JPH0568259A - White balance correcting device - Google Patents

Abstract

PURPOSE:To provide a white balance correcting device which can faithfully- reproduce a color without regard to the kind of a photographing light source. CONSTITUTION:The device is provided white a light source information output means 3 outputting information of the photographing light source and a neural net 4 which consists of an input layer consisting of plural units to be connected to the output of the photographing light source information and the input color image information of the light source information output means 3, an intermediate layer consisting of plural units mutually connected to the respective units of the input layer and an output layer consisting of plural units mutually connected to the respective units of the intermediate layer and outputting the white- balance-corrected color image information of an input color image.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a white balance correction device suitable for use in a color image output device such as a color video printer or a color video camera.

[0002]

2. Description of the Related Art Conventionally, as a white balance correction apparatus of this type, there is one shown in FIG. 4, for example.

In the color video camera, the light source determination for determining the photographing light source and the color signal conversion according to the light source are automatically performed. As a light source determination method, a method (sensor method) of directly measuring the color temperature of the photographing light source shown in FIG. 4A and a method of FIG. 4B are shown.
A method for estimating the color temperature of a light source from a captured image (TTL
Method).

In both types, CC is used through the imaging lens 41.
The image information is read at D42, and the image information is converted into RGB color signals by the signal processing circuit 43. Then, the read color signal is subjected to white balance correction based on the color temperature of the photographing light source, and the corrected color signal is converted into a color difference by the matrix 45.

In the sensor system, the photographing lens 41
Separately, the photometric sensor 46 detects color temperature information of the photographing light source, and the color temperature calculated by the color temperature calculation circuit 47 is input to the variable gain amplifier 44. On the other hand, in the TTL method,
In the color temperature detection circuit 48, the color difference signals RY and BY
The color temperature calculated based on the averaged voltage is input to the variable gain amplifier 44. In both methods, the conversion of the color signal in the variable gain amplifier 44 is performed by R, so that the light source color becomes neutral gray (R = G = B) from the detected color temperature.
Adjusting B gain.

Therefore, under a light source of color temperature K ₁ ,
The colors R ₀ , G ₀ , B ₀ in the standard light source of an object having the colors R ₁ , G ₁ , B ₁ are

[0007]

[Equation 1]

It is represented in the form: Note that fr (K ₁ ), f
b (K ₁ ) is determined so that the light source color becomes neutral gray at the color temperature K ₁ .

Here, 24 color charts from black to fore age (green of leaves) by the Macbeth color checker are obtained when the white balance correction is not performed and when the white balance correction is performed by the above-mentioned conventional method. Use A, D50, D55, D65, D75 5
Tables 1 and 2 show the results of evaluation by CIELAB (referred to as the L ^* a ^* b ^* color space defined by the International Lighting Society) of the color difference from the standard light source under various types of light sources. In addition, the light source A,
D50, D55, D65 and D75 are respectively correlated color temperatures 2856K, 5003K, 5503K, 6504K,
The light source having 7504K (K is Kelvin) is defined by JIS (JIS Z 8720), and D65 is adopted as the standard light source.

[0010]

[Table 1]

[0011]

[Table 2]

From Table 1, when the white balance correction is not performed, particularly in the light source A having a very low color temperature, the average color difference and the color difference dispersion are 40.837 and 4 respectively.
It is 1.965, which means that a large color difference occurs with respect to the standard light source.

On the other hand, when the white balance correction by the conventional method is performed, as shown in Table 2, the color difference with respect to the standard light source is smaller than that in the case where the white balance correction is not performed. ..

However, in this conventional method, the color difference from black to neutral 3.5, neutral 5, neutral 6.5, neutral 8 and white is almost 0, but with a light source other than the standard light source,
Color charts other than the above color charts have not been completely corrected, and especially in the A light source, the average color difference is 7.869 and the color difference dispersion is 9.709, and the white balance correction is insufficient. It was

A and D5 when white balance correction by the above-mentioned conventional method is performed using Munsell color chart
Table 3 shows the results of the CIELAB evaluation of the color difference with the standard light source under the five types of light sources of 0, D55, D65, and D75. The color chart is 10 evenly from the hue circle from red (5 red) to magenta (5 red purple) in the Munsell book.
I chose the individual and gray. The results of Table 3 are the same as those of Table 2,
The average color difference is 11.619 and the color difference dispersion is 1 in the light source.
It was 2.352, and the white balance correction was insufficient.

