JPH09147098A - Method and device for processing color image - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain a high quality image by correcting the white balance of plural input images at a digital image processor. SOLUTION: Concerning all the images in plural input image memories, following processing is performed. Based on image data inputted to an input image buffer 1, white and black RGB values are acquired by a white/black extracting means 2 and inputted to an RGB table preparing means 3. A reference white RGB value is held in a reference white buffer 7 and a reference black RGB value is held in a reference black buffer 8 in advance. Based on the output values from the white/black extracting means 2 and the values of the reference white buffer 7 and reference black buffer 8, the RGB table preparing means 3 generates the respective look-up tables(LUT) of RGB and holds them in an RLUT 4, GLUT 5 and BLUT 6. Based on the data in the RLUT 4, GLUT 5 and BLUT 6, an RGB table transforming means 20 transforms picture element values to a table and outputs it to an output image buffer 12.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a technique for improving the quality of an image in a digital image processing apparatus by correcting the white balance of a plurality of input images and unifying disparate color tones.

[0002]

2. Description of the Related Art Conventionally, in this type of image processing apparatus, in order to correct the white balance of an image, each operator manually corrects the pixel value, and the operator interactively corrects the pixel value. However, for a single image, Japanese Patent Laid-Open No. 62-2810
Some devices, such as the device 70, perform automatic correction based on the maximum brightness value in the image. With reference to FIG.
An example of a conventional image processing apparatus will be described based on -281070.

An example of a conventional device is an image memory 10.
0, maximum brightness value extraction unit 101, correction amount calculation unit 102, correction ROM 103, correction ROM 104, masking circuit 1
It consists of 05.

From the input Y, RY and G-Y signals,
By the maximum brightness value extraction unit 101, the maximum brightness value in the image,
Alternatively, the average value Ymax of 10 points having high brightness is obtained. At the same time, in the correction amount calculation unit 102, the color components of the pixel having the maximum luminance value Ymax, R-Y and G-Y, are obtained as ΔRY and ΔGY, respectively. When Ymax is the luminance average, ΔRY and ΔGY are also averaged.

From the correction amount thus obtained, the RY and GY signals are converted using the following equation.

(R−Y) ′ = (R−Y) ΔRY × Y / Ymax (G−Y) ′ = (G−Y) ΔGY × Y / Ymax (1) Obtained y, (R−Y) Based on the ', (G-Y)' signals,
The color is converted to CMY and output.

[0007]

The conventional image processing apparatus has a problem that it is not possible to perform color correction or white balance correction for a plurality of images in an integrated manner.

In the conventional image processing apparatus, the video signals Y, RY, and GY are used as the input signals, but RGB signals are used in digital image processing on a personal computer, for example. Many, Y, R-
Overflow (of exceeding the RGB dynamic range) often occurs in the process of converting the Y, G-Y signals to RGB, and the colors may be unnatural when displayed on a monitor or the like.

Further, in the conventional image processing apparatus, since the color conversion is performed on the basis of only the information of the high brightness portion, the quality of the low brightness portion cannot be improved.

[0010]

In order to solve the above-mentioned problems, the color image processing method of the present invention comprises the steps of extracting the values of white pixels and black pixels in an image from input image data, and extracting the values. A color that includes a step of creating a conversion table based on a white value, a black value, and a predetermined reference white value and a standard black value, and a step of performing table conversion on all pixels based on the created table. It is characterized in that the correction processing is simultaneously performed on a plurality of input image data.

Further, the color image processing apparatus of the present invention has a plurality of input image memories for holding a plurality of input images, a plurality of output image memories for holding a plurality of output images, and a color correction process for all image data. To control the input image data, an input image buffer that holds the input image data, a white-black extraction unit that extracts the RGB values of white pixels and black pixels from the image, and a position designation unit that designates the target area to be extracted. And a reference white buffer that holds the reference white,
A reference black buffer that holds a reference black, a table creating unit that creates a lookup table for color conversion, and a table conversion for each of RGB.
It is characterized in that it is provided with a B look-up table conversion means and an output image buffer for holding image data after color conversion.

