JPH1023279A - Image-processing unit - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide the image-processing unit in which an image is subject to emphasized contrast with a saturation adaptive to the image. SOLUTION: An image data division section 102 classifies image data, forming an input image into a plurality of groups, depending on the luminance of each picture element. A histogram measurement section 103 calculates a saturation histogram, representing the distribution of saturation of the image data for each group. A histogram modification condition setting section 104, a histogram modification section 107 and a level conversion condition setting section 108 generate a lookup table (LUT), to correct saturation of the image data, so that the saturation with high frequency and the saturation with low frequency in the saturation histogram, keep a current frequency, and the frequency of other saturation is close to flatness. A level conversion section 109 uses the LUT, to correct the saturation of the image data.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a saturation correction function for performing a correction adapted to an image, and more particularly, to an image processing apparatus having a saturation contrast enhancement function.

[0002]

2. Description of the Related Art In recent years, digital copiers have been required to have not only faithful reproduction but also higher quality reproduction. For example, if the white part of the background such as a document is dirty, it is better to reproduce the dirt without copying the dirt as it is, and if the character of the pencil is faint, sharpen it sharply It is better to reproduce. In addition, when taking pictures of fruits indoors, due to various shooting conditions such as the intensity of light and the direction of light, the color of the fruits may be lower in saturation than the actual fruits and may be cloudy. Sometimes I take a picture. When such an image is input, it is preferable that the image be reproduced with a sharp and pleasant image by increasing the saturation and increasing the brightness if necessary.

[0003] Further, with the progress of networking / systemization, it has come to face a case where it is necessary to perform a correction process on an image whose environment is unknown under what environment it was created. In this case, external information such as photographing conditions cannot be reflected in the correction processing. Therefore, a correction process must be performed using information from the image itself as a main feature amount.

[0004] In addition, specialized knowledge on image processing, such as what kind of correction processing should be performed when an image shows characteristics, and how much correction should be performed, etc. If not, it is difficult to perform “optimal correction processing”.

[0005] Above all, values having a strong correlation with colors, such as saturation and color balance, have a great influence on an image, so that it is a difficult task for ordinary users to operate the values. With the advent of the color management system recently, the concept of color has become easier for general users to handle,
In order to obtain satisfactory results, empirical, trial and error operations are often required.

[0006] Usually, a feature quantity extracted from the image itself,
Image quality is corrected by automatic processing or manual operation using information provided by an external input. As this type of method, there is a method of correcting the image quality by correcting the luminance of each pixel constituting an image. Further, as the method of correcting the luminance, there are already some contrast enhancement methods using a histogram as a feature amount. Or have been presented. These various methods of performing contrast enhancement on luminance can also be applied as a saturation contrast enhancement method. Hereinafter, these techniques will be described.

First, there is a technique generally called dynamic range conversion. This method obtains the gradation distribution of each pixel constituting the input image, and applies a correction to the gradation of each pixel to linearly extend the gradation distribution to a wider gradation region. The correction is performed.

Here, the procedure of this correction will be described with reference to FIGS. 9 (a) to 9 (c). First, the gradation of each pixel constituting the input image is checked, and a distribution of the gradation, that is, a histogram H (i) representing the number of pixels having the gradation for each gradation i is created. FIG. 9A illustrates this histogram H (i).

Next, based on the distribution range of the histogram H (i), an LUT (gradation conversion table) for associating the gradation of the pixel of the input image with the gradation of the corrected pixel is created. FIG. 9B illustrates this LUT.
In the figure, the horizontal axis is the input gradation of the LUT, that is, the gradation of the pixels constituting the input image, the vertical axis is the output gradation of the LUT,
In other words, it represents the gradation of the pixel after the correction. FIG.
In the example shown in (a), the input image is represented by M gradations, and the gradations of all pixels constituting the input image are minimum values x1 to x1.
It is distributed within the range up to the maximum value x2. Therefore, FIG.
As shown in (b), the points (x1, 0) and (x2, M
An LUT corresponding to one straight line passing through -1) is created.

