JP2000201277A - Image correcting method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain an image of appropriate hue all the time even about an image in which sufficient correction of white balance can not be performed with the conversion of a density histogram. SOLUTION: Each density histogram of RGB is extended to improve image contrast and to correct 1st white balance (s9 to s11), and after that, the average value of pixel data of each color of the RGB is calculated about white pixels extracted in the original image only for an image where any maximum value of pixel values in each density histogram is larger than a reference value and the correction of the 1st white balance due to the density histogram extension is not effective, the conversion coefficient of each color of the RGB for making the white pixels whiter is calculated by using the average value, and 2nd white balance correction in which respective pixel data of each color of the RGB of all pixels of the original image are multiplied by the conversion coefficient is carried out (s12 to s15).

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a color still image correcting method for correcting the contrast and white balance of a color still image obtained by a digital still camera, an image scanner or a video capture circuit.

[0002]

2. Description of the Related Art As a means for converting a video signal into a color still digital image, a method using a digital still camera or an image scanner equipped with a solid-state image pickup device such as a CCD or C-MOS area sensor, and a video camera or the like are used. There is a method using a video capture circuit that converts an analog image signal being captured into digital data in real time. Above all, reflecting the recent reduction in the price of high pixel density CCDs, high quality still images can be easily obtained with digital still cameras equipped with CCDs.

However, a CCD has a lower sensitivity than a silver halide photograph, and when the illuminance of a subject is not sufficient, the CCD is used.
There is a problem that the contrast of the image obtained by the above becomes insufficient, and the color tone becomes extremely red or blue due to improper white balance. Further, in an image read by an image scanner, there is a problem that if the setting of the reading device driver is not appropriate, the color tone is distorted and the brightness becomes dark.

For this reason, in a conventional imaging apparatus such as a digital still camera, an internal data signal processor performs white balance correction, adjusts the amplification width of RB according to the color temperature of a light source, and obtains an image having an appropriate white balance. I'm trying to get For example, JP-A-9-559
No. 48 discloses a configuration for correcting white balance based on digital data of a high-luminance portion of an image captured by a CCD.

In the case of a still image, as a means for correcting the image quality, a density histogram is created for each of the RGB colors, and the density histogram is expanded so that the minimum and maximum pixel values in the density histogram are enlarged. Accordingly, a density histogram conversion method for improving the contrast of an image and correcting a white balance is used. As a method of expanding the density histogram, for example, the R component of the original image and the B
If the components are distributed over the entire pixel value, the G component is distributed over the low pixel values, and the original image has a magenta hue as a whole, the distribution of the G component in the density histogram By extending the range to the entire pixel value, the components of each color of RGB are balanced to convert the image into a natural color tone image.

[0006]

However, in the above-described method of correcting contrast and white balance by converting the density histogram, when the white balance of the original image is extremely inappropriate and there is a large difference between the density histograms of RGB colors. Is effective, but when the white balance of the original image is not so much disturbed and the maximum value of the pixel value in the density histogram of each color of RGB does not have a large difference, the conversion coefficient of the density histogram for each color of RGB is substantially equal. Even though the contrast is changed, the balance of each color of RGB hardly changes, and there is a problem that the white balance cannot be sufficiently corrected.

For example, when a highlight having extremely high luminance exists in the original image and the maximum value of the density histogram is extracted in the highlight portion, or when the brightness of the original image is sufficient, the color tone collapses. RGB even if there is a bias
The density histogram of each color is distributed over substantially the entire range of pixel values, and the color tone cannot be sufficiently improved even if the density histogram is expanded.

[0008] An object of the present invention is to execute another white balance correction to make the RGB average values of white pixels equal for an image for which sufficient white balance correction cannot be performed by conversion of the density histogram. Another object of the present invention is to provide an image correction method for a color still image that can always obtain an image having an appropriate color tone regardless of the state of an original image.

[0009]

According to the first aspect of the present invention, the contrast of a color original image is improved and the first contrast is improved.
In an image correction method for expanding a density histogram for white balance correction, a color original image whose color tone cannot be properly corrected by only the density histogram expansion A second white balance correction for converting all pixel data based on a color tone of a white pixel is performed.

According to the first aspect of the present invention, for a color original image whose color tone cannot be properly corrected only by expanding the density histogram, the color tone of white pixels in the color original image is changed after the density histogram is expanded. All pixel data is converted as a reference. Therefore, even if the density histogram is expanded, a sufficient change in the color tone is provided even if the density histogram is expanded, for example, as in a color original image having a bias in the color tone even though the density histograms of the RGB colors are distributed over substantially the entire value range. For an image that cannot be processed, another white balance correction based on the color tone of the white pixel is executed, and the color tone bias in the color original image is reliably corrected.

In the invention described in claim 2, prior to the second white balance correction, the maximum value of the density histogram of each color of RGB in the original color image is compared with a reference value, and any one of the RGB color tone is used. The second white balance correction is performed when the color is biased and the maximum value of each of the density histograms of the RGB colors exceeds a reference value.

According to the second aspect of the present invention, a color original image in which the color tone is biased to any one of the RGB colors and the maximum value of each of the density histograms of the respective colors of RGB exceeds the reference value is defined as a density image. It is determined that the image cannot give a sufficient change in color tone only by expansion of the histogram, and the second white balance correction is performed on this image. Therefore, the second white balance correction is accurately weighted only for an image for which a sufficient change in color tone cannot be given even if the density histogram is expanded, and a sufficient change in color tone is given by expanding the density histogram. The second white balance correction is not executed with weight for the image that can be obtained, and the color tone is always properly corrected in the shortest time regardless of the characteristics of the image.

According to a third aspect of the present invention, in the second white balance correction, a white pixel is extracted from a color original image, an average value of pixel data of each color of RGB of the white pixel is calculated, and Calculating a conversion coefficient for each of the RGB colors in which all of the average values of the pixel data of each of the RGB colors match the average luminance of the color original image, and multiplying the pixel data of all the pixels of the color original image by the conversion coefficient. Features.

According to the third aspect of the present invention, white pixels are extracted from a color original image in which a sufficient change in color tone cannot be given even if the density histogram is expanded, and RGB colors of the extracted white pixels are extracted. The average value of the pixel data is calculated, and the pixel data of each color of RGB in all the pixels of the color original image is converted by the conversion coefficient determined so that the average value of the image data of each color matches the average luminance of the color original image. Is done. Therefore, for a color original image in which a sufficient change in color tone cannot be given even when the density histogram is expanded, the color tone of the entire color original image is appropriate so that white pixels are whiter in a range where the luminance of the color original image is not changed. Is corrected to

According to a fourth aspect of the present invention, in the second white balance correction, the average luminance of the pixels included in the peripheral portion of the original color image and the average luminance of the pixels included in the central portion are compared. The processing target is changed based on the result.

In the invention described in claim 4, the second white balance correction is performed based on the result of comparing the average luminance of the pixels included in the peripheral portion of the original color image with the average luminance of the pixels included in the central portion of the color original image. Is changed. Therefore, the case where the color tone of the central portion is preferentially corrected, such as when a backlight portion is included in the central portion of the image, and the case where the entire image is corrected, such as a landscape photograph, are accurately distinguished. When the color tone is preferentially corrected, no correction is performed on the peripheral pixels, and the image processing time is reduced.

According to a fifth aspect of the present invention, after the second white balance correction, a gradation conversion process by gamma correction is performed.

In the invention described in claim 5, the brightness of the color original image after the second white balance correction is corrected by the γ correction process. Therefore, the brightness is further corrected for the image data whose contrast and color tone have been made appropriate by the improvement of the contrast, the first white balance correction and the second white balance correction, and the image is output in a more appropriate state. .