[0016]

[Table 3]

[0017]

Therefore, in the above-mentioned conventional method, the accuracy of the white balance correction is poor and the color reproduction is unnatural in the room light having a very low color temperature such as the A light source.

The present invention has been made in view of the above circumstances, and when the white balance of an input image is corrected by a color image output device, faithful color reproduction is achieved regardless of the type of photographing light source. An object of the present invention is to provide a white balance correction device that can perform the above operations.

[0019]

SUMMARY OF THE INVENTION The present invention is a color image output device for outputting a color image based on input color image information obtained by photographing a subject, and a light source information output for outputting information of a photographing light source. Means, a photographing light source information of the light source information output means, and an input layer composed of a plurality of units connected to the output of the input color image information, and a plurality of units mutually connected to each unit of the input layer. And a neural network including an output layer including a plurality of units that are mutually coupled to each unit of the intermediate layer and that outputs color image information of the input color image whose white balance is corrected. A white balance correction device characterized in that

[0020]

According to the present invention, even if an image signal that is not white-balanced is input, it is corrected and output so as to be in an ideal white-balanced state regardless of the type of photographing light source.

[0021]

DESCRIPTION OF THE PREFERRED EMBODIMENTS This embodiment will be described below with reference to the drawings showing the embodiments.

FIG. 1 is a schematic block diagram of a video printer showing an embodiment of the present invention. Reference numeral 1 is a CPU (microcomputer), and this CPU 1 controls a head driver control unit 6 and the like. The input color image signal is temporarily stored in the image memory 2. Next, under the control of the CPU 1, the color image signal R, which is stored in the image memory 2,
One screen of G and B is input to the light source color determination circuit 3 which outputs information on the photographing light source, and the color temperature of the photographing light source is calculated by the light source color determination circuit 3 as described later. Then, the calculated color temperature and the color image signal of the image memory 2 are input to the neural network 4 described later, and the corrected image signal is supplied to the line buffer 5. The head driver control circuit 6 controls the corrected image signal for one line stored in the line buffer 5 so as to generate a heating pulse to be supplied to the thermal head 7. The temperature of the thermal head 7 is detected by the temperature detector 8 and the detection result is given to the gradation table ROM 9. By this, the gradation data in the gradation table ROM 9 is corrected and fed back to the heat generation pulse control circuit 10. To be done.

Next, the operation of the light source color judgment circuit 3 will be described with reference to the flowchart of FIG.

First, the RGB values are read from the image memory 2 (S1), the read RGB values are summed up, and the average values Ra, Ga, Ba are calculated (S2). Then, it is determined whether the RGB values for one screen have been read (S3). In this step S3, one screen of RGB
If it is not determined that the values have been read, the process returns to step S1, and if it is determined that the RGB values for one screen have been read, the process proceeds to step S4. Note that the color average values Ra, Ga, and Ba for one screen are calculated in step S3 because these values are almost equal to the color values of the photographing light source.

Then, in the next step S4, the color average value Ra for one screen is calculated by using the color temperature approximation formula showing the relationship between the RGB value and the color temperature determined based on the color temperature locus of the perfect radiator. , Ga, Ba, the color temperature K of the photographing light source is calculated.

As shown in FIG. 3, the neural network 4 has color image signals R, G, B from the image memory 2,
And the input layer 31 to which the color temperature K calculated by the light source color temperature determination circuit 3 is input, and each neuron A of the input layer 31.
The signals from _i (i = 1 to 4) are coupled coefficients W _ij (j = 1 to 1).
An intermediate layer 32 composed of neuron B _j input by weighting 15), signals from each neuron B _j of the intermediate layer 32 by weighting with the coupling coefficient V _jk (k = 1~3) And an output layer 33 composed of input neurons C _k . Each neuron C _k output of the output layer 33 represents the white balance-corrected color image signals R ′, G ′, B ′.