Next, using this LUT, the gradation of each pixel constituting the input image is converted. By this gradation conversion, the gradation of each pixel of the input image distributed in the range of x1 to x2 is converted into the gradation of 0 to M-1. As described above, the contrast distribution can be enhanced by extending the gradation distribution concentrated in a specific range over a wide range. The histogram G (i) after the gradation conversion is shown by a broken line in FIG. Note that the histogram H (i) before the conversion is shown by a solid line in FIG.

The dynamic range conversion described above is
An LUT is created from the feature amounts of the histogram (in the above example, the minimum and maximum values of the gradation), and the LUT is used to perform gradation conversion to improve the contrast. The following two methods are used. Transforms the histogram itself from the feature amount of the gradation histogram of each pixel constituting the image, and uses the transformed histogram to calculate L
The UT is created.

First, referring to FIGS. 10 (a) to 10 (c),
A technique called histogram flattening, which is widely known as a contrast improvement method, will be described. This is a method of changing the gradation of each pixel constituting an image so that the gradation histogram is distributed over the entire gradation area and the frequency is constant throughout the entire gradation area.

First, FIG. 10A exemplifies a histogram H (i) of gradations of pixels constituting an input image. In the histogram H (i), the number of frequencies in each gradation is normalized so that the total frequency in the entire gradation range is 1. Also in this example, the input image is expressed by M gradations, as in the example of FIG. 9 described above.

In this contrast improvement method, conversion is performed on each pixel of the input image forming such a gradation distribution (that is, histogram H (i)). This conversion is performed after the gradation conversion. This is performed so that each pixel forms a histogram G (i) represented by the following equation (1).

(Equation 1)

FIG. 10B shows the relationship between the frequency of each gradation before and after the conversion. As shown in this figure, the present method ensures that the frequency at each gradation i after gradation conversion has a constant value (the average value of the frequency) regardless of the frequency at each gradation i in the input image. Things. A histogram G (i) after gradation conversion is shown by a broken line in FIG. This method has an advantage that the contrast can be enhanced regardless of the gradation distribution in the input image.

Next, referring to FIGS. 11 (a) to 11 (c), the contrast of the infrequent gradation region is not reduced so much.
A method referred to as parametric conversion for enhancing the contrast of a frequent gradation region will be described.

First, FIG. 11A shows an example of a gradation histogram H (i) of an input image. Also in the histogram H (i), the frequency number in each gradation is normalized so that the total frequency in the entire gradation range becomes 1.
The input image in this example is also expressed in M gradations.

This contrast improvement method also applies gradation conversion to each pixel of an input image. In this conversion, the gradation of each pixel after conversion is represented by a histogram G expressed by the following equation (2).
(I).

(Equation 2)

In the above equation (2), if p = 0, G
(I) = 1 / M, and in this case, the gradation conversion with the highest degree of contrast enhancement that makes the converted gradation histogram G (i) a constant value regardless of the gradation i is performed. . In the above equation (3), if p = 1, G (i) = H (i), and the converted grayscale histogram G (i) is obtained from the input image grayscale histogram H (i). The gradation conversion that does not change and has the lowest degree of contrast enhancement is performed. Thus, the present method adjusts the parameter p in the range of 0 to 1 to
There is an advantage that the degree of contrast enhancement performed by gradation conversion can be continuously adjusted. FIG.
FIG. 11B illustrates the relationship between the frequency of each gradation before and after the conversion in this method, and FIG. 11C illustrates the histogram G (i) of the converted gradation.

[0020]

As described above, there are three methods of applying contrast enhancement using a histogram as methods applicable to saturation contrast enhancement. However, these methods have the following problems. I have.

(1) First, the dynamic range conversion has no effect when the histogram of the input image is distributed in all gradations from the beginning. Also, when the distribution of the histogram is biased, there is also a problem that the bias of the distribution cannot be corrected because the region where the gradation is distributed is simply enlarged.