[0019]

FIG. 1 is a block diagram showing a configuration of an image processing system to which an image correction method according to an embodiment of the present invention is applied. The image correction device 1 forms an image processing system together with the image output device 2 and the image forming device 3. The image output device 2 is, for example, a digital video camera, an image scanner, a video capture board, or the like, and the image forming device 3 is, for example, an image forming unit or a printer of a digital copying machine. The image correction device 1 includes an image output device 2, an image scanner and an image forming device 3.
In some cases, one digital copying machine is configured as an image processing system together with the image forming unit.

The image correction apparatus 1 includes a ROM 12 and a RAM
An image data input unit 14, an image memory 15, and an image data output unit 16 are connected to a CPU 11 having a CPU 13. The CPU 11 reads image data output from the image output device 2 via the image data input unit 14, temporarily stores the read image data in the image memory 15, and performs predetermined image processing in accordance with a program previously written in the ROM 12. And outputs the image data from the image data output unit 16 to the image forming apparatus 3 via the image memory 15.

FIG. 2 is a flowchart showing a processing procedure of the image correction method. CPU 11 of image correction device 1
Reads an image file in an image format such as Windows BMP, JPEG, GIF or PNG captured by the image output device 2 such as a digital still camera, an image scanner or a video capture board via the image data input unit 14 (s1); It is developed as a bit map on the image memory 15. Next, the CPU 11 divides the bitmap developed on the image memory 15 into five sections in each of the vertical and horizontal directions, and as shown in FIG. Ga2
It is divided into four (s2).

As described above, by dividing the original image into a predetermined number of areas, the original data is extracted from the attribute data representing the attributes of the image such as the average value of the pixel data of each color of RGB, the I and Q values, and the color difference signals. When the characteristics of the subject in the image are determined and the optimal image correction processing is selected, the attribute correction data is calculated for only a part of the area, so that the image correction processing can be shortened.

Thereafter, the CPU 11 determines the nine central areas Ga6 to Ga6 shown in FIG.
The average luminance Yc of 8, Ga11 to 13, and Ga16 to 18 is calculated (s3). The average luminance Yc is calculated for each of the pixels included in the nine central areas by calculating the luminance Y of each pixel calculated by the following equation 1 in consideration of the human visual sensitivity characteristics based on the RGB components of each pixel: It is obtained by dividing by the number of pixels included in the nine central areas.

[0024]

(Equation 1) Further, the CPU 11 controls the sixteen peripheral areas Ga0 to Ga4, Ga5, Ga9, Ga10, and Ga shown in FIG.
The average luminance Ys of 14, Ga15, Ga19, and Ga20 to 24 is calculated in the same manner as the average luminance Yc (s4), and the average luminance Yc in the central area and the average luminance Ys in the peripheral area are calculated.
Is compared with (s5). If the average luminance Yc in the central area is lower than the average luminance Ys in the peripheral area, CP
U11 determines that the subject located at the center of the image is an image in a backlight state, sets the central area as a target of correction processing described later (s6), and sets the average luminance Y of the central area.
If c is equal to or higher than the average luminance Ys of the peripheral area,
It is determined that the image is a landscape image or a group image of a plurality of persons, and the entire image is set as a target of the correction processing (s7).

Next, the CPU 11 creates a density histogram for each color of RGB for the area to be corrected set in s6 or s7, and obtains the maximum pixel value Rmax, Gmax, B in each density histogram.
max and minimum values Rmin, Gmin, Bmin are obtained (s8). Thereafter, the CPU 11 expands each density histogram and performs image contrast correction and first white balance correction (s9 to s11).

