JP2000209437A - Image correction method and storage medium storing image correction program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To automatically acquire a correction coefficient optimum to an original image by obtaining a mean luminance of a part especially desirably emphasized in the original image as an evaluated luminance, multiplying the evaluated luminance with a value of a monotonously decreasing function whose variable is the evaluated luminance so as to calculate a mean luminance after correction as an expected luminance, calculating a coefficient given by a curve on the basis of the evaluated luminance and the expected luminance thereby correcting all pixels of the original image. SOLUTION: An image display monitor that is an image output device 9 displays an original image and an image after gamma correction. A floppy disk driver 11 reads a medium on which an image correction program is recorded and a CPU 1 loads the program to a RAM 3 via an interface 10. The original image is read and expanded on a memory as a bit map, the entire image is divided into small areas and a mean luminance of a center area is obtained. Furthermore, the mean luminance of the surrounding areas and the mean luminance of the entire screen are found, the means luminance of the center area is compared with the mean luminance of the surrounding area, and it is decided that an object in the middle of the screen is in a back light state when the center part is darker than the surrounding part.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a gamma correction method for a color / monochrome still image obtained by a digital still camera, an image scanner, a video capture circuit, or the like. The present invention relates to a method of performing γ correction by obtaining a suitable γ correction coefficient.

[0002]

2. Description of the Related Art In order to convert a video signal into a still digital image irrespective of color / black and white, a still image is captured by a digital still camera or image scanner equipped with a solid-state imaging device such as a CCD or C-MOS area sensor. 2. Description of the Related Art A technique of capturing a still image by a video capture board that converts an analog video signal being captured by a video camera or the like into digital data in real time and performing image processing has become widespread.

[0003] These digitized still images are CR
Display on a display device such as T or liquid crystal display,
The image file as digital data is transmitted by any transmission means.

[0004] In recent years, with the spread of the Internet, there are many opportunities to be used as image data to be pasted into multimedia text represented by an HTML document.

On the other hand, devices that capture still images are also
With the price reduction of digital still cameras having CCDs of more than 100,000 pixels, higher quality still images can be easily obtained.

[0006] As the quality and versatility of the means for capturing still images, the means for displaying still images, and the means of application have been increased, there has been a demand for further improving the image quality of still images captured by image capturing devices. Is growing.

In particular, the gradation characteristics of an image are important in determining the image quality, such as the brightness, vividness, and contrast of the image. As an effective method for correcting the gradation, γ correction processing has been conventionally used. Are known.

Originally, γ correction is a method used for the purpose of correcting the luminance characteristics of a cathode ray tube of a CRT display, and the γ curve, which is the input / output characteristic by the γ correction, shows the amplification factor when the input is small. In many cases, the characteristic is increased so that the amplification factor approaches 1 as the input increases. As a result, it is possible to perform tone correction without impairing the dynamic range of the original image, and this is an effective method for providing appropriate tone characteristics regardless of the output device and the method of using the image. .

As a gradation correction method using such γ correction, for example, in Japanese Patent Laid-Open No. Hei 5-76036, a gradation conversion characteristic of a γ curve is given from a setting means, and one of 16 types of γ curves is obtained by referring to a table. A method has been proposed in which a γ curve is selected, γ correction is performed on a luminance signal, and then gradation conversion of a color signal is performed based on a ratio of the luminance signal before and after γ correction.

[0010]

However, in such a conventional tone correction method, it is necessary to specify the shape of the γ curve by another means independent of the image quality characteristics of the original image. In order to select the gradation correction curve, it was necessary to repeat the judgment of the person operating the apparatus and the trial until obtaining an appropriate correction curve.

Further, the γ correction coefficient of the γ curve according to the prior art needs to be selected from several kinds of fixed values prepared in advance, and there is a problem that an arbitrary γ curve optimal for the original image cannot be obtained.

An object of the present invention is to solve the above-mentioned problems and to automatically obtain a γ coefficient for γ correction optimal for an original image, and a program recording medium storing an image correction program. Is to provide.

[0013]

According to the present invention, an average luminance (hereinafter, referred to as an "evaluation luminance value") of a portion to be particularly emphasized in an original image is obtained, and the average luminance is monotonously reduced using the evaluation luminance value as a variable. The calculation result of the function is multiplied by the evaluation luminance value, and the average luminance after the γ correction is calculated as the expected luminance value.
Then, the γ coefficient of the γ curve is calculated from the evaluation luminance value and the expected luminance value, and γ correction is performed using the calculated γ coefficient. For example, a range of values that can be taken by pixel values of an original image (hereinafter, “value range”)
That. The values after γ correction are obtained for all the values in the parentheses, a γ correction table is created, and the γ correction of the pixel values of all the pixels of the original image is performed with reference to the γ correction table.

For example, in the case of an image in which a person is photographed in the center, it is generally desirable that the gradation of the person in the center is corrected to a desired gradation. The average luminance of the area is used as the evaluation luminance value. On the other hand, in an image of a landscape, the entire screen is generally important, and the average luminance of the entire screen is used as the evaluation luminance value with the entire screen as the attention area.

According to the above configuration, since the γ correction coefficient is calculated from the average luminance of the portion of the original image to be emphasized, the gradation characteristic originally possessed by the original image is not largely changed, and It is possible to automatically calculate an appropriate γ correction coefficient for each image. Further, by changing the function formula of the monotonically decreasing function, and the parameters of the function formula, what value the average luminance of the original image has, the highest amplification factor by γ correction, in other words, It is also possible to tune whether to actively increase the brightness by gamma correction when the average luminance of the attention area of the original image is relatively low, or to further increase the brightness when the average luminance of the original image is relatively bright. .

[0016]

FIG. 9 is a block diagram showing the configuration of an image correction apparatus to which the image correction method according to the present invention is applied. In FIG. 9, the CPU 1 executes a program previously written in the ROM 2 and a program read from the hard disk drive device 7 to the RAM 3 to perform a gamma correction process on an original image described later. The RAM 2 is also used as a buffer for temporarily storing image data and a working area during the processing. The image input device 5 is, for example, a digital still camera, a video capture board, an image scanner, or the like.
The image data is read through the hard disk, and given a predetermined file name in the hard disk and stored as an original image file. The hard disk drive device 7 is used to store image files after gamma correction in addition to the above-mentioned programs and original image files, and the CPU 1 writes / reads files via the interface 6. The image output device 9 is, for example, an image display device (monitor), and displays an original image and an image after γ correction. The floppy disk drive 11 reads a medium (floppy disk) on which an image correction program is recorded, and the CPU 1 loads the RAM 3 via the interface 10.

Note that the image input device is not indispensable. For example, an image file may be received using a communication unit (not shown) and stored in the hard disk. Further, the image output device is not indispensable. For example, the purpose is to generate an image file after γ correction, and the image file may be transmitted by a communication unit (not shown).

FIG. 1 is a flowchart showing the procedure of a processing program executed by the CPU of the image correction apparatus.
First, an original image file is read in step s1. This original image file is an image file captured as a digital image by a digital still camera, a video capture board, an image scanner, or the like.
For example, in the case of a still image from a digital still camera,
Windows BMP and JPEG formats. On the Internet, GIF format is often used in addition to JPEG. In recent years, PNG has been used as a network image format instead of GIF.
Are also attracting attention. The image file of such various formats is read by designating the corresponding format.

Next, in step s2, a bit map is developed on the memory, and the vertical size and the horizontal size of the bit map are divided into five, respectively, and the entire screen is divided into 25 small areas. By dividing the image into a plurality of small areas in this way and calculating image attributes such as luminance, RGB average value, IQ value, and color difference signals such as BY and RY for each small area, It predicts whether the image is an image and how to perform image correction according to the image.

FIG. 6 shows an example of the screen division.
The “central 9 areas” are area numbers 6, 7, 8, 11, 12,
13, 16, 17, and 18, and the “peripheral 16 areas” are peripheral areas excluding the central area from the entire screen.