(2) In the histogram flattening, since the frequencies of all the gradations are distributed so as to be uniform (the average value of the frequencies), the contrast in the gradation region where the frequency is low in the histogram of the input image is lost. there is a possibility. Also,
Since the degree of contrast enhancement cannot be adjusted, contrast correction adapted to the image cannot be performed, and the contrast may be enhanced more than necessary.

(3) Parametric conversion is proposed to solve such a problem of histogram flattening. According to this method, a desired degree of contrast enhancement can be performed by adjusting parameters as well as the intensity enhancement that completely flattens the converted tone histogram. However, in the parametric transformation, the input image is flattened to a high-frequency gradation region which can be said to be an important feature.Therefore, the processed image may not retain the features of the input image. Not be. Further, the parametric conversion has a problem that, when the gradation distribution is concentrated in a certain range, the contrast enhancement is performed only within that range.

(4) In addition to the above problems, when applying the above three methods as saturation contrast enhancement, the color gamut of the output device must be considered. FIG. 12 shows an example of the color gamut of the monitor and the color printer on the a * -b * plane when the L * value in the L * a * b * space is fixed. As shown in FIG. 12, the color gamut of a color printer or a color copier is usually smaller than the color gamut of a monitor. For this reason, it is a phenomenon that the impression and the color to be received often differ between the image viewed on the monitor and the image actually output. Image data of a certain pixel (L *, a
*, B *) is (L1, a1, b1), the saturation value of the pixel is the point (L1, a1) in the L * a * b * space.
1, b1) is expressed as the distance from the point (L1, 0, 0). As a result of modulating this distance by enhancing the saturation contrast, the saturation of the image exceeds the color reproduction range of the output device. It can happen. If such a situation occurs, it cannot be said that the correction processing is optimal.

The present invention has been made in view of the circumstances described above, and it is an object of the present invention to provide an image processing apparatus capable of performing saturation contrast enhancement adapted to an image.

[0026]

According to the present invention, a saturation histogram of image data constituting an input image is obtained, and a high-frequency saturation and a low-frequency saturation maintain the current frequency in the saturation histogram. Further, the present invention provides an image processing apparatus characterized in that the saturation of each image data is corrected so that the frequency of other saturation approaches flat.

Further, according to the present invention, image data dividing means for classifying image data constituting an input image into a plurality of groups based on luminance of each pixel, and calculating a saturation histogram of each image data for each of the groups. The saturation histogram calculation means, and for each of the groups, the high-frequency saturation and the low-frequency saturation in the saturation histogram of the group maintain the current frequency, respectively, and the frequency of the other saturation approaches flat. Thus, there is provided an image processing apparatus including a correction means for correcting the saturation of each image data belonging to the group.

According to these image processing apparatuses, it is possible to enhance the contrast of the saturation without lowering the contrast of the infrequent portion and while preserving the important features of the image. The correction of the saturation is automatically performed based on the saturation histogram obtained from the image data. Therefore, it is possible to perform a correction adapted to an image without performing excessive emphasis.

[0029]

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

A. First Embodiment FIG. 1 is a block diagram showing a configuration of an image processing apparatus according to a first embodiment of the present invention. In FIG. 1, an image input device 101 is input means for inputting image data of an input image to the image processing device, and is constituted by a digital multi-value image input device such as a scanner, a memory, and the like.

The image data dividing section 102 divides image data input via the image input device 101 in a color space based on luminance. That is, the image data is
The image data is composed of image data representing each pixel of the input image. The image data dividing unit 102 classifies each image data corresponding to each pixel into a plurality of groups according to the intensity of each pixel.

The histogram measuring unit 103 creates a saturation histogram of each pixel belonging to each group obtained by the division. In other words, each of the groups is composed of image data corresponding to each pixel whose luminance is close to each other, but the histogram measuring unit 103 determines, for each group, the saturation value of the image data of each pixel belonging to each group. The saturation histogram is created by counting the number of frequencies (the number of pixels having the saturation value) for each saturation value.

The histogram transformation condition setting section 104 and the histogram transformation section 107 maintain the current frequency for the high-frequency saturation and the low-frequency saturation in the above-mentioned saturation histogram, and flatten the other saturation frequencies. Is a means for performing a histogram deformation process of generating a corrected saturation histogram approaching the above from the saturation histogram.