As shown in FIG. 4, for example, when the density histogram of any one of the RGB colors in the original image is represented by the curve A, the minimum pixel value Amin and the minimum pixel value By expanding Amax to the minimum value and the maximum value of the range of the pixel data, the density histogram is converted into the state shown by the curve B. This pixel value conversion is performed by setting the pixel value of the density histogram (curve A in the example shown in FIG. 4) of the original image to x, and the pixel value of the density histogram after expansion (curve B in the example shown in FIG. 4) to y. As shown in FIG. 5, this can be performed using the general formula y = ax + b.

The minimum value and the maximum value of the range of the pixel data are, for example, 8 bits for each color of RGB for each pixel.
In the case of the resolution of t, the pixel data takes a value range from “0” to “255”, the minimum value of the pixel data range is “0”, and the maximum value of the pixel data range is “255”.

By this processing, the value of each color of the pixel is distributed over the entire range from the minimum value to the maximum value of the value range of the pixel data, and the contrast is corrected. As an example, as shown in FIG. 6, when the distribution range of the G component is wide, the distribution range of the R component and the B component is biased toward a range in which the pixel value is low, and when the image is a red tone image as a whole, RGB Is different in the distribution range of the density histogram of each color. In this case as well, the distribution ranges of the density histograms of the respective colors of RGB coincide with each other, and the color tone of the image is also corrected.

Thereafter, the CPU 11 determines the maximum pixel value Rmax, Gmax, Bma of the pixel value in each density histogram.
x is compared with a reference value Yth (s12), and a second white balance correction process is selectively executed according to the comparison result. That is, the CPU 11 determines in the comparison that s9
An image for which the first white balance correction processing by extending the density histogram in 〜s11 is not effective is extracted. Therefore, as the reference value Yth, a value representing a feature common to images whose color tone cannot be sufficiently corrected depending on the expansion of the density histogram is selected. For example, when the value range of the pixel data is “0” to “255”, “245” is set as the reference value Yth.

The CPU 11 finds in the comparison of s12 that the color tone of the color original image is biased, and that the maximum value Rmax, Gmax of the pixel value in each density histogram is obtained.
If both x and Bmax are larger than the reference value Yth, it is determined that the image is not effective for the first white balance correction processing by expanding the density histogram. FIG. 7A shows an example of such an image.
As shown in the figure, although the highlight portion having a large area exists in the image and the overall color tone is biased toward G, the maximum value of the density histogram of each color of RGB is “255”.
7B, the density histogram of each color of RGB is distributed over the entire value range, and the maximum value of each is close to “255” as shown in FIG. 7B. Some images are slightly biased toward G.

For an image as shown in FIG. 7A or 7B, the CPU 11 performs the second white balance correction in s13 to s15. That is, a white pixel which should be originally white in the image is extracted, and the pixel data of each color of RGB of the extracted white pixel is individually added and divided by the total number of white pixels, thereby obtaining the RGB of each color of white pixel. The average values Wr, Wg, Wb of the pixel data are calculated (s13). Here, the extraction of the white pixel focuses on the fact that the saturation is extremely low in the white pixel, and selects a pixel in which the values of I and Q calculated by the following equations 2 and 3 are both smaller than the reference value S as the white pixel. I do.

[0032]

(Equation 2) Next, the CPU 11 calculates the average value W calculated in s13.
Using r, Wg, and Wb, the conversion coefficients Kr, Kg, and Kb for each of the RGB colors for making the white pixels whiter are calculated (s14). Here, if it is assumed that the average luminance Wy of the original image is not changed after the conversion of the white pixels, the following equation 4 is established, and the relationship shown in the following equation 5 is obtained from the equation 4.

[0033]

(Equation 3) Further, the CPU 11 multiplies each of the RGB pixel data of all the pixels of the original image by the conversion coefficients Kr, Kg, Kb calculated in s14 (s15). As a result, the color tone of the entire image is corrected to a natural state so that the pixel data of each of the RGB colors of the originally white pixels match.

After the CPU 11 selectively executes the processes of s13 to s15, it performs the γ correction process in s16 to s18. The γ correction is effective as a correction method when the density histogram peak is in a low value range in the image after the density histogram expansion process in s9 to s11 and the entire image is dark. In this gamma correction, generally,
The input signal X is corrected to the output signal Y by the correction coefficient γ according to the following Expression 6.

[0035]

(Equation 4) As is apparent from Equation 6, when the value of the correction coefficient γ is smaller than 1, the output signal y becomes larger than the input signal x, and when the value of the correction coefficient γ is larger than 1, the output signal y becomes It becomes smaller than the input signal x. This relationship is shown in FIG.

The CPU 11 determines γ in s16 to s18.
In the correction process, the pixel data of each pixel is used as the input signal X in Expression 6, and a gradation conversion table is created using the set correction coefficient γ. Tone conversion table ta for each color of RGB
The array of bleR [], tableG [], and tableB [] has the same number of elements as the resolution of the range of each color of RGB, and the value of each element of the array is a value that can be obtained in the range of RGB (here, In this case, “0” to “255”) are substituted into i.

[0037]

(Equation 5) The CPU 11 converts the values of the pixel data r, g, and b of each color of RGB for each pixel using the gradation conversion table created by the above equation (7). In this example, the same correction coefficient γ is used for each color of RGB. Therefore,
The pixel data r ′ of each color of RGB after gradation conversion
g ′ and b ′ are represented by the following equation 8.

[0038]

(Equation 6) The CPU 11 temporarily stores the pixel data of each pixel after the γ correction processing in s16 to s18 in the image memory 15 as image data after the correction processing (s19),
The image data is output to the image forming apparatus 3 via the image data output unit 16 in response to a predetermined image data output request (s20).

By the above processing, the pixel data of each pixel of the original image is subjected to the improvement of the contrast by the expansion processing of the density histogram, the first white balance correction, and the correction of the halftone brightness by the γ correction processing. The second white balance correction is performed by equalizing the pixel data of each color of RGB in the white pixels according to the characteristics of the original image. In other words, only for an image in which the color tone of the original image cannot be sufficiently corrected only by the first white balance correction by expanding the density histogram, the second white balance by uniforming the pixel data of each color of RGB in the white pixels For an image whose color tone can be sufficiently corrected only by the first white balance correction based on the expansion of the density histogram, the second white balance correction is not performed, and the shortest processing time according to the characteristics of the original image is performed. The color tone of the original image can always be corrected to an appropriate state.

If the subject located in the center of the original image is an image in a backlight state, only the central area is set as an object of the image correction processing by the processing of s5 and s6. This makes it possible to reduce the range to be subjected to the image correction processing according to the characteristics of the original image, and to further reduce the time required for the image correction processing.

[0041]

According to the first aspect of the present invention, for a color original image whose color tone cannot be properly corrected only by the expansion of the density histogram, the white pixels in the color original image are expanded after the expansion of the density histogram. By converting all pixel data based on the color tone of, for example,
Despite the fact that the density histograms of the RGB colors are distributed over substantially the entire range of the value range, a sufficient change in the color tone cannot be given even if the density histogram is expanded, as in a color original image having a color tone bias. For the image, another white balance correction based on the color tone of the white pixel can be executed, and the color tone bias in the original color image can be corrected reliably.

According to the second aspect of the present invention, a color original image in which the color tone is biased to one of the RGB colors and the maximum value of each of the density histograms of the RGB colors exceeds the reference value is obtained. It is determined that the image cannot provide a sufficient change in color tone only by the expansion of the density histogram, and a second white balance correction process is performed on this image to provide a sufficient change in color tone by expanding the density histogram. For images that can be used, the second white balance correction is not executed with weighting, and the color tone can always be properly corrected in the shortest time regardless of the characteristics of the image.

According to the third aspect of the present invention, white pixels are extracted from a color original image in which a sufficient change in color tone cannot be given even if the density histogram is expanded, and the RGB values of the extracted white pixels are extracted. The average value of the pixel data is calculated, and the pixel data of each color of RGB in all the pixels of the color original image is converted by the conversion coefficient determined so that the average value of the image data of each color matches the average luminance of the color original image. Thus, for a color original image in which a sufficient change in color tone cannot be given even when the density histogram is expanded, the color tone of the entire color original image is adjusted so that white pixels become whiter in a range where the luminance of the color original image is not changed. Correction can be made appropriately.

According to the fourth aspect of the present invention, based on the result of comparing the average luminance of the pixels included in the peripheral portion of the color original image with the average luminance of the pixels included in the central portion, the second image is obtained.
By changing the range of the white balance correction, the color tone at the center is preferentially corrected, such as when the image contains a backlight, and when the entire image is Can be accurately distinguished from the case where the correction is made, and when the color tone in the center portion is corrected with priority, the correction is not performed on the pixels in the peripheral portion.
The correction processing time can be reduced.

According to the fifth aspect of the invention, the brightness of the color original image after the second white balance correction is corrected by the γ correction to improve the contrast. White balance correction and second
It is possible to further correct the brightness of the image data in which the contrast and the color tone are made appropriate by the white balance correction, and output an image in a more appropriate state.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of an image processing system to which an image correction method according to an embodiment of the present invention is applied.

FIG. 2 is a flowchart showing a processing procedure of the image correction method.

FIG. 3 is a diagram showing a division state of an original image in the image correction method.

FIG. 4 is a view for explaining improvement of contrast and expansion of a density histogram for the first white balance correction in the image correction method.

FIG. 5 is a diagram illustrating a conversion formula of pixel data used for expanding the density histogram.

FIG. 6 is a diagram illustrating an example of a density histogram of an image to be subjected to a first white balance correction by expanding the density histogram.

FIG. 7 is a diagram illustrating an example of a density histogram of an original image to be subjected to a process of equalizing pixel data of white pixels for a second white balance correction in the image correction method.

FIG. 8 is a diagram illustrating a correction coefficient γ used for γ correction in the image correction method.

[Explanation of symbols]

 1-Image Correction Device 2-Image Output Device 3-Image Forming Device 11-CPU 12-ROM 13-RAM 14-Image Data Input Unit 15-Image Memory 16-Image Data Output Unit

Continued on front page F-term (reference) 5C079 HB01 HB04 LA12 LA23 LA31 MA02 MA04 MA11 NA01

Claims (5)

[Claims] In an image correction method for improving contrast and expanding a density histogram for a first white balance correction on a color original image, a color tone cannot be properly corrected only by expanding the density histogram. An image correction method, wherein a second white balance correction for converting all pixel data based on a color tone of a white pixel in the color original image is performed on the color original image after expanding the density histogram. 2. Prior to the second white balance correction, the maximum value of the density histogram of each color of RGB in the color original image is compared with a reference value, and the color tone is biased toward any one of RGB colors. , RGB if the maximum value of the density histogram of each color exceeds the reference value.
2. The image correction method according to claim 1, wherein the white balance correction is performed. 3. In the second white balance correction, a white pixel is extracted from a color original image, and the R of the white pixel is extracted.
The average value of the pixel data of each color of GB is calculated, and R of the white pixel is calculated.
A conversion coefficient is calculated for each of the RGB colors in which all of the average values of the pixel data of each of the RGB colors match the average luminance of the color original image, and the pixel data of each of the RGB colors of all the pixels of the color original image are multiplied by the conversion coefficient. The image correction method according to claim 1. 4. In the second white balance correction, an average luminance of a pixel included in a peripheral portion of the color original image is compared with an average luminance of a pixel included in a central portion of the color original image. The image correction method according to claim 1, wherein: 5. After the second white balance correction, γ
The image correction method according to any one of claims 1 to 4, wherein gradation conversion by correction is performed.