The operation of the image processing apparatus of the present invention will be described.

The processing contents of the present invention will be described with reference to the flowcharts of FIGS. 3 and 4. This is to perform white correction on a plurality of images to be processed in an integrated manner.

The operator first specifies the reference white value and the reference black value, and the system stores them (step S1). A plurality of image data to be white-corrected is held, images are input one by one from the buffer to the buffer (step S2), and the following white correction processing is performed to perform color correction (steps S3 to S5). When the process for one image is completed, the white correction process is sequentially performed for the remaining images by the determination process (step S6). In this way, the processing is completed after processing all the plurality of images to be processed. In the white correction processing, first, RGB values of white and black are extracted from the input image data (step S3). A method for extracting white and black RGB values will be described with reference to FIG.

First, a low pass filter is applied to the image data (step A1). An example of the low pass filter is a local average filter that adopts the average value of neighboring pixels as a new pixel value.

Pixel values at positions i and j are represented by I (i,
j), when a 3 × 3 local average filter is operated, its output I ′ (i, j) is I ′ (i, j) = (I (i−1, j−1) + I ( i, j-1) + I (i + 1, j-1) + I (i-1, j) + I (i, j) + I (i + 1, j) + I (i-1, j + 1) + I (i, j + 1) + I ( i + 1, j + 1)) / 9 (2). If there are three pixel values like RGB,
A local average filter may be applied to each of R, G, and B. The purpose of the low-pass filtering step (step A1) is to reduce noise, but this step A1 is deleted when high-speed processing is performed. Next, a histogram of the brightness of the image is created (step A3). The brightness here refers to a value indicating the brightness of the RGB pixel by a numerical value.
As a method of obtaining a luminance value from RGB values, CIE-XY
There is a method of using the Y value of the Z value. Assuming the RGB values are NTSC signals, Y = 0.299 * R + 0.587 * G + 0.114 * B (3) In addition, the average value of R, G, B,
Alternatively, the maximum value of R, G, B or the like can be adopted as the luminance (brightness) Y of the pixel. Histogram 5
In order to obtain 0, the image data is scanned from the end, and when the value of the brightness Y is obtained, the value of the array hist [Y] is incremented. Further, in order to shorten the processing time, it is possible to create a histogram by thinning out pixels and searching every few pixels instead of searching all pixels (step A2).

Maximum luminance pixel RGB value (Rma
x, Gmax, Bmax) is calculated (step A4), and then the luminance minimum pixel RGB value (Rmin, Gmi) in the image is obtained.
n, Bmin) is obtained (step A5). By referring to the histogram 50 in FIG. 5, the maximum luminance value and the minimum pixel in the image can be extracted. Maximum brightness Ymax51
Then, the maximum luminance pixel RGB value can be obtained as the RGB value when the luminance is Ymax51 by scanning the image again. Similarly, if the minimum luminance value Ymin53 is obtained, the minimum luminance pixel RGB value is determined by scanning the image once more, so that the luminance is Ymi.
It can be obtained as an RGB value when n53.

When determining the RGB value of the maximum luminance pixel,
The number of pixels of α% (α = 0 to 0.1) of the total number of pixels is Na
Then, the Na-th high-luminance pixel can be obtained as the maximum luminance value in consideration of noise. At this time, the luminance value Yma lower than the histogram 50 by the number of α% pixels is used.
x'52 is used as the maximum luminance value. Then, the image is scanned again, and the RGB value of the maximum luminance pixel can be determined by averaging the RGB values of the pixels of Ymax′52 or more. Even when determining the minimum luminance pixel RGB value,
First, if the number of pixels of β% (β = 0 to about 0.1) of the total number of pixels is Nb, the Nb-th low luminance pixel can be obtained as the minimum luminance value in consideration of noise. At this time,
Brightness value Y that is higher from the histogram 50 by the number of β% pixels
min'54 is used as the minimum luminance value. Then, by scanning the image again and averaging the RGB values of the pixels of Ymin′54 or less, the minimum luminance pixel RGB value can be determined.