First, the histogram deformation condition setting unit 104
Is a means for obtaining a non-linear function used in the above-mentioned histogram deformation processing, a feature amount extracting unit 105 for extracting a feature amount from the saturation histogram of each group, and a correction coefficient constituting the nonlinear function from the feature amount. It comprises a correction coefficient setting unit 106 to be calculated.

The histogram transformation unit 107 converts the frequency of the saturation histogram corresponding to each group using a nonlinear function using the correction coefficient as a parameter, and converts the corrected saturation histogram obtained by modifying each saturation histogram into each group. Generated every time. The conversion of the frequency from the saturation histogram to the corrected saturation histogram is performed such that the low frequency and the high frequency each maintain the same frequency as the current frequency, and the other frequencies are the flattened frequencies.

The level conversion condition setting section 108 sets a level conversion condition for performing optimal contrast correction. More specifically, the level conversion condition setting unit 10
8 is a chroma conversion table for converting the chroma required for transforming the chroma of the chroma histogram into the same histogram as the corrected chroma histogram. To create. In other words, the above-described histogram transformation unit 107 can obtain an ideal corrected saturation histogram on which the optimum contrast correction has been performed, and the level conversion condition setting unit 108 converts the current saturation histogram into such an image. Saturation level conversion condition for transforming to ideal corrected saturation histogram (ie, saturation conversion table)
Is what you want.

The level conversion unit 109 executes the level conversion of the saturation of each pixel of the input image (saturation conversion processing) according to the level conversion conditions, and converts each image data of each pixel having an ideal corrected saturation histogram. It is a means for generating.

The image output device 110 is a means for outputting image data processed by the image processing device to the outside, and is composed of, for example, an image output device such as a printer or a memory. The memory 100 includes a histogram measuring unit 103, a histogram deformation condition setting unit 10
4. It is connected to the histogram transformation unit 107 and the level conversion condition setting unit 108 via a bidirectional bus.

Next, the operation of this embodiment will be described with reference to the flowchart shown in FIG. First, image data of a digital multivalued image is input by the image input device 101 (step S200). When the image input device 101 is a memory, the image data input process is performed by reading out image data stored in advance in the memory, and the image input device 101 has a scanner or an image reading function. In the case of means such as a digital copying machine, the image is scanned by such means.

The image processing after the input processing of the image data is performed in accordance with a process called prescan for analyzing the input image data and creating optimum level conversion conditions, and a level conversion condition obtained by the prescan. And a process called main scan for performing level conversion of the image data of the input image. Steps S201 to S20 in FIG.
8 constitutes a pre-scan, steps S209 and S209
210 constitutes the main scan.

First, the prescan will be described. At first,
The image data dividing unit 102 divides the image data of the input image supplied via the image input device 101 on a color space based on a plurality of divided luminance ranges. For example, in image data having a luminance of 256 gradations, when dividing an image into four, pixels having L * = 0 to 63 are grouped into one group, and similarly, L * = 64 to 127, 128 to 191,
By creating a group of pixels for each of the luminance ranges of 192 to 255, the image data is divided on the color space based on the luminance.

The image data division unit 102 creates a saturation histogram H (i) of the image data for each of the groups thus divided, and stores it in the memory 100 (step S202). In the saturation histogram H (i) created here, the horizontal axis represents the saturation level, and the vertical axis represents the frequency.
FIG. 3A illustrates the saturation histogram H (i) of a certain group created in this way.

The feature value extraction unit 105 extracts the feature value of the image from the saturation histogram (step S203). From the saturation histogram H (i) stored in the memory 100, the minimum value MIN of the frequency, the maximum value MAX of the frequency, the average value AVE of the frequency, and the standard deviation σ of the frequency are calculated as the image feature amounts.

