JP3082931B2 - Color image processing method - Google Patents

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a color image processing apparatus such as a color copier, a color scanner, and a color facsimile, and more particularly, to a color image processing apparatus having a function of performing color / monochrome discrimination. It is.

[Conventional technology]

In a conventional system for reading a color original, chromatic / achromatic is determined using a ratio of three components of three primary colors (RGB, XYZ, etc.) of an input image signal. For example, if the input levels of the three primary colors have similar values, it is determined that the color is an achromatic color, and if the levels vary, the color is determined to be a chromatic color. A technique is known in which the determination for each pixel is repeated for all the pixels of the document, and if a chromatic color pixel exists, the document is colored, and if not, the document is black and white.

[Problems to be solved by the invention]

However, in the above prior art, when the reading accuracy of the color sensor for image input is not good, pixels around the black characters of the input image are often erroneously determined to be chromatic. Therefore, there is a problem that a monochrome document is erroneously determined as a color document. If a monochrome document is erroneously determined as a color document, the following problem occurs.

In the case of a color copying machine, when a black-and-white document is copied by superimposing color inks (CMY), characters, lines, and halftone dots are poorly tightened due to color misregistration and differences in the spectral characteristics of the ink. In addition, in the case of a color facsimile, in addition to the above-described disadvantage that the print quality is poor, there is also a disadvantage that transmission of three primary colors takes a long time to transmit and increases communication cost even though the original is a black-and-white document. .

The present invention has been made in view of the above problems, and is an image processing method that performs processing by a computer, and can perform color / monochrome determination processing of an image with high accuracy and with a reduced processing load. It is an object to provide a color image processing method.

[Means for solving the problem]

In order to achieve the above object, the present invention is an image processing method for performing processing by a computer, wherein the computer determines whether a pixel of interest in image information according to an image is chromatic or achromatic. A first determination step, and referring to peripheral pixels of the target pixel, cause the computer to determine whether or not the peripheral pixels include a first pixel whose brightness and chromaticity are both equal to or lower than a certain level. A second determination step, and when it is determined in the second determination step that the first pixel is included in the peripheral pixels of the pixel of interest, the determination result of the first determination step for the pixel of interest is When the target pixel is determined to be achromatic in the correction step of causing the computer to correct and the first determination step, the reference of the peripheral pixel of the target pixel in the second determination step is referred to as the peripheral pixel in the peripheral pixel. All of Characterized by a control step of controlling the computer so as not against the elements.

[Embodiment] (Principle of the present invention) An original is read from a color scanner, and the data is read by NT.
Convert to SC RGB, and furthermore, brightness data Y and chromaticity data I
And Q. FIGS. 6 and 7 show IQ plotted on a graph.

FIG. 6 shows the distribution when a monochrome document is read, and FIG. 7 shows the distribution when a color document is read. It can be seen that when a black-and-white document is read, it is distributed near the origin, and for a color document, it is distributed away from the origin. Therefore, the origin (I =
(0, Q = 0), the color original is determined if there are many pixels farther away, and the monochrome original is determined if there are fewer pixels. The original can be determined based on such a concept.

(Scanner Reading Accuracy) When black and white characters are read from the scanner, if the accuracy of the scanner is not good, the input image will be an image having a color shift as shown in FIG. In this case, a pixel having a color misregistration is determined as a color pixel, and as a result, there is a possibility that a color original is erroneously determined in spite of a black and white original. Therefore, if a black pixel (black and white pixel and a dark pixel) exists around the input pixel, it is determined that the input pixel is a color-shifted pixel and is actually a black and white pixel. By doing so, it is possible to reduce the reading error of the scanner.

First Embodiment FIG. 9 shows a flowchart of the process. The processing is large, and the pixel discrimination processing (S1 to S4) and the pixel correction processing (S5 to S1)
1) and original discrimination processing (S12 to S15).
The pixel determination process is a process of determining whether the pixel of interest is color or monochrome, and the document determination process is a process of determining whether the entire document is a color document or a monochrome document. The details of the algorithm will be described below according to the flow.

<Pixel Determination Process> First, the color image determination process will be described. The color pixel determination is processed in S1 to S4. Whether the input pixel is a color pixel or a monochrome pixel is determined for each pixel.

In color space conversion processing S1, color space conversion processing is performed. That is, NTSC-RG
The data normalized to B is converted into a luminance signal Y and chromaticity signals I and Q according to equation (1).