The brightness maximum pixel RGB value thus obtained is set as a white RGB value (Rw, Gw, Bw), and the brightness minimum pixel RGB value is set as a black RGB value (Rb, Gb, B).
b). As the black and white RGB values, R, G and B may be set to the same value and stored. As an example in which R, G, and B are the same, the maximum value of the luminance minimum pixel RGB values (Rmin, Gmin, Bmin) is represented by Rb, as shown in Expression (4).
There is a method adopted as Gb and Bb.

Rb = Gb = Bb = MAX (Rmin, Gmin, Bmin) (4) Next, a conversion LUT (look-up table) is created based on the reference white and the reference black (step S4). FIG.
An example of the method for creating the LUT is shown in FIG. The reference white 60 and the reference black 61 are given in advance by the user. Reference white 60, reference black 61, and white RG obtained from the image
Using the B value 62 and the black RGB value 63, the R conversion characteristic 6
4, G conversion characteristic 65 and B conversion characteristic 66 are obtained. Taking the R conversion characteristic 64 as an example, an example of how to obtain it is shown in Expression (5).

R ′ = a × R + b a = (Rw−Rb) / (Rw0−Rb0) (5) b = Rb0−a × Rb Next, each RGB conversion characteristic is stored as a lookup table (LUT). . Examples of LUTs for handling 8-bit pixel values are LUT67, LUT68, LUT
69.

Finally, the white correction processing is completed by converting the pixel values of all the pixels using the created RGB LUTs (step S5).

[0023]

DETAILED DESCRIPTION OF THE INVENTION A color image processing apparatus according to an embodiment of the present invention will be described below with reference to the drawings.

1 and 2 are block diagrams showing an embodiment of the color image processing apparatus of the present invention.

In the embodiment of the present invention, a multi-input image memory 14 that holds a plurality of input images, a multi-output image memory 16 that holds a plurality of output images, and color correction processing are performed on all image data. 18 to control
And color correction means 19. The color correction unit 19 is an input image buffer 1 that holds input image data.
A white and black extraction means 2 for extracting RGB values of white and black pixels from the image, a reference white buffer 7 for holding a reference white, and a reference black buffer 8 for holding a reference black.
And R to create a lookup table for color conversion
It is composed of a GB table creating means 9, an RGB table converting means 20 for performing table conversion for each of RGB, and an output image buffer 12 for holding image data after color conversion.

Further, the white / black extraction means 2 is a brightness value extraction means 35 for calculating a brightness value from RGB values, a pixel thinning search means 43 for thinning out pixel values when obtaining a histogram, and a histogram of image brightness. And a maximum brightness extracting means 36 for obtaining the maximum brightness from the histogram, and a minimum brightness extracting means 37 for obtaining the minimum brightness from the histogram.

A plurality of image data input from the image input device 13 such as a scanner is held in the multiple input image memory 14. Multiple input image memory 1 by CPU 18
Data is input to the color correction means 19 one by one from 4 and processed.

In the color correction means 19, the input image buffer 1
The white and black extraction means 2 obtains the white RGB value and the black RGB value based on the image data input to, and inputs them to the RGB table creation means 3.

In the white / black extraction means 2, the data of the input image buffer 1 is temporarily held in the image data buffer 32, and the histogram calculation means 34 calculates the histogram of the brightness values.

The histogram calculating means 34 is connected to the brightness value calculating means 35 for calculating the brightness value from RGB.

Before the histogram is calculated, noise components are removed from the contents of the image data buffer 32 by the low pass filter 33, and the image data buffer 32 is again restored.
It is also possible to adopt a configuration in which noise is removed by rewriting the data in the.