The correction coefficient setting unit 106 includes the feature amount extracting unit 1
Saturation histogram H using the feature amount extracted in step 05
Two correction coefficients γ1 and γ2 for transforming (i) are calculated (step S204). The correction coefficients γ1 and γ2 are
Each is calculated by the following equations (3) and (4).

(Equation 3)

[0045]

(Equation 4)

The histogram deformation unit 107 uses the two correction coefficients γ1 and γ2 calculated by the histogram deformation condition setting unit 104 to convert the saturation histogram H (i) stored in the memory 100 into the corrected saturation histogram G
It transforms to (i) (step S106). The frequency conversion formula used here is represented by the following (5) using γ1 and γ2.
It is defined by equation (6). The curve shown in FIG. 3B shows this frequency number conversion equation in a two-dimensional coordinate system in which the horizontal axis represents the frequency of saturation before conversion and the horizontal axis represents the frequency of saturation after conversion. It is an expression.

(Equation 5)

[0047]

(Equation 6)

With this conversion formula, the low frequency or the high frequency is held as it is, the frequency conversion is performed so that the frequencies of the other levels approach the average, and the saturation histogram H
A corrected saturation histogram G (i) obtained by modifying (i) is obtained. Here, the corrected saturation histogram G (i) is normalized by the total frequency to obtain G (i) ′. The normalized corrected saturation histogram G (i) ′ becomes a broken line in FIG. 3C and is stored in the memory 100.

The level conversion means setting section 108 stores the memory 1
The level conversion condition is set using the pre-transformation saturation histogram H (i) and the post-transformation correction saturation histogram G (i) ′ stored in 00 (step S207). That is, in this step S207, the saturation histogram H
A saturation conversion table for converting the saturation required when transforming the saturation of (i) into the same histogram as the corrected saturation histogram G (i) ′. Create Specifically, H
Cumulative histogram Ha of both (i) and G (i) ′
(I) and Ga (i) are calculated, and an LUT (saturation conversion table) is created by associating the levels having the same cumulative frequency in the cumulative histograms Ha (i) and Ga (i). This LUT creation method is shown in FIG.

In the cumulative histogram Ha (i) shown in FIG. 4, when i is the saturation in, the cumulative frequency is F (arrow). At this time, the saturation out at which the cumulative frequency is F in the cumulative histogram Ga (i) is obtained (arrow).
Thereafter, the i is moved by the number of possible gradations, and a pair of saturation in and saturation out is sequentially obtained. A large number of pairs of saturation in and saturation out thus obtained are collected, and the saturation of the input pixel (corresponding to saturation in) and the corrected saturation of the pixel (corresponding to saturation out) are associated with each other. Create an LUT.

The operations from step S202 to step S207 described above are executed for each group divided in step 201 (step S208). As a result, an LUT (saturation conversion table) is obtained for each group obtained by the division in step 101.

Next, the main scanning will be described. The level conversion unit 109 uses the LUT created by the level conversion condition setting unit 108 to perform level conversion of information relating to saturation in the image data of the input image supplied from the image input device 101 (step S209). . In the image output device 210, the level conversion unit 209 outputs image data that has been subjected to level conversion for the information on the saturation (step S210).

As described above, in the present embodiment, the low-frequency saturation and the high-frequency saturation in the input image are maintained as they are even after the level conversion, and the intermediate frequencies other than these are maintained. Saturation level conversion is performed so that the frequency of each saturation after the level conversion approaches the average, so that the overall saturation contrast is enhanced while maintaining the characteristics of the input image. Becomes

In the above embodiment, the frequency conversion formulas used in the histogram transformation unit 107 are (5) and (6).
Although defined by the equation, it may be approximated by another function or the like. For example, two spline functions may be combined to form a curve as shown in FIG. 3B and used as a frequency conversion formula.

B. Second Embodiment FIG. 5 is a block diagram showing a configuration of an image processing apparatus according to a second embodiment of the present invention. This image processing apparatus is obtained by adding an external information designating unit 511 for designating color gamut data of an output device to the above-described first embodiment. Reference numeral 50 other than the external information designation unit 511 in FIG.
Elements denoted by 0 to 510 are denoted by reference numerals 10 in FIG.
It corresponds to each element numbered 0 to 110. Note that the image input device 501 and the image data dividing unit 502 are the same as those in the first embodiment, and thus the description is omitted.

The memory 500 stores color gamut data of several types of image output devices in advance. The external information specifying unit 511 specifies color gamut data of the image output device 510 to be used from several types of color gamut data stored in the memory 500. The specifying method is such that the external information specifying unit 511 interacts with the user to obtain information specifying the image output device, and specifies the corresponding color gamut data based on the obtained information. Alternatively, a configuration may be employed in which, when the image output device 510 is determined by another method, a corresponding one of the color gamut data stored in the memory 500 is automatically specified.

The histogram measuring unit 503 creates a saturation histogram H (i) normalized in the color reproduction range of the output device. Details of this processing will be described with reference to FIG.

First, the histogram measuring unit 503 calculates the luminance, hue, and saturation of each pixel constituting the input image. Here, the image data p of a certain pixel is (l, a, b)
Then, the luminance L of the pixel becomes 1, and the hue h and the saturation S _{l, h} are calculated from a and b.

Next, the histogram measuring unit 503 obtains the maximum reproducible saturation of each pixel using the color gamut data specified by the external information specifying unit 511. FIG.
(A) is a table showing the maximum color saturation of color reproduction. This table is a two-dimensional array table using the quantized lightness L as a row index and the quantized hue H as a column index. Each array element SG _{L, H} (L = 0 to 255, H =
0 to 255) stores the maximum reproducible chroma.

The histogram measuring section 503 calculates the maximum reproducible saturation SG _{l, h} corresponding to the luminance 1 and the hue h calculated from the image data p for each pixel in FIG.
Search from the table.

Next, the histogram measuring unit 503 calculates the normalized saturation SN _l by normalizing the saturation S _{l, h} of each pixel by the maximum reproducible saturation SG _{l, h} of each pixel. . For example, the image data dividing unit 502
When divided into six, the histogram measuring unit 503 sets
For each luminance of 55, a histogram H (l) of the normalized saturation SN _l of each pixel having the luminance. FIG.
(B) illustrates the normalized saturation histogram H (l) created for the 256 types of luminance 1 in this manner.

The histogram deformation condition setting unit 504, feature amount extraction unit 505, and correction coefficient setting unit 506 perform the first
Since the configuration is the same as that of the embodiment, the description is omitted.

As in the first embodiment, the histogram transformation section 507 includes a histogram transformation condition setting section 504.
The histogram is modified using the correction coefficient supplied from. In the level conversion condition setting unit 508, the created LU
Since T is a normalized value, a procedure reverse to the normalization procedure shown in FIG. 6 is executed to return the value of the LUT to the original scale. The level converter 509 and the image output device 510 are the same as those in the first embodiment, and a description thereof will be omitted. In this embodiment, the same effects as those of the first embodiment can be obtained.

C. Third Embodiment FIG. 7 is a block diagram illustrating a configuration of an image processing apparatus according to a third embodiment of the present invention. In the present embodiment, processing for adjusting the saturation value of the entire image up and down as well as enhancement of the saturation contrast is added. Further, the image data is not divided as in the first and second embodiments, and the saturation of the entire image is corrected by a single process.

In FIG. 7, the image input device 701 and the memory 700 are the same as those in the second embodiment, and the description is omitted.

The external information specifying section 712 has a function of specifying an adjustment coefficient used in a saturation adjusting section 713 described later, in addition to the function of the external information specifying section 511 in the second embodiment. As the adjustment coefficient, the user may directly input a numerical value, or may specify a desired coefficient from a plurality of adjustment coefficients stored in the memory 700 in advance. In order to prevent over-emphasis, an upper limit value of the adjustment coefficient may be input.

The color space normalizing section 702 is provided for the image input device 7
01 is normalized using the color gamut data of the image output device 711 designated by the external information designation unit 712. Here, referring to FIG.
This color space normalization will be described.