In step s3, the average luminance Yc of the nine central areas is obtained. The average luminance Yc of the central 9 areas is
The luminance of each pixel is determined from the RGB components in accordance with the human visual sensitivity characteristics, and is obtained by [Equation 1], which is integrated for all the pixels in the central 9 areas, and divided by the total number of pixels in the central 9 areas.

Y = 0.3R + 0.59G + 0.11B (Equation 1) Similarly, in step s4, the average luminance Ya of the surrounding 16 areas
In step s5, the average luminance Yf of the entire screen is determined.

In step s6, the above Yc and Ya are compared. If Yc is smaller than Ya, that is, if the central portion is darker than the peripheral portion, it is determined that the subject at the center is backlit, and In order to preferentially correct the gradation of the part, Yc is set as the evaluation luminance value of the original image in step s7. Conversely, Yc
If it is greater than or equal to Ya, it is determined that the image should be given importance to the entire screen such as a landscape or a group photo of people, and Yf is set as the evaluation luminance value of the original image in step s8.

The gamma curve which is the basis of the tone correction according to the present invention is represented by the following equation (2) when the input signal is x, the output signal is y, and the gamma correction coefficient (hereinafter simply referred to as "gamma coefficient") is gamma. ].

Y = x ＾ γ (0 ≦ x ≦ 1) [Formula 2] When γ is smaller than 1, y is larger than x, and when γ is larger than 1, y is smaller than x. There are features. FIG. 8 shows an example of a γ curve when the value of γ is changed by several types in the range of γ <1. As is clear from this figure, it can be seen that the smaller the γ, the higher the halftone amplification factor. In addition, the amplification factor is 1 at the maximum / minimum value range.
Thus, it can be understood that the dynamic range itself of the original image does not change even if γ changes.

Here, a method of deriving the γ coefficient from the evaluation luminance value of the original image will be described. When the evaluation luminance value Y of the original image is obtained, when Y is smaller than the value range of Y,
For example, when the RGB of the original image can take a value of 0 to 255 for each 8 bits, the value of Y can take a value of 0 to 255 similarly. At this time, when the value of Y is a small value such as 20 or 30, it is desirable to increase the luminance by applying a smaller γ coefficient (to increase the luminance amplification factor). On the other hand, when the evaluation luminance value Y is larger than the value range, for example, a value equal to or more than the median value in the value range of 0 to 255, if the luminance amplification rate is increased, the image after γ correction has a higher luminance value. Since the image is biased and the quality of the original image may be impaired, it is desirable to increase the γ coefficient value or to output the input pixel value as it is without performing luminance correction with γ = 1. In this embodiment, the evaluation luminance value Y of the original image is used as a variable for the calculation of the γ coefficient, and the luminance value after the γ correction (hereinafter referred to as “expected luminance value Y ′”) is calculated by the equation shown in [Equation 3]. Ask.

In the equation at the upper stage of [Equation 3], f (Y) is Y
Is a function whose value monotonously decreases from a predetermined value to 1 with respect to an increase from 0 to c. Further, the following expression means that when the evaluation luminance value Y is equal to or more than c, no γ correction is performed.

Y ′ = Y · f (Y) (0 ≦ Y <c) (Equation 3) Y ′ = Y (c ≦ Y <Ymax) Here, as an example of f (Y), FIG. 4] is introduced.

F (Y) = (a + b) exp (−Y / T) −b + 1 (Equation 4) As apparent from FIG. 2, the point (0,1) + a) indicates an exponential function that intersects the vertical axis. In addition, the time constant T of the exponential function shown in [Equation 4] indicates that if the evaluation luminance value is equal to or greater than c, no γ correction is performed. That is, Y (c) and f (Y) = 1.0
By substituting Y = cf (Y) = 1.0 into [Equation 4], the following is derived.

T = c / log {b / (a + b)} (Equation 5) The graphs of the equations shown by [Equation 4] and [Equation 5] are as shown in FIG. FIG. 5 shows a case where the constants a and b are changed by several types. For example, (a = 1.0, b = 0.2
The curve of ()) is relatively close to the straight line Y '= Y, but has a bulge on the most positive side near Y = C / 2. Here, Y ′ is the luminance (expected luminance value) after the γ correction that desires that the average luminance of the attention area of the image obtained as a result of the γ correction of the original image whose evaluation luminance value is Y is Y ′. Above (a = 1.
0, b = 0.2), if the γ correction is performed using the γ coefficient obtained from the relationship between Y and Y ′, the absolute value of the gradation conversion amount is small, but the evaluation luminance value of the original image is c / 2 In the vicinity, the correction effect by the γ correction is the largest, and the image can be made slightly brighter.

In the example of (a = 2.5, b = 0.4), the effect of brightening the image is (a = 1.0, b = 0.2) when the evaluation luminance value of the original image is between 0 and C. The effect of correcting the image to be bright when the evaluation luminance value of the original image is low and the evaluation luminance value of the original image is low is high.

(A = 2.5, b = 4.0) is an example in which a and b are extremely large. Contrary to the example of (a = 2.5, b = 0.4), the evaluation of the original image is The effect of making the image brighter is higher when the luminance value is larger.

As described above, when the value of b is made smaller than the value of a, the effect of brightening the image after the γ correction of the image having a low evaluation luminance value of the original image is enhanced. Further, when the value of b is increased as compared with the value of a, the effect of brightening the image after the γ correction of the image having the high evaluation luminance value of the original image is enhanced.

When the expected luminance value after gamma correction is obtained from the luminance evaluation value of the original image according to the above [Equation 3] and [Equation 5], the graph of the expected luminance value with respect to the evaluation luminance value range is generally shown in FIG.
become that way. As described above, when (0 <Y <c), the expected luminance value can be obtained by giving various characteristics to the parameters a and b in [Equation 5]. Further, by automatically adjusting the value of c, that is, by changing the original image evaluation luminance value for which the γ correction is valid, it is possible to provide a feature to the automatic generation of the γ coefficient.

Next, an example will be described in which a linear expression having the evaluation luminance value Y of the original image as a variable, such as Expression 6, is introduced into f (Y) shown in Expression 3.

F (Y) = − ｛(e−1) / c｝ Y + e (Expression 6) [Expression 6] when the expression of Expression 6 is used for f (Y) shown in Expression 3 3] is shown as a graph in FIG. In this case, the expected luminance value Y ′ after the γ correction is f (Y) · Y = − ｛(e−1) / c｝ Y ² + eY (Equation 7), which is 2 of the evaluation luminance value Y of the original image. It becomes the following function.

FIG. 7 is a diagram showing the relationship between the original image evaluation luminance value Y and the expected luminance value Y 'after the .gamma. As described above, when the evaluation luminance value of the original image is relatively high in comparison, the effect of brightening the image after the γ correction increases.

In the case where the expected luminance value after γ correction is obtained from the luminance evaluation value of the original image according to [Equation 3] and [Equation 6],
A graph of the expected luminance value with respect to the evaluation luminance value range is substantially as shown in FIG. As described above, when (0 <Y <c), the expected luminance value can be obtained by giving various characteristics by the parameter e in [Equation 6]. Further, by adjusting the value of c itself, it is possible to give a feature to the automatic generation of the γ coefficient.

As described above, the expected luminance value Y 'is obtained in step s9 of FIG.

Thereafter, in step s10 of FIG. 1, a γ coefficient that determines the γ correction characteristic is obtained from the evaluation luminance value Y of the original image and the expected luminance value Y ′ obtained in step s9.

Here, it is only necessary that the result of the γ correction of the evaluation luminance value Y be the expected luminance value Y ′. Y ′ = Y ＾ γ (Equation 7) γ = logY ′ / logY (Equation 8) The γ coefficient is obtained from the relationship. Here, Y ′ and Y are 0
It must be normalized within the range of 〜1.

In step s11 of FIG. 1, a table for γ correction is created by using the γ coefficient obtained in step s10. Now, let the array of the gamma correction tables of RGB be tableR [], tableG [], and tableB []. These arrays have the same dynamic range and the same number of elements as the evaluation luminance values. The value of each element of the array can be obtained by substituting each value of the RGB dynamic range into i according to [Equation 9]. In [Equation 9], the RGB dynamic range is set to 8 bits (0
To 255).

TableR [i] = 255 * (i / 255) ＾ γ tableG [i] = 255 * (i / 255) ＾ γ (Equation 9) tableB [i] = 255 * (i / 255) ＾ γ In step s12, the RGB values of all pixels of the original image are converted with reference to the gamma correction table created in step 10 above. In this example, the same gamma correction carp is used for RGB.

The RGB values of one pixel of the original image are
Assuming that r, g, b, the RGB values r ′, g ′, b ′ after the γ correction are obtained as shown in [Equation 10].

R ′ = tableR [r] g ′ = tableG [g] (Equation 10) b ′ = tableB [b] In step s13, the image after the γ correction obtained in step s12 is stored as a file.

[0046]

According to the present invention, an electronic still camera using a solid-state imaging device such as a CCD or a C-MOS area sensor,
In a still image captured from an image scanner, a video capture board, or the like, an expected luminance after γ correction is obtained by referring to a monotone decreasing function such as an exponential function or a linear function using the average luminance of the attention area as a variable. Therefore, it is possible to derive a γ coefficient for γ correction steplessly and automatically. This makes it possible to perform appropriate gamma correction without particularly requiring a gamma curve designating means and without deteriorating the brightness atmosphere of the original image.

[Brief description of the drawings]

FIG. 1 is a flowchart showing a processing procedure in an image correction apparatus according to the present invention.

FIG. 2 is a diagram showing an exponential function for obtaining an expected luminance value from an original image evaluation luminance value.

FIG. 3 is a diagram showing a linear function for obtaining an expected luminance value from an original image evaluation luminance value.

FIG. 4 is a diagram showing a relationship between an original image evaluation luminance value and an expected luminance value according to the present invention.

FIG. 5 is a diagram showing a relationship between an original image evaluation luminance value and an expected luminance value.

FIG. 6 is a diagram showing an example of area division.

FIG. 7 is a diagram showing a relationship between an original image luminance evaluation value and an expected luminance value.

FIG. 8 is a diagram showing an example of a γ correction curve.

FIG. 9 is a diagram illustrating a configuration of an image correction device.

[Explanation of symbols]

 4, 6, 8-interface 7- hard disk drive 11- floppy disk drive

 ──────────────────────────────────────────────────続 き Continued on the front page F term (reference) 5C021 PA63 PA80 RA07 RB03 XA34 5C066 AA00 AA01 BA00 CA08 EC05 GA01 GA02 GB00 KA11 KD06 KE05 KP02 5C077 LL16 MM02 MP08 PP15 PP28 PP32 PP44 PP46 PP68 PQ12 PQ22 PQ23 TT

Claims (4)

[Claims] 1. An average luminance of an attention area of an original image is obtained as an evaluation luminance value, and an expected luminance value after γ correction is obtained from the evaluation luminance value by a monotonic decreasing function using the evaluation luminance value as a variable. An image correction method in which a luminance value is set as an expected luminance value, a γ coefficient is obtained from the evaluation luminance value and the expected luminance value, and a value of each pixel of the original image is γ corrected based on the γ coefficient. 2. The image correction method according to claim 1, wherein the monotone decreasing function includes an exponential function having the evaluation luminance value as a variable. 3. The image correction method according to claim 1, wherein the monotone decreasing function includes a linear function having the evaluation luminance value as a variable. 4. A computer that executes image processing obtains an average luminance of an attention area of an original image as an evaluation luminance value,
From the evaluation luminance value, an expected luminance value after γ correction is obtained by a monotonically decreasing function using the evaluation luminance value as a variable, the luminance value is set as an expected luminance value, and γ is calculated from the evaluation luminance value and the expected luminance value. A recording medium storing a program for obtaining a coefficient and performing an image correction process for performing a γ correction of a value of each pixel of the original image based on the γ coefficient.