Distance calculation In S2 and S6, in order to create saturation information from I and Q, the distance calculation unit Is calculated. This indicates how far the target color is from the origin on the IQ chromaticity diagram. If this value is large, it can be said that the saturation is high and the probability of being a color pixel is high.

The saturation comparison unit S3 compares the saturation of the target pixel with a threshold. As shown in FIG. 10 (A), if the target pixel C is a part of a character or a base (limited to white), the saturation is low, In order to satisfy the condition, the process proceeds to S4, and the number of black pixels BPLX is set to 1
Just increase. Then, the process returns to S2 to read the next pixel and perform the same processing.

When the target pixel is located at a position considered to be a color pixel as shown in FIG. The process proceeds to S5 without satisfying the condition of.

Pixel correction processing S5 to S11 are pixel correction processing units. This is because even if the S3 saturation comparison unit determines that a pixel is a color pixel, it may be erroneously determined to be a color pixel due to blurring of black characters caused by a color shift of the sensor. Is provided to correct the error. Hereinafter, the pixel correction unit will be described.

In S5, the eight peripheral pixels S ₁ to S ₈ of the pixel of interest C of FIG. 3 1
Input for each pixel. In S6, RGB → YIQ conversion is performed by color space conversion (similar to S1).

S7 saturation information Is calculated (similar to S2). S8: Saturation of surrounding pixels And the luminance Y are compared with a threshold value. For example, if the surrounding pixel S ₁ is a character portion, the saturation brightness is low low In addition, the condition of Y <γ is satisfied. As a result, the target pixel C was determined to be a color pixel in the pixel determination process. However, it is actually due to a color shift of the sensor, and it is determined that the pixel of interest is a black pixel. Just increase. For surrounding pixel S ₆ And S ₆ in order not satisfy the condition of Y <gamma proceeds to S10 is judged not to be a black pixel. In step S10, a check is made to determine whether all the surrounding eight pixels have been input. In this way, it is checked whether or not the surrounding eight pixels include a black pixel by repeating the processing of S5 to S10 up to the eight pixels S1 to S8 around the target pixel. If at least one black pixel is included in the eight surrounding pixels, the target pixel C is determined as a black pixel and the process proceeds to S9. If no black pixel is included in the surrounding eight pixels, the target pixel C is not a color pixel due to color misregistration but proceeds to S11 as a true color pixel, and the number CPIX of color pixels is increased by one.

<Color / Monochrome Document Discrimination> The above-described pixel discrimination (S1 to S11) is performed on all the pixels of the document, and the ratio between the color pixels CPIX and the monochrome pixels BPIX is obtained. If the condition of CPIX> δ is satisfied in S13, the target document is determined to be a color document, and if not, it is determined to be a monochrome document.

The above is a flowchart in the case where the processing is performed by, for example, computer software. The circuit configuration for performing the above processing will be described below. Referring to FIG. 1, an original is placed on a scanner constituted by 101 CCDs, and when scanned, RGB data normalized to the NTSC standard is output. The YIQ converter 102 converts the RGB data output from 101 into a lightness (luminance) signal Y and chromaticity signals I and Q. The pixel discriminating unit 103 discriminates whether the inputted one pixel is a monochrome pixel (saturation is low) or a color pixel. Reference numeral 104 denotes a document discriminating unit, which counts the discrimination results for each pixel, obtains the ratio of the number of color pixels to the number of all pixels,
Depending on the ratio, whether the original is color or black and white
5 Output the document discrimination signal. 106, an image processing unit; 107, a display for displaying images; 108, a facsimile transmission unit for transmitting images; and 109, a printer for reproducing images on a recording medium. The image processing unit 106 performs different processes on a black-and-white document and a color document according to the document discrimination signal 105.

For example, the image signal to be sent to the facsimile transmission unit 108 is compressed and encoded so that each of R, G, and B signals can be stored in the case of a color original, and is converted into a density (luminance) signal in the case of a black and white original. Compression coding such as normal MH, MR, MMR, etc.

For the image signal sent to the printer 109, in the case of a color original, logarithmic conversion is performed on the R, G, and B signals,
R, M, C, K
In the case of a black-and-white original, only a K (black) signal is generated, and black monochrome printing is performed.

The type of printer may be a laser beam printer, an ink jet printer, a thermal transfer printer, a dot printer, or any other printer capable of performing color printing.

FIG. 2 is a diagram showing a YIQ conversion unit, and calculates equation (1). 201, 202, and 203 are R, G, and B input signals, respectively. 204 to 212 are registers for setting the coefficient of equation (1), 213 to 221 are multipliers for multiplying the input data by the coefficient,
222, 223, 225 and 228 are adders, and 226 and 229 are subtractors. The result of the operation is Y at 224, I at 227, and Q at 230.
Is output.

For example, when calculating Y, register 204
0.3, 0.59 is set in the register 207, and 0.11 is set in the register 210. The R data 201 and the contents of the register 204 are multiplied by a 213 multiplier. Similarly, a multiplier 216 uses the G data and the contents of the register 207, and a multiplier 21 uses the B data and the contents of the register 210.
Multiply by 9. Then, the three multiplication results are added to the adders 222 and 22.
3 and finally Y is obtained. Similarly, I and Q
Can be calculated.

 FIG. 3 is a diagram showing a pixel determination unit.

Y data is input to 301, I data is input to 302, and Q data is input to 303.
I is squared by the multiplier 310. Q is squared by the multiplier 311. I ² and Q ² are added in adder 312. next Is calculated by referring to a table.
For example, when the input of the ROM 313 is 2, the second address of the ROM is accessed, and data 1.414 is output. Further ROM
313 output Are input to the comparators 305 and 306. The threshold value β is set in the register 308, and the threshold value α is set in the register 309. Then, the threshold value β and the Are compared, In this case, 1 is output, and otherwise, 0 is output. In the comparator 306, the threshold α and Are compared, At the time of 1, 1 is output, and otherwise, 0 is output. This comparator 30
The output result of 6 is referred to as a first determination result.

Next, for the lightness (luminance) signal, a threshold value γ is set in the register 307, and the comparator 304 compares Y with γ. Outputs 1 when Y <γ, 0 otherwise
Is output. The outputs of comparators 304 and 305 are AND gate 314
Is ANDed. That is, the brightness (brightness) is low,
When the saturation is low, the AND gate 314 outputs 1, and otherwise outputs 0. The output result of the AND gate 314 is referred to as a secondary determination result. The output of the AND gate 314 is input to the line memories of flip-flops (FF) 318 and 315. The output of the line memory 315 becomes the input of the line memory 316 and the FF 312. Therefore, FF321
, A pixel delayed by one line is input to FF324. The FFs 318 to 326 hold the output result of the AND gate 314 in synchronization with the pixel clock. Therefore, assuming that the secondary determination result of the target pixel is held in FF322 at a certain point in time, FF318, 319, 3
20, 321, 323, 324, 325, and 326 hold the secondary determination results of the surrounding eight pixels. The OR results of the secondary determination results of these eight pixels are obtained by OR gates 329, 330, 331, and 332. The output of the OR gate 332 becomes 1 if there is at least one pixel whose secondary determination result is 1 among the 8 surrounding pixels. That is, if there is at least one black pixel in eight pixels around the target pixel, the 332 outputs 1. The output of the OR gate 322 is called a pixel correction result.

The output of the comparator 306 (the primary determination result) enters the line memory 317, and is delayed by one line and flows to FF327 and FF328. That is, the FF328 and FF322 hold the primary determination result and the secondary determination result of the same pixel, respectively.

Finally, the NOR of the pixel correction result and the primary determination result is determined by a NOR gate. When it is determined that the pixel is a color pixel, the color pixel determination signal 334 becomes 1, and when it is determined that the pixel is monochrome, it becomes 0.

 FIG. 4 is a diagram showing a document discriminating section.

Reference numeral 406 denotes an up counter, which has as input an initial count data DAT 403, a pixel clock 402, a color pixel judgment signal 401, and a carry bit 407 as output. Four
The counter 06 performs count-up from the initial count data in synchronization with the pixel clock. During the period in which the count-up is enabled, the color pixel determination signal is counted only during 1 and the color pixel determination signal is counted as a result. Will be. For example, when 406 is a 10-bit counter and the document discrimination threshold δ is 255, DAT is set to 769. Then, when the number of pixels determined to be color becomes 255, the carry bit 407 becomes 1, and the number of color pixels in the document becomes 255 or more. Therefore, a signal indicating that the target document is a color document (document determination signal 105) becomes 1.

With the above processing, it is possible to determine whether the original is color or monochrome.

Second Embodiment FIG. 5 is a diagram for explaining a second embodiment. First
In the embodiment, the configuration of FIG. 2 is used to realize the calculation of Expression (1). However, in the second embodiment, the calculation of Expression (1) is approