JP2003250164A - Digital camera - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a digital camera capable of properly adjusting white bal ance. <P>SOLUTION: An optimum gain for white balance adjustment is found on the basis of R data, G data, and B data the white balance of which is adjusted according to a reference gain. In this case, a CPU 30 first obtains a color evaluation value for a plurality of division areas forming an object image and selects a narrower variable range of the optimum gain with respect to the reference gain as the variance of the color evaluation values is greater. Then the CPU 30 discriminates whether or not an optimum gain tentatively determined on the basis of a plurality of the color evaluation value is included in the variable range and corrects the optimum gain when the tentatively determined optimum gain is deviated from the variable range. Thus, as a plurality of the color evaluation values are greatly dispersed, the obtained optimum gain indicates a value closer to the reference gain. <P>COPYRIGHT: (C)2003,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a digital camera, and more particularly to a digital camera for obtaining an optimum gain of white balance adjustment based on a subject image signal whose white balance is adjusted according to a reference gain.

[0002]

2. Description of the Related Art In a conventional digital camera of this type, a color evaluation value is obtained for each of a plurality of regions forming a subject image, and among the plurality of obtained color evaluation values, the color evaluation values included in the pull-in range are calculated. The average value was calculated, and the optimum gain was determined so that the average value converged on the intersection of the RY axis and the BY axis.

[0003]

However, the white balance adjustment based on the average value becomes unsuitable as the obtained variance of the plurality of color evaluation values is large. This is due to the difference in importance between the plurality of colors of the subject. For example, when the main subject is a person, the skin color of this person is more important than the colors of the surrounding subjects. Then, even if there is a large difference between the average value and each color evaluation value, if the optimum gain is determined so that the average value converges on the intersection, there is a possibility that appropriate white balance adjustment will not be performed.

Therefore, a main object of the present invention is to provide a digital camera capable of properly adjusting the white balance even when the variance of a plurality of color evaluation values is large.

[0005]

SUMMARY OF THE INVENTION According to the present invention, in a digital camera for obtaining an optimum gain for white balance adjustment based on a subject image signal whose white balance has been adjusted in accordance with a reference gain, a color for evaluating colors of a plurality of parts of a subject. Variable range changing means for narrowing the variable range of the optimum gain with respect to the reference gain as the variance of the plurality of color evaluation values obtained by the evaluation means and the color evaluation means is large, and the optimum gain is provisionally determined based on the plurality of color evaluation values. A digital camera, comprising: a tentative determination unit; and a correction unit that corrects the optimum gain provisionally determined by the provisional determination unit when the optimum gain deviates from a variable range.

[0006]

The optimum gain for white balance adjustment is obtained based on the subject image signal whose white balance is adjusted according to the reference gain. At this time, the color evaluation means first evaluates the colors of a plurality of parts of the subject to obtain a plurality of color evaluation values. The variable range changing unit narrows the variable range of the optimum gain with respect to the reference gain as the variance of the plurality of color evaluation values increases. The optimum gain is provisionally determined by the provisional determination means based on the plurality of color evaluation values. Then, when the temporarily determined optimum gain is out of the variable range, the optimum gain is corrected by the correction means. Therefore, the greater the variance of the plurality of color evaluation values, the closer the optimum gain obtained is to the reference gain.

The variable range changing means preferably sets the range related to the distance from the color region to the reference point as the variable range when the plurality of color evaluation values are dispersed in the specific color region and the other color regions. Set. More preferably, when the number of color evaluation values belonging to the specific area is equal to or less than the threshold value, the variable range changing unit is disabled by the disabling unit. If the region including the skin color is set as the specific region, the change in the skin color is suppressed for the subject including the skin color, and the white balance adjustment is performed for the subject not including the skin color regardless of the variance of the color evaluation values.

The color evaluation means preferably integrates the color information signal of the object in each of the plurality of parts, and removes the brightness-related component from the plurality of integrated values thus obtained. In this way, a plurality of color evaluation values are obtained.

The reference gain is preferably the optimum gain for a subject having a reference color temperature.

The subject image signal whose white balance is adjusted according to the optimum gain is preferably recorded on the recording medium by the recording means.