When the color space of the image to be handled is the L * a * b * space, the image data of each pixel constituting the input image has a shape as shown in FIG. 8A in the L * a * b * space. It is distributed in. The color space normalizing unit 702 normalizes the image data of each pixel having such a distribution so as to adapt to the color gamut of the designated output device, and calculates the distribution of the image data in the color space. The shape is changed to a cylindrical shape as shown in FIG. This is color space normalization.

In order to perform such a color space normalization, the color space normalization unit 702 in FIG.
The normalized chroma of each pixel is obtained from the image data p = (l, a, b) of each pixel and the maximum color reproduction value SG corresponding to the output device by exactly the same procedure as described with reference to FIG. S
_Nl is calculated and supplied to the saturation adjustment unit 713. When the saturation of all the pixels is normalized, the image data originally distributed in the color space as shown in FIG.
Is distributed in the color space in the shape as shown in FIG.
As described above, by performing normalization in the color space, it is possible to collectively handle saturation values having different reproducible ranges for each luminance and hue.

The saturation adjusting section 713 includes a color space normalizing section 70.
The normalized saturation value supplied from 2 is adjusted by the adjustment coefficient. This adjustment may be performed using an adjustment coefficient specified by the external information specifying unit 712, or processing may be performed by adaptively calculating an adjustment coefficient from an input image. Further, the processing may be performed using the upper limit value supplied by the external information specifying unit 712 and the value determined as the upper limit of the adjustment coefficient.

The contrast emphasizing method adopted by the present invention is a contrast emphasizing method in which the average value of the saturation of the entire image is substantially equal before and after the processing, and has an effect of producing sharp colors. . However, for example, in the case of an original image having low saturation, contrast enhancement is performed only in low saturation. Therefore, the process of adjusting the average value of the saturation of the entire image up and down is also used, and when the image saturation is low, the average value is increased and the image is vivid, and when the image saturation is high, the average value is decreased. If the image becomes calm and the color is sharpened further by the contrast enhancement processing, it will be a very effective saturation correction. For this reason, in the present embodiment, after the saturation of the entire image is adjusted by the saturation adjustment unit 713, contrast enhancement is performed.

An example of a process for adaptively calculating the adjustment coefficient will be described below. First, the maximum value is detected from the normalized saturation values supplied from the color space normalization unit 702. And
The normalized saturation value is multiplied by the ratio between the maximum saturation value and the maximum reproducible saturation value of the output device (that is, 1.0) as an adjustment coefficient to adjust the saturation value of the entire image. Here, when the calculated adjustment coefficient is larger than the upper limit value specified by the external information specifying unit 712, the upper limit value is substituted for the adjustment coefficient and multiplied by the normalized saturation value.

In the histogram measuring section 703, one normalized chroma histogram is created based on the chroma values of the image data adjusted by the color space normalizing section 702 and adjusted by the chroma adjusting section 713.

In the above-described second embodiment, FIG.
As shown in FIG. 6, a normalized saturation histogram is calculated for each luminance. However, in the present embodiment, such a treatment is not performed in units of luminance, and thus a plurality of normalized saturation histograms in FIG. Collectively, only one normalized saturation histogram of the entire image needs to be created.

The configuration from the histogram deformation condition setting unit 704 to the level conversion unit 709 is the same as that in the first embodiment, and the description is omitted.

The color space inverse conversion unit 710 converts the image supplied from the level conversion unit 709, that is, the image normalized in the output device color reproduction range and further subjected to the saturation contrast enhancement processing as shown in FIG. Then, inverse conversion is performed to return to the original color space. The image output device 711 is the same as that in the first embodiment, and a description thereof will be omitted.

In this embodiment, the saturation adjusting section 713 is arranged before the histogram measuring section 703. However, the same effect can be obtained by arranging after the histogram measuring section 703. In this case, an adjustment coefficient is calculated from the normalized histogram supplied from the histogram measurement unit 703, the normalized histogram is deformed using the adjustment coefficient, and supplied to the histogram deformation condition setting unit 704. In this embodiment, the same effects as those of the second embodiment can be obtained.