Although the number of neurons in the intermediate layer is 15 in the above embodiment, this is set in consideration of the accuracy of white balance correction, and the accuracy of white balance correction must be sufficient. It need not be specific to this number.

On the other hand, each coupling coefficient W _ij and V of the intermediate layer 32 is
_{The jk} is stored in a ROM (not shown) as a value that is learned in advance and set as will be described later.

The learning of the neural network 4 is carried out by the Macbeth color checker A, D5 in which the color temperature of 24 color charts from black to foreage (green of leaves) is known.
0, D55, D65, and D75 are used to measure the color, and the RGB values of the color chart and the color temperature of the photographing light source are input, and the colorimetric value of the standard light source (D65) is used as the teacher The signal was measured by the backpropagation method. That is,
120 pairs of input and teacher signal pairs were learned.

Tables 4 and 5 show the results of white balance color correction performed by the neural net 4 thus learned. Table 4 shows color differences as a result of color correction with respect to learned data, and Table 5 shows color differences as a result of color correction with respect to unlearned Munsell color chart data.

[0031]

[Table 4]

[0032]

[Table 5]

From these tables, in the case of learning data,
The average color difference and the color difference dispersion in the light source A are 2.0 and 2.0, respectively.
It is 17, 2.557, and it can be seen that the white balance correction is performed accurately. Also, in the case of unlearned data, the average color difference and the color difference variance are 2.336 and 2.961 in the A light source, respectively, and the white balance correction is performed accurately as in the case of the learned data. I understand.

In the present embodiment, the case where the present invention is applied to the color video printer has been described, but the present invention can also be applied to other color video output devices such as a color video camera.

Further, the light source color determination circuit calculates the color temperature information of the photographing light source on the basis of the input color image color signal, but instead of this, the color temperature information is directly detected by using a photometric sensor or the like. Alternatively, the operator may directly input the photographing light source information.

[0036]

The white balance correction apparatus of the present invention is
Light source information output means for outputting information of the photographing light source, photographing light source information of the light source information output means, an input layer composed of a plurality of units connected to output of input color image information, and each unit of the input layer From an intermediate layer composed of a plurality of units coupled to each other, and an output layer composed of a plurality of units coupled to each unit of the intermediate layer and outputting white image-corrected color image information of an input color image. The input color image information that has not been white-balance corrected is corrected and output so that it will be in an ideal white-balance state even if the image is captured under various light sources. To be done.

Therefore, faithful color reproduction is performed even under a photographing light source having a low color temperature, and a good color image can always be obtained.

[Brief description of drawings]

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

FIG. 2 is a diagram showing a flowchart for explaining an operation of a light source color determination circuit in the embodiment of FIG.

FIG. 3 is a configuration diagram of a neural network applied to the embodiment in FIG.

FIG. 4 is a schematic block diagram of a white balance correction device in a conventional color video camera.

[Explanation of symbols]

 1 CPU 2 Image Memory 3 Light Source Color Judgment Circuit (Light Source Information Output Means) 4 Neural Network 31 Input Layer 32 Intermediate Layer 33 Output Layer 41 Photographing Lens 42 CCD 43 Signal Processing Circuit 44 Variable Gain Amplifier 46 Photometric Lens

Claims (3)

[Claims] 1. A color image output device for outputting a color image based on input color image information obtained by photographing a subject, and a light source information output means for outputting information of a photographing light source, and the light source information output means. Of the photographing light source information, and an input layer composed of a plurality of units connected to the output of the input color image information, an intermediate layer composed of a plurality of units mutually coupled to each unit of the input layer, and the intermediate layer. , A neural network comprising an output layer composed of a plurality of units which are mutually connected to each other to output color image information of the input color image subjected to white balance correction, and a white balance correction characterized by comprising: apparatus. 2. The back net propeller is the neural net.
2. The white balance correction apparatus according to claim 1, wherein the white balance correction apparatus is a hierarchical neural network that performs learning by a method of division. 3. The input layer includes a plurality of units connected to output color temperature information of a photographing light source and color information of an input color image, and the output layer includes white-balance-corrected colors of the input color image. The white balance correction apparatus according to claim 1 or 2, comprising a plurality of units for outputting information.