[0011]

According to the present invention, the more the plurality of color evaluation values are dispersed, the narrower the variable range is, and the optimum gain is determined so as to fall within the variable range. Here, the variable range is a range that the optimum gain can take with respect to the reference gain, and by narrowing this range, the optimum gain does not change to an unintended value. Thereby, the white balance can be adjusted more appropriately.

The above-mentioned objects, other objects, features and advantages of the present invention will become more apparent from the detailed description of the embodiments below with reference to the drawings.

[0013]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT Referring to FIG. 1, a digital camera 10 of this embodiment includes a lens 12. The light image of the subject is incident on the light receiving surface of the image sensor 16 through the lens 12 and the primary color filter 14 mounted on the front surface of the image sensor 16. As shown in FIG. 2, the primary color filter 14 has R, G, and B filter elements corresponding to respective pixels. Therefore, the amount of charges generated by each light receiving element formed on the light receiving surface reflects the amount of R, G, or B light.

The timing generator (TG) 34 is
The charges (pixel signals) generated by the image sensor 16 are read by the raster scan method. The read pixel signal is converted into a digital signal, that is, pixel data by the A / D converter 18, and the converted pixel data is input to the first signal processing circuit 20. The first signal processing circuit 20 is
Signal processing such as CDS or AGC is performed on the input pixel data, and color separation is performed on the pixel data subjected to such processing. As a result, each pixel has all the color information of R, G, and B, and the R data, G data, and B data of the same pixel are stored in the first signal processing circuit 20.
Are simultaneously output to the white balance adjusting circuit 22.

The R data is given Rgain by the amplifier 24a, and the B data is given Bgain by the amplifier 24b. On the other hand, no gain is given to the G data. The R data and B data to which the gains have been added by the amplifiers 24a and 24b and the G data output from the first signal processing circuit 20 are subjected to a matrix operation by a matrix circuit 26. As a result, Y data, U data and V data are generated.

The R data, G data and B data are also input to the integrating circuit 28. As shown in FIG. 7, the screen is divided into eight in the horizontal and vertical directions, and 64 divided areas are formed on the screen. Therefore, the integrating circuit 28 integrates the R data, G data, and B data for each color information and for each divided area. As a result, the integrated values Rij, Gij and Bij of each divided area are generated over one frame period. It should be noted that i and j respectively indicate the positions of the divided areas in the horizontal direction and the vertical direction, and are assigned any one of the values “1” to “8”.

When the shutter button 32 is operated, the CPU 30 pre-exposes the image sensor 16, and white balances based on the integrated values Rij, Gij and Bij output from the integration circuit 28 based on the pre-exposure. After the adjustment and after the white balance adjustment is completed, the image sensor 16 is subjected to main exposure. The Y data, U data and V data based on the pixel signal obtained by the main exposure are
The recording medium 4 undergoes compression processing by the second signal processing circuit 38.
It is recorded at 0.

When the white balance adjustment is performed in response to the operation of the shutter button 32, the CPU 30 processes the flowcharts shown in FIGS. The control program corresponding to this flow chart is stored in the ROM 42.

First, at step S1, the counters 30a, 30
b, 30c, 30i and 30j and registers 30r,
Initialize 30g and 30b. Count value ACNT of counter 30a, count value BC of counter 30b
NT, the count value CCNT of the counter 30c, the count value i of the counter 30i, and the count value j of the counter 30j are set to "0", and the register value ΣR of the register 30r, the register value ΣG of the register 30g, and the register value ΣB of the register 30b. Is also set to "0".

When the shutter button 32 is operated, pre-exposure is first executed, and R obtained by the pre-exposure is executed.
The data, G data and B data are subjected to white balance adjustment according to the initial value of Rgain and the initial value of Bgain. Integration values Rij, Gij, and Bi of R data, G data, and B data subjected to such white balance adjustment
j is obtained by the integrating circuit 28.

The initial values of Rgain and Bgain are
This is a value at which the white balance is appropriately adjusted when a subject having a color temperature (reference color temperature) of 5100K is photographed. Since there is an error in the characteristics of each manufactured image sensor, the initial values of Rgain and Bgain are calculated individually when manufacturing each digital camera. When such initial values of Rgain and Bgain are set in the amplifiers 24a and 24b and the same subject is photographed, the R data and B data output from the amplifiers 24a and 24b match each other between the digital cameras.

In step S3, Rgain and Bg
Integrated values Rij, Gij of R data, G data, and B data subjected to white balance adjustment according to the initial value of ain
And Bij, the integrated values Rij, Gij, and Bi of the divided areas specified by the count values i and j.
Take in j. In step S5, Yij, (R- is calculated based on the integrated values Rij, Gij, and Bij thus fetched.
Calculate G) ij and (BG) ij. Yij indicates the integrated value of the Y signal detected from the divided area (i, j), and can be defined as the brightness evaluation value. On the other hand, (R-
G) ij indicates the difference between the integrated values of the R signal and the G signal detected from the divided area (i, j), and (B−G) ij is the B signal and G detected from the divided area (i, j). The difference of the integrated value of a signal is shown.

In the following step S7, fy is calculated according to the equation 1.
(RG) ij and fy (BG) ij are obtained.

[0024]

Fy (R−G) ij = (R−G) ij / k1 * Yij fy (B−G) ij = (B−G) ij / k2 * Yij k1, k2: constants (R−G) The numerical values indicated by ij and (BG) ij include
The exposure amount at the time of pre-exposure is reflected. That is, when the exposure amount is large, the numerical value is large, but when the exposure amount is small, the numerical value is small. If (R−G) ij and (B−G) ij having such characteristics are defined as the color evaluation value of each divided area, the color evaluation value varies depending on the exposure amount. On the other hand, the color of the subject does not inherently depend on the exposure amount, and the color of the subject is always the same unless the subject and the light source change. Therefore, even if the exposure amount is changed, the color evaluation value should keep the same value. Therefore, according to Equation 1, (RG)
ij and (B−G) ij are each related to the exposure dose k
It is divided by 1 * Yij and k2 * Yij, and the division values fy (RG) ij and fy (BG) ij are defined as color evaluation values. As a result, the color evaluation value does not depend on the exposure amount, and the color of the subject can be evaluated accurately.

In step S9, the color evaluation value fy (R-
It is determined where G) ij and fy (BG) ij are located in the color temperature diagram shown in FIG. 8B, and the weighting coefficient Wij corresponding to the determined position is shown in the table shown in FIG. 8A. It detects from 30t. In the following step S11, the detected weighting coefficient Wij is multiplied by the integral values Rij, Gij and Bij fetched in step S3, and the multiplication values Rij * Wij, Gij * Wij and Bij * W are obtained.
Find ij.

According to FIG. 8A, the weighting coefficient Wij
Indicates any value of “0”, “6”, “7” and “8”. Of these, the areas to which “6”, “7”, and “8” are assigned define the pull-in range, and the divided area whose color evaluation value is outside the pull-in range is the step S11.
Is invalidated by the processing of. When focusing on the subject image shown in FIG. 7, the 14 divided areas indicated by diagonal lines are the blue sky of the same color, and the color evaluation value is out of the pull-in range indicated by the diagonal lines in FIG. 8A. Therefore, the 14 divided areas are invalidated and the remaining 50 divided areas are activated.

In step S13, the weighted values Rij * Wij, Gij * Wij and Bij * Wij calculated in step S11 are integrated for each color information. Specifically, weighting values Rij * Wij, Gij * Wij and Bij * W are added to the current register values ΣR, ΣG and ΣB.
Add ij.

In step S15, the value indicated by the weighting value Wij detected in step S9 is determined. Where W
When ij indicates a numerical value other than "0", the divided area of interest is regarded as an effective divided area, and steps S17 to S2 are performed.
After step 7, the process proceeds to step S29. On the other hand, when Wij indicates “0”, the divided area of interest is regarded as an invalid divided area, and the process directly proceeds to step S29.

In steps S17, S19 and S21, it is determined which of the areas A to C shown in FIG. 9 the color evaluation value obtained in step S7 belongs to. The area surrounded by the alternate long and short dash line in FIG. 9 corresponds to the area to which “6”, “7” or “8” is assigned in FIG. 8A, and this area also defines the pull-in range. Area A
To C are formed within this pull-in range. Area A is an area where the light source color of daylight fluorescent lamp, day white fluorescent lamp or outdoor light is distributed, area B is an area where the light source color of white fluorescent lamp is distributed, and area C is the light source color of incandescent lamp. This is the area of distribution. The skin color belongs to the area C.

When the color evaluation value belongs to the area A, YES is determined in the step S17, and the counter 3 is determined in a step S23.
The count value ACNT of 0a is incremented. When the color evaluation value belongs to the region B, YES is determined in the step S19, and the count value BCNT of the counter 30b is incremented in a step S25. When the color evaluation value belongs to the area C, YES is determined in the step S21, and the count value CCNT of the counter 30c is incremented in a step S27. When the process of step S23, S25 or S27 is completed, the process proceeds to step S29. If the color evaluation value does not belong to any of the areas A to C, the process directly proceeds to step S29.

In step S29, it is determined whether or not the processes of steps S3 to S29 have been completed for all the divided areas. If NO is determined here, at least one of the counters 30i and 30j is updated in step S31 and then the process returns to step S3, but if YES is determined, the process proceeds to step S33.

Therefore, the count values ACNT, BCNT and C at the time when YES is determined in step S29.
The CNT indicates the distribution state of the color evaluation values of the effective divided area. Further, the register values ΣR, ΣG, and ΣB at the time when YES is determined in step S29 indicate the weighted sum of the integrated values Rij, Gij, and Bij of the effective divided area.

In step S33, Rga is calculated according to the equation 2.
The optimum values (optimal gain) of in and Bgain are provisionally determined.

[0034]

## EQU00002 ## Rgain = k3 * .SIGMA.G / .SIGMA.R Bgain = k4 * .SIGMA.G / .SIGMA.B k3, k4: According to the constant number 2, dividing the register value .SIGMA.G by the register value .SIGMA.R and multiplying the divided value by the constant k3. Then, the gain Rgain is obtained. Further, the gain Bgain is obtained by dividing the register value ΣG by the register value ΣB and multiplying the divided value by the constant k4.

When the optimum gain is provisionally determined, the range in which the optimum gain can change with respect to the reference gain (variable range)
In order to determine, the processing of steps S35 to S45 is executed.

First, in step S35, RgainMAX is set.
(+) And RgainMAX (-) to R_GAIN
(4) and R_GAIN (3), set Bgain
MAX (+) and BgainMAX (-) to B_GA
Set to IN (3) and B_GAIN (4). Rg
ainMAX (+) is the maximum value in the positive direction that Rgain can take, and RgainMAX (−) is Rgain.
Is the maximum negative value that can be taken. Also, Bga
inMAX (+) is the maximum value in the positive direction that Bgain can take, and BgainMAX (-) is the maximum value in the negative direction that Bgain can take. by this,
The variable range is initialized.

As shown in FIG. 9, the upward direction is the plus direction for the RG axis, and the right direction is the plus direction for the BG axis. Therefore, the vector from the second quadrant of the plane shown in FIG. 9 toward the origin is represented by the negative polarity Rgain and the positive polarity Bgain, and the vector from the fourth quadrant of the plane toward the origin is the positive polarity Rgain and the negative polarity. And Bgain.

In steps S37 to S45, the variable range is changed based on the dispersion state of the color evaluation values. In particular,
The count values ACNT, BCNT, and CCNT are set to the threshold value n.
1, n1 and n3, if necessary, Rgain
Change MAX (-) and BgainMAX (+). It should be noted that the threshold values n1 to n3 all indicate values larger than “0” and smaller than “64”.

When the conditions of ACNT> n1 and CCNT> n3 are satisfied, steps S39 to S39.
45, go to RgainMAX (-) and Bgain
MAX (+) to R_GAIN (1) and B_GAIN
Set to (1). Also, ACNT ≦ n1, BCNT>
When the conditions of n2 and CCNT> n3 are satisfied, the process proceeds from step S41 to step S43, and RgainMAX
(-) And BgainMAX (+) to R_GAIN
(2) and B_GAIN (2).

Note that Rga set in step S45
in (1) and Bgain (1) are related to the distance from the position of the area A to the origin, and Rgain (2) and Bgain (2) set in step S43 are related to the distance from the position of the area B to the origin. To do.

On the other hand, the condition of CCNT> n3 is satisfied, or ACNT ≦ n1, BCNT> n2 and C
When the condition of CNT> n3 is satisfied, YES is determined in step S37 or S41, and the process proceeds to step S47 without changing the variable range.

That is, when many color evaluation values are included in the areas A and C shown in FIG. 9, a subject including human skin is photographed under a daylight fluorescent lamp, a neutral white fluorescent lamp or outdoor light, and It is determined that the color of B belongs to the region C, and the vector amount from the second quadrant to the origin is corrected to R_GAIN (1) and B_GAIN (1).

When the color evaluation values are large in areas B and C shown in FIG. 9, an object including human skin is photographed under a white fluorescent lamp, and the color of human skin belongs to area C. That the vector amount from the second quadrant to the origin is R_
Modify to GAIN (2) and B_GAIN (2).

Further, when the distribution of color evaluation values is biased to the area C shown in FIG. 9, an object including human skin is photographed under an incandescent lamp, and most of the color of the object belongs to the area C. It is determined that the variable range that has been set is enabled, and the variable range set by default is activated. That is, when most of the color of the subject belongs to the region C, it is impossible to distinguish the flesh color and the light source color.
Activate the default variable range.

The initially set variable range is validated even when the color evaluation value hardly exists in the area C shown in FIG. This is to avoid a situation in which the variable range is limited even when shooting a subject that does not include a skin color such as vegetation.

In steps S47 to S53, the optimum value of Rgain which is provisionally determined is corrected as necessary, and then, in step S47.
In 55 to S61, the optimal value of Bgain that is provisionally determined is corrected as necessary.

First, in steps S47 and S49, Rga
in to RgainMAX (-) and RgainMAX
Compare with (+). And Rgain <RgainM
If the condition of AX (-) is established, YE is determined in step S47.
It is determined to be S, and Rgain is set to Rgain in step S51.
Correct to MAX (-). In addition, Rgain> Rgai
If the condition of nMAX (+) is satisfied, YES is determined in the step S49, and Rgain is set to Rga in a step S53.
Correct to inMAX (+). When the correction is completed, the process proceeds to step S55. On the other hand, RgainMAX (−) ≦ R
When the condition of gain ≦ RgainMAX (+) is satisfied, NO is determined in both steps S47 and S49, and the process proceeds to step S55 without modifying Rgain.

In steps S55 and S57, Bga
in to BgainMAX (-) and BgainMAX
Compare with (+). And Bgain <BgainM
If the condition of AX (-) is established, YE is determined in step S55.
Sgain is determined, and Bgain is set to Bgain in step S59.
Correct to MAX (-). Also, Bgain> Bgai
If the condition of nMAX (+) is satisfied, YES is determined in the step S57, and Bgain is set to Bga in a step S61.
Correct to inMAX (+). When the correction is completed, the process proceeds to step S63. On the other hand, BgainMAX (−) ≦ B
When the condition of gain ≦ BgainMAX (+) is satisfied, NO is determined in both steps S55 and S57, and the process proceeds to step S63 without modifying Bgain.

In step S63, the Rgain and Bgain corrected as necessary are stored in the amplifier 2 shown in FIG.
4a and 24b. When the setting is completed, the process returns to the upper hierarchy routine.

As can be seen from the above description, the optimum gain for white balance adjustment is obtained based on the R data, G data and B data whose white balance has been adjusted according to the reference gain. At this time, the CPU 30 first obtains the color evaluation values of the plurality of divided areas forming the subject image, and sets the variable range of the optimum gain to the reference gain narrower as the variance of the color evaluation values increases. CPU30 continues,
Whether or not the optimum gain provisionally determined based on a plurality of color evaluation values is included in the variable range is determined, and when the variable gain is out of the variable range, the optimum gain is corrected. Therefore, the greater the variance of the plurality of color evaluation values, the closer the optimum gain obtained is to the reference gain.