[0078]

As described above, according to the present invention, it is possible to enhance the contrast of saturation without lowering the contrast of a low-frequency portion and maintaining important features of an image. I can do it. In addition, since the saturation correction is performed based on the saturation histogram obtained from the image data, it is possible to prevent excessive emphasis, and the correction automatically adapted to each image without performing trial and error. (Claims 1 to 3). In addition, since the saturation histogram is corrected based on the color gamut data of the device that outputs the corrected image data, effective saturation contrast enhancement can be performed within the color gamut of the device. .

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of an image processing apparatus according to a first embodiment of the present invention.

FIG. 2 is a flowchart showing an operation of the embodiment.

FIG. 3 is a diagram showing a saturation correction method in the embodiment.

FIG. 4 is an LUT for saturation correction in the embodiment.
FIG. 4 is a diagram showing a method of creating a.

FIG. 5 is a block diagram illustrating a configuration of an image processing apparatus according to a second embodiment of the present invention.

FIG. 6 is a diagram showing a procedure for creating a normalized saturation histogram in the embodiment.

FIG. 7 is a block diagram illustrating a configuration of an image processing apparatus according to a third embodiment of the present invention.

FIG. 8 is a diagram illustrating color space normalization and color space inverse conversion in the embodiment.

FIG. 9 is a diagram illustrating a conventional method of enhancing the contrast of image data.

FIG. 10 is a diagram illustrating a conventional method of enhancing the contrast of image data.

FIG. 11 is a diagram illustrating a conventional method of enhancing the contrast of image data.

FIG. 12 is a diagram illustrating an example of a color gamut of an output device.

[Explanation of symbols]

102, 502... Image data division unit (image data division means), 702... Color space normalization unit, 103, 503
703: Histogram measurement unit (saturation histogram calculation unit), 104, 504, 704: Histogram deformation condition setting unit (correction unit), 107, 507, 707 ...
Histogram transformation unit (correction means), 108, 508, 7
08: level conversion condition setting unit (correction means), 109,
509, 709... Level converter (correction means)

Claims (5)

[Claims] 1. A saturation histogram of image data constituting an input image is obtained. In the saturation histogram, a high-frequency saturation and a low-frequency saturation maintain their current frequencies, respectively, and the other saturation frequencies An image processing apparatus characterized in that the saturation of each image data is corrected so as to approach flat. 2. An image data dividing means for classifying image data constituting an input image into a plurality of groups based on luminance of each pixel, and a saturation histogram calculation for calculating a saturation histogram of each image data for each of the groups. Means, for each of the groups, in the saturation histogram of the group, the high-frequency saturation and the low-frequency saturation maintain the current frequency, respectively, and the other groups in such a manner that the frequencies of the other saturations approach flatness. Correction means for correcting the saturation of each image data belonging to the above. 3. A means for normalizing saturation based on color gamut data of a device for outputting corrected image data, and calculating the saturation histogram using the normalized saturation. The image processing apparatus according to claim 1, wherein the saturation of the image data is corrected based on a saturation histogram normalized by. 4. A method for correcting the saturation of the image data, comprising: a. A histogram modification in which the high-frequency saturation and the low-frequency saturation in the saturation histogram maintain the current frequency, respectively, and generate a corrected saturation histogram from the saturation histogram in which the frequency of other saturation approaches flatness. Processing; b. By converting the saturation of the saturation histogram, a saturation conversion table for converting the saturation required when the saturation histogram is transformed into the same histogram as the corrected saturation histogram is created. Conversion condition setting processing; c. 4. A saturation conversion process for converting the saturation of each image data corresponding to each of the pixels by the saturation conversion table. Image processing device. 5. The method according to claim 1, wherein the histogram deformation processing includes: a1. Extracting feature values of the saturation histogram; a2. A process of obtaining a non-linear function that converts the low frequency and the high frequency into the same frequency as the current frequency, and converts the other frequencies into the flattened frequency; The image processing apparatus according to claim 4, further comprising: converting a frequency in the saturation histogram using the nonlinear function.