Further, when the histogram is calculated, the pixel thinning-out searching means 43 may be used to search the image while acquiring pixel values every few pixels instead of searching the entire image to create a histogram. .

From the histogram obtained by the histogram calculating means 34, the maximum luminance extracting means 36 extracts the luminance value Ymax of the pixel having the maximum luminance in the image.

As described in the section of means for solving the problem, the corrected maximum luminance extracting means 44 is used instead of the maximum luminance extracting means 36 to change the number of α% of all pixels to Na.
Then, the luminance value of the pixel having the Nath largest luminance value can be adopted and output.

Similarly, from the histogram obtained by the histogram calculating means 34, the minimum luminance extracting means 37 extracts the luminance value Ymin of the pixel having the minimum luminance in the image.

As described in the section of means for solving the problem, the corrected minimum luminance extraction means 45 is used instead of the minimum luminance extraction means 37 to change the number of β% of all pixels to Nb.
Then, the luminance value of the pixel having the Nb-th smallest luminance value can be adopted and output.

Based on Ymax, the maximum brightness pixel RGB value calculation means 38 is used to calculate and output white RGB values. The maximum brightness pixel RGB value calculation means 38 scans the data in the image data buffer 32, and performs a process of obtaining an average of RGB values of pixels of Ymax or more.

Based on Ymin, the minimum brightness pixel RGB value calculation means 39 is used to calculate and output a black RGB value. The minimum brightness pixel RGB value calculation means 39 scans the data in the image data buffer 32, and performs processing for obtaining the average of the RGB values of pixels of Ymin or less.

Finally, the black value correcting means 42 performs a process of setting the RGB values of the black value to the same value to obtain the black RGB value.
Output the value. As an example of the black value correction means 42, there is a device that realizes the expression (4) in the term of means for solving the problem. In advance, by the reference white black designating means 15,
The reference white RGB value is held in the reference white buffer 7, and the reference black RGB value is held in the reference black buffer 8.

The RGB table creating means 3 creates a lookup table for each of RGB based on the output value of the white / black extracting means 2 and the values of the reference white buffer 7 and the reference black buffer 8, and RLUT4, GLUT5, BLUT6.
To hold.

The RGB table conversion means 20 is an RLUT.
The pixel value of the input image buffer 1 is converted into a table based on the data of 4, GLUT5, and BLUT6 and output to the output image buffer 12. The processed data is output image buffer 12
Is further output to and stored in the multiple-output image memory 16, but after one image data is processed, the CPU 18
Is checked to see if any unprocessed images remain in the multiple input image memory 14. If there is unprocessed data, that data is input to the color correction means 19. In this way, after all the data in the multi-input image memory 14 has been processed, the result in the multi-output image memory 16 is output to the image output device 17 such as a hard disk or a monitor.

[0042]

As described above, the color image processing apparatus of the present invention has the effect that it is possible to correct the color balance of the image displayed in combination by simultaneously processing a plurality of image data. is there.

Further, since color correction can be performed without converting the RGB image data into another color coordinate system, there is an effect that processing can be performed at high speed.

Further, by correcting not only the high-luminance point (white point) but also the low-luminance point (black point), there is an effect that the output image quality is improved as compared with the prior art.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an embodiment of a color image processing apparatus of the present invention.

FIG. 2 is a block diagram showing an embodiment of a color image processing apparatus of the present invention.

FIG. 3 is a flowchart of a color image processing method of the present invention.

FIG. 4 is a flowchart of a white / black extraction unit of the color image processing method of the present invention.

FIG. 5 is an explanatory diagram of maximum brightness and minimum brightness extraction.

FIG. 6 is an explanatory diagram of a lookup table.

FIG. 7 is a block diagram showing an embodiment of a conventional color image processing device.

