item name] 0032

[Correction method] Change

[Correction contents]

Further, according to the invention corresponding to claim 7, the color
Image input means for inputting an image, and the image input means
Pixel values r of all pixels of the color image input by the means,
g and b are converted to chromaticity values rg and yb and a luminance value l in the color space.
Color space conversion means for converting, and the color space conversion means
Chromaticity values rg, yb and luminosity of all pixels of the obtained color image
Average each of the degree values l to obtain the overall average value rg_mean, Y
b_mean, L_meanMeans for calculating the average, and the color
For each pixel of the color image obtained by the space conversion means,
Each chromaticity value rg, yb and luminance value l of the pixel and its vicinity
Chromaticity values rg1 to rgi, yb of i pixels in the vicinity
Neighboring images for extracting 1-ybi and luminance values 11-li
Extraction means, and the neighborhood extracted by the neighborhood image extraction means.
Chromaticity values rg1 to rgi, yb1 to ybi of neighboring pixels
Average the respective luminance values l1 to li to obtain a neighborhood average value r
g_ave, Yb_ave, L_aveMeans for calculating the neighborhood average
And the overall average value calculated by the overall average calculation means
rg_mean, Yb_mean, L_mean From each of the chromaticity values rg, y
b and each lower limit rg of the luminance value l _L , Yb _L , L _L Pull
ValueAnd the neighborhood average calculated by the neighborhood average calculation means
Value rg_ave, Yb_ave, L_ave From the lower limit r
g _L , Yb _L , L _L MinusCorresponding to the ratioPositive
Pixel value correctioncoefficientαrg _posi , Αyb _posi , Αl _posi To
And the chromaticity values rg, yb and the luminance value l
Each upper limit rg _U , Yb _U , L _U From the total average calculator
Overall average value rg calculated by step _mean , Yb _mean , L
_mean Minus the above upper limit rg _U , Yb _U , L _U From
The neighborhood average value rg calculated by the neighborhood average calculation means
_ave , Yb _ave , L _ave Net corresponding to the ratio
Gative pixel value correction coefficient αrg _nega , Αyb _nega , Αl
_nega Coefficient calculating means for calculating
The neighborhood average value of the pixel obtained inrg _ave , Y
b _ave , L _ave From each of the lower limit values rg _L , Yb _L , L _L To
For the value subtracted,Calculated by coefficient calculation meansPo
Exclusive pixel value correctioncoefficientαrg _posi , Αyb _posi , Αl
_posi Multiply byAnd the positive image pixel value variable δrg_posi, Δ
yb_posi, Δl_posiPositive image pixel value variate to determine
Determining means and the image obtained by the neighborhood average calculating means.
Elementary neighborhood meanrg _ave , Yb _ave , L _ave The above each
Limit value rg _U , Yb _U , L _U From the value subtracted fromcoefficient
Calculated by the calculation meansNegative pixel value correctioncoefficientα
rg _nega , Αyb _nega , Αl _nega Multiply byAnd negative
Image pixel value variation δrg_{ne ga}, Δyb_nega, Δl_negaDetermine
Negative image pixel value variable determining means;
Image pixel value variable determined bypositive
imagePixel value variation δrg_posi, Δyb_posi, Δl_posi The picture
From the respective chromaticity values rg and yb of the element and the luminance value l, the above lower limits are obtained.
Value rg _L , Yb _L , L _L This relative to the relative value minus
Modify the values,The negative image pixel value variable determination
Determined by meansNegative imagePixel value variation δrg
_nega, Δyb_nega, Δl_nega Is the upper limit rg _U , Y
b _U , L _U From each chromaticity value rg, yb and luminance of the pixel
Correct these relative values in addition to the relative value minus the value l, and
These modified relative valuesThe logarithm of the ratio of
Pixel value for calculating (rg), kyb (yb), KL (l)
Correction means, and all of the pixels calculated by the pixel correction means.
Of the pixel correction values krg (rg), kyb (yb), and kl (l)
Of which the maximum value is krg_max, Kyb_max, Kl_maxAnd the minimum value kr
g_min, Kyb_min , Kl_minAnd the correction value to calculate the maximum and minimum
Value calculation means and the correction value maximum / minimum value calculation means.
Maximum value issued krg_max, Kyb_max, Kl_maxAnd minimum
Value krg_min , Kyb_min, Kl_minBased on the pixel value
Pixel correction value krg (rg), k calculated by the correction means
Normalize yb (yb) and kl (l) within the output range
And the normalized conversion values Vrg (rg), Vyb (yb), Vl
Pixel value normalizing means for calculating (l);
Conversion value Vrg (r) of each pixel calculated by the
g), Vyb (yb), Vl (l)
Image output means for constructing and outputting a color image
This is an image quality improvement device.

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Shunichi Kato 1-1-4 Umezono, Tsukuba, Ibaraki Pref. Within the Institute of Electronics and Technology (72) Inventor Yuichi Kobayashi 1-5-1, Taito 1-chome, Taito-ku, Tokyo No. Toppan Printing Co., Ltd. F term (reference) 5B057 AA11 BA02 BA11 BA28 CA01 CA08 CA12 CB01 CB08 CB12 CC02 CE17 CE18 CH08 5C066 AA03 BA20 CA08 DA09 DD07 EA08 GA01 GA21 GB02 HA02 JA03 KA12 KD04 KD06 KE01 KE19 KM15 LA NN03 PP15 PP31 PP32 PP36 PP41 PP46 PP68 PQ12 PQ20 TT02 TT06 TT09 TT10 5C079 HB01 HB08 HB11 LA10 LA12 LB01 NA03 PA02 PA03

Claims (8)

[Claims] 1. A pixel value r, at each pixel of a color image,
g and b are defined as chromaticity values a * and 上 の on the uniform color space CIE-LAB.
b * and a luminance value l *, and an average l * _mean of the luminance values l * of the respective pixels is obtained. For each of the pixels, the respective luminance values l * 1 of i pixels in the vicinity of the pixel are determined. The average value l * _ave of ｌl * i is obtained, and the average value l * _{mean of the} whole and the average value l * _{ave of the} neighborhood are obtained.
Is obtained by multiplying each coefficient by a relative value of the average value l * _ave of the neighborhood in the positive image to obtain a positive pixel variable δl * _posi
With seek, seeking negative pixel variables .DELTA.l * _{ne ga} multiplied by the coefficient of the each relative value of the average value l * _ave of the vicinity of the negative image, the of the positive pixel variable .DELTA.l * _posi a positive image This relative value is corrected in addition to the relative value of the luminance value l * of the pixel, and the relative value is added by adding the negative pixel variable δl * _{ne ga} to the relative value of the luminance value l * of the pixel in the negative image. The corrected luminance value l * is corrected by the logarithm of the ratio of the corrected relative values, the corrected result kl * (l *) is normalized within the output range, and the normalized result Vl * ( l *) and the respective chromaticity values a *, b
*, A color image quality improving method comprising: converting * from the uniform color space CIE-LAB to the rgb color space; and constructing and outputting the entire image based on the obtained inverse conversion result. 2. The color image quality improving method according to claim 1, wherein the normalization is a correction result k of the overall average value l * _mean.
A method for improving image quality of a color image, wherein l * (l * _mean ) is executed so as to correspond to an intermediate value Vl * _mid of the output range. 3. Image input means for inputting a color image, and pixel values r, g, and b of all pixels of the color image input by the image input means are converted to respective chromaticity values in a uniform color space CIE-LAB. a *, b * and a color space converting means for converting into a luminance value l *; and a luminance value l * of all pixels obtained by the color space converting means.
And an overall average value calculating means for calculating an overall average value l * _mean, and for each pixel obtained by the color space converting means, each luminance value l * 1 of i pixels in the vicinity of the pixel. Ｌl * i, a neighborhood image extracting means for extracting 輝 度 l * i, and a neighborhood average for calculating a neighborhood average value l * _ave by averaging the brightness values l * 1 to l * i of the neighboring pixels extracted by the neighborhood image extraction means. Calculating means, and the overall average value l * calculated by the overall average calculating means
coefficient corresponding to the ratio of the vicinity of the average value l * _ave calculated by _mean and the neighboring average calculation means αl * _posi, αl * _nega
And a coefficient αl * calculated by the coefficient calculating means, and a nearby average value l * _{ave of the} pixel obtained by the neighboring average calculating means.
Based on the _posi , the pixel value variance of the positive image δl * _posi
And a neighborhood average value l * _{ave of the} pixel obtained by the neighborhood average calculation means and a coefficient αl * calculated by the coefficient calculation means.
Based on _nega , the pixel value variable δl * _{nega of the} negative image
A negative image pixel value variable determining means for determining the pixel value variable δl * _posi determined by the positive image pixel value variable determining means and a pixel value variable δl * _{ne ga} determined by the negative image pixel value variable determining means. Pixel value correction means for correcting the luminance value l of the pixel based on the pixel value, calculating the logarithm of the ratio of the correction result to calculate a pixel correction value kl * (l *), and calculating the pixel correction value kl * (l *). Of all the pixel correction values kl * (l *), the maximum value kl * _max and the minimum value kl * _min
Correction value maximum / minimum value calculation means for calculating the maximum value kl * _max and minimum value kl * _min calculated by the correction value maximum / minimum value calculation means, and a pixel calculated by the pixel value correction means. The correction value kl * (l *) is normalized within the range of the output range, and the normalized conversion value Vl * (l *)
, A normalized conversion value Vl * (l *) of each pixel calculated by the pixel value normalizing means, and each chromaticity value a *, obtained by the color space converting means. b * is the uniform color space CIE-LAB
And an image output means for constructing and outputting the entire image based on the inverse conversion result obtained from the image data to the rgb color space. 4. The color image quality improving apparatus according to claim 3, wherein the overall average value l * calculated by said overall average calculating means instead of said pixel value normalizing means and said image output means.
_mean and coefficient αl * calculated by the coefficient calculating means
_posi, based on .alpha.l * _nega, the overall average correction value calculating means for calculating a correction value kl * (l * _mean) of the overall average value l * _{m ean,} of each pixel calculated by the pixel value correcting means The pixel correction value kl * (l *) is compared with the correction value kl * (l * _mean ) calculated by the overall average correction value calculating means, and the pixel correction value kl * (l *) is compared with the pixel correction value kl * (l *). Correction value kl * (l
* _mean ) is large, the pixel correction value kl * (l *) is linearly converted to an output range section smaller than the intermediate value Vl * _mid within the corresponding output range, and the pixel correction value kl * (l *) is obtained.
And the corresponding correction value kl * (l * _mean ) are equal, the pixel correction value kl * (l *) is converted to the intermediate value Vl * _mid within the corresponding output range, and the pixel correction value kl * ( l
When the corresponding correction value kl * (l * _mean ) is smaller than *), the pixel correction value kl * (l *) is linearly converted into an output range section larger than the intermediate value Vl * _mid in the corresponding output range. By doing so, each pixel correction value kl * (l *) is normalized within the range of the output range, and the normalized conversion value Vl * (l *)
), A normalized conversion value Vl * (l *) of each pixel calculated by the pixel correction value conversion means, and a chromaticity value a * obtained by the color space conversion means. , B * in the uniform color space CIE-LA
A color image quality improving apparatus, comprising: a normalized image output unit that performs an inverse conversion from B to an rgb color space and forms and outputs the entire image based on the obtained inverse conversion result. 5. A pixel value r, at each pixel of a color image,
g and b are chromaticity values rg and yb and a luminance value l in a color space.
And the average value of each of the chromaticity values rg, yb and the luminance value l
rg_mean, Yb_mean, L_meanIs calculated for each pixel of the color image.
Chromaticity values rg1 to rgi, yb1 to ybi of i pixels
And the average value rg of each of the luminance values 11 to li_{av e}, Yb
_ave, L_aveIs obtained, and the average value rg of the whole is obtained._mean, Yb_mean, L_meanAnd said near
Average value rg_ave, Yb_ave, L_aveCorresponding to the ratio
The respective coefficients are obtained, and the respective coefficients are average values of the neighborhood in a positive image.
rg_ave, Yb_ave, L_aveMultiplied by the relative value of
Extrinsic pixel variable δrg_posi, Δyb_posi, Δl_{pos i} Ask for
And each of the coefficients is
Average value rg _ave, Yb_ave, L_aveRelative value of each
Multiplied by the negative pixel variable δrg_nega, Δyb_nega, Δl
_nega And the positive pixel variable δrg_posi, Δyb_posi, Δl_posi 
Are the same as the respective chromaticity values rg, yb of the pixel in the positive image.
In addition to these relative values,
The negative pixel variable δrg_nega, Δyb
_nega, Δl_negaIs the chromaticity of the pixel in the negative image
In addition to the relative values of the values rg, yb and the luminance value l,
These relative values are corrected, and each of the pixels is determined by the logarithm of the ratio of these corrected relative values.
The chromaticity values rg and yb and the luminance value 1 are respectively corrected, and these correction results krg (rg), kyb (yb), and kl
(L) is normalized within the range of each output range.
Color image quality improvement method. 6. The color image quality improving method according to claim 5, wherein the normalization is performed by using the overall average values rg _mean , yb _mean ,
l _mean each correction result of krg (rg), kyb (y
b _mean ), kl (l _mean ) is the intermediate value V of the output range.
A method for improving image quality of a color image, wherein the method is performed so as to correspond to rg _mid , Vyb _mid , and Vl _mid , respectively. 7. An image input device for inputting a color image.
A step, and all pixels of the color image input by the image input means
Pixel values r, g, b of the color space are represented by respective chromaticity values rg, yb and
Color space conversion means for converting the color image into a luminance value 1;
Average each chromaticity value rg, yb and luminance value l to obtain
Average value rg_mean, Yb_mean, L_meanCalculate the overall average
Calculating means, for each pixel of the color image obtained by the color space converting means
, The chromaticity values rg, yb and the luminance values l and
Chromaticity values rg1 to rg of i pixels in the vicinity thereof
Extract i, yb1-ybi and luminance values l1-li
Neighboring image extracting means; and each color of neighboring pixels extracted by the neighboring image extracting means
Degree values rg1 to rgi, yb1 to ybi and luminance value l1
Ｌｉ li is averaged and the neighborhood average value rg_ave, Yb_ave,
l_aveAnd a total average value rg calculated by the total average calculation means.
_mean, Yb_mean, L_{me an}And the neighborhood average calculating means
The calculated neighborhood average value rg_ave, Yb_ave, L _aveWith
Coefficient αrg corresponding to the ratio_posi, Αyb_posi, Αl_posi , Αrg
_nega, Αyb_nega, Αl_nega And a neighborhood average value rg calculated by the neighborhood average calculation section.
_ave, Yb_ave, L_{av e}And calculated by the coefficient calculation means.
Coefficient αrg_posi, Αyb_posi, Αl_posi Based on
Pixel value variance δrg_posi, Δyb_posi, Δl
_posi Positive image pixel value variable determining means for determining
And a neighborhood average value rg calculated by the neighborhood average calculation means.
_ave, Yb_ave, L_{av e}And calculated by the coefficient calculation means.
Coefficient αrg_nega, Αyb_nega, Αl_nega Based on
Pixel value variance δrg of gative image_nega, Δyb_nega, Δl
_nega Negative image pixel value variable determining means for determining
Determined by the positive image pixel value variable determining means
Pixel value variable δrg_posi, Δyb_posi, Δl_posi And the said
Pixels determined by the gative image pixel value variable determination means
Value variate δrg_nega, Δyb_nega, Δl_nega Based on each of
Each chromaticity value rg, yb and luminance value l of the pixel are complemented.
Correction, the logarithm of the ratio of each correction result is obtained, and the pixel correction value k
Pixels for calculating rg (rg), kyb (yb), kl (l)
Value correction means, and all pixel correction values k calculated by the pixel correction means.
The maximum value of rg (rg), kyb (yb), and kl (l)
krg_max, Kyb_max, Kl_max And minimum value krg_{mi n}, Kyb
_min, Kl_min Correction value maximum / minimum value calculation means for calculating
And a maximum value calculated by the correction value maximum / minimum value calculation means.
krg_max, Kyb_max, Kl_max And minimum value krg_min, K
yb_min, Kl_min On the basis of
Pixel correction values krg (rg), kyb (yb),
kl (l) is normalized within the range of the output range,
Calculate the replacement values Vrg (rg), Vyb (yb), Vl (l)
Pixel value normalizing means, and normalization of each pixel calculated by the pixel value normalizing means.
Based on the conversion values Vrg (rg), Vyb (yb), Vl (l)
And image output means for composing and outputting the entire image.
A color image quality improvement device characterized by the following. 8. The color image quality improving apparatus according to claim 7, wherein the overall average value rg calculated by said overall average calculation means instead of said pixel value normalization means and said image output means.
_{I mean,} yb _mean, l _{me an} and the coefficients calculated by the coefficient calculating means _{αrg posi, αyb posi, αl posi} , αr
g _nega, αyb _nega, based on .alpha.l _nega, the overall average value rg _mean, yb _mean, the correction value of l _mean krg (r
g _mean ), kyb (yb _mean ), kl (l _mean ), and an overall average correction value calculating means, and pixel correction values krg (rg), kyb (yb) of each pixel calculated by the pixel value correcting means. , Kl (l) and the correction value krg (rg) calculated by the overall average correction value calculation means.
_mean ), kyb (yb _mean ), and kl (l _mean ), respectively, and the pixel correction values krg (rg), kyb (yb), k
l (l), the corresponding correction value krg (rg _mean ),
When kyb (yb _mean ) and kl (l _mean ) are large, the pixel correction values krg (rg), kyb (yb), and kl (l) are converted into intermediate values Vrg _mid , Vyb _mid , and Vl _mid in the corresponding output range.
linearly converted to an output range section smaller than l _mid, and the correction values krg (rg _mean ), kyb (y) corresponding to the pixel correction values krg (rg), kyb (yb), kl (l)
When b _mean ) and kl (l _mean ) are equal to each other, the pixel correction values krg (rg), kyb (yb), and kl (l) are set to the intermediate values Vrg _mid , Vyb _mid , and Vl _mid within the corresponding output range.
And the pixel correction values krg (rg), kyb (y
b), kl (l), the corresponding correction value krg (rg
_mean ), kyb (yb _mean ), and kl (l _mean ) are small, the pixel correction values krg (rg), kyb (yb), and kl
(L) is the intermediate value Vrg _mid , within the corresponding output range.
By linearly converting to an output range section larger than Vyb _mid and Vl _mid , each pixel correction value krg (rg), kyb
(Yb), kl (l) are normalized within the range of the output range, and normalized conversion values Vrg (rg), Vyb (yb), Vl
A pixel correction value conversion unit for calculating (l), and an entire image based on the normalized conversion values Vrg (rg), Vyb (yb), and Vl (l) of each pixel calculated by the pixel correction value conversion unit. And a normalized image output unit configured to output the image.