In setting the variable range, especially area A
The distribution to C and C or the distribution to regions B and C is noted. This is because the skin color belongs to the area C.
Then, when the distribution to the areas A and C is significant, the variable range is limited to the extent that the color evaluation values belonging to the area A can be moved to the vicinity of the origin. The variable range is limited to the extent that the color evaluation value to which it belongs can be moved near the origin. This makes it possible to prevent the skin color from changing significantly.

When the color evaluation value hardly belongs to the area C, the variable range is not limited and the variable range (maximum variable range) in the initialized state is validated. As a result, the white balance adjustment is performed for the subject that does not include the skin color, regardless of the variance of the color evaluation values.

As described above, the more the plurality of color evaluation values are dispersed, the narrower the variable range is, and the optimum gain is determined so as to be within the variable range. Here, the variable range is a range that the optimum gain can take with respect to the reference gain, and by narrowing this range, a situation in which the optimum gain is largely changed to an unintended value is avoided. Thereby, the white balance can be adjusted more appropriately.

In this embodiment, the variable range is limited when the color evaluation values are widely dispersed.
Alternatively, the gain variation may be prohibited. That is, the reference gain may be determined as the optimum gain when the color evaluation values are dispersed in the areas A and C shown in FIG. 9 or in the areas B and C shown in FIG.

In this embodiment, three types of variable ranges are prepared, but if the number of types is increased, more appropriate white balance adjustment can be performed.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention.

FIG. 2 is an illustrative view showing a primary color filter.

FIG. 3 is a flowchart showing a part of the operation of the CPU.

FIG. 4 is a flowchart showing another portion of the operation of the CPU.

FIG. 5 is a flowchart showing another portion of the operation of the CPU.

FIG. 6 is a flowchart showing yet another portion of behavior of the CPU.

FIG. 7 is an illustrative view showing divided areas formed on a screen.

FIG. 8A is an illustrative view showing a weighting coefficient and a pull-in range, and FIG. 8B is an illustrative view showing a plane formed by the BG axis and the RG axis.

FIG. 9 is an illustrative view showing a pull-in range formed on a plane formed by a B-G axis and an R-G axis.

[Explanation of symbols]

10 ... Digital camera 20 ... First signal processing circuit 22 ... White balance adjustment circuit 24a, 24b ... Amplifier 26 ... Matrix circuit 28 ... Integrating circuit 30 ... CPU 38 ... Second signal processing circuit 40 ... Recording medium

Claims (7)

[Claims] 1. A digital camera that obtains an optimum gain for white balance adjustment based on a subject image signal whose white balance has been adjusted according to a reference gain. Color evaluation means for evaluating colors of a plurality of portions of a subject; Variable range changing means for narrowing the variable range of the optimum gain with respect to the reference gain as the variance of the obtained plurality of color evaluation values is larger, and provisional determining means for tentatively determining the optimum gain based on the plurality of color evaluation values. , And a correction means for correcting the optimum gain provisionally determined by the provisional determination means when the optimum gain deviates from the variable range. 2. The variable range changing means sets, when the plurality of color evaluation values are dispersed in a specific color area and other color areas, a range related to a distance from the color area to a reference point as the variable range. The digital camera according to claim 1, wherein the digital camera is set as. 3. The digital camera according to claim 2, further comprising a disabling unit that disables the variable range changing unit when the number of color evaluation values belonging to the specific area is equal to or less than a threshold value. 4. The specific color area is an area including a skin color.
The digital camera according to claim 2 or 3. 5. The color evaluation means removes a brightness-related component from the integration means for integrating the color information signal of the subject in each of the plurality of parts, and the plurality of integrated values obtained by the integration means. The digital camera according to claim 1, further comprising a removing unit that obtains the plurality of color evaluation values. 6. The digital camera according to claim 1, wherein the reference gain is an optimum gain for a subject having a reference color temperature. 7. The digital camera according to claim 1, further comprising recording means for recording a subject image signal whose white balance is adjusted according to the optimum gain on a recording medium.