[Explanation of symbols]

 1 Input Image Buffer 2 White / Black Extracting Means 3 RGB Table Creating Means 4 RLUT 5 GLUT 6 BLUT 7 Standard White Buffer 8 Standard Black Buffer 9 LUT Converting Means 10 LUT Converting Means 11 LUT Converting Means 12 Output Image Buffer 13 Image Inputting Equipment 14 Multiple Input image memory 15 Reference white / black designation means 16 Multiple output image memory 17 Image output device 18 CPU 19 Color correction means 20 RGB table conversion means 30 White / black extraction means 31 Input RGB image data 32 Image data buffer 33 Low-pass filtering filter 34 Histogram Calculation means 35 Luminance value extraction means 36 Maximum luminance value extraction means 37 Minimum luminance value extraction means 38 Maximum luminance pixel RGB value calculation means 39 Minimum luminance pixel RGB value calculation means 40 White RGB value 41 Black RGB value 42 Black Value correction means 43 Pixel thinning search means 44 Corrected maximum brightness value extraction means 45 Corrected minimum brightness value extraction means 50 Histogram 51 Maximum brightness value Ymax 52 Corrected maximum brightness value Ymax '53 Minimum brightness value 54 Corrected minimum brightness value Ymin' 60 Reference white 61 reference black 62 white RGB value 63 black RGB value 64 R conversion characteristic 65 G conversion characteristic 66 B conversion characteristic 67 R lookup table 68 G lookup table 69 B lookup table 100 image memory 101 maximum luminance value extraction unit 102 correction amount Calculation unit 103 Correction ROM 104 Correction ROM 105 Masking circuit

Claims (8)

[Claims] 1. A step of extracting values of a white pixel and a black pixel in an image from input image data, based on the extracted white value and black value and a predetermined reference white value and reference black value. , A color image processing method comprising the steps of creating a conversion table and performing a table conversion for all pixels based on the created table. 2. The color tones are unified by simultaneously applying to a plurality of input image data.
The described color image processing method. 3. An input image buffer for holding input image data, a histogram calculating means for calculating a luminance histogram from the image, and RGB of white pixels and black pixels from the image.
White / black extraction means for extracting values, reference white buffer for holding reference white, reference black buffer for holding reference black, RGB table creation means for creating a lookup table for color conversion, and RGB for each An RGB table conversion means for performing table conversion and an output image buffer for holding image data after color conversion are provided, and a brightness value extraction for calculating a brightness value from RGB values is included in the white / black extraction means. A color image processing apparatus comprising: a means, a histogram calculating means for taking a histogram of luminance, a maximum luminance extracting means for obtaining a maximum luminance from the histogram, and a minimum luminance extracting means for obtaining a minimum luminance from the histogram. 4. A multi-input image memory for holding a plurality of input images, a multi-output image memory for holding a plurality of output images, and means for controlling color correction processing to be performed on all image data. The color image processing apparatus according to claim 3, further comprising: 5. The color image according to claim 3, further comprising a black value correction unit that corrects R, G, and B values of the minimum luminance RGB values to the same value to obtain a black value. Processing equipment. 6. The color image processing apparatus according to claim 3, further comprising pixel thinning-out search means for controlling so as to search every few pixels when calculating a luminance histogram. 7. A corrected maximum luminance for obtaining the luminance value of the Na-th pixel from the highest luminance, where Na is the number of α% of the number of pixels instead of the maximum luminance extraction means for extracting the maximum luminance value from the histogram. Corrected minimum luminance for obtaining the luminance value of the Nb-th pixel from the lowest luminance, where Nb is the number of β% of the number of pixels, instead of the minimum luminance extraction means for extracting the minimum luminance value from the histogram. 7. The color image processing apparatus according to claim 3, further comprising an extracting unit. 8. A low-frequency filtering filter is further provided to remove noise from the image data and improve the accuracy of calculating the luminance histogram.
Alternatively, the color image processing device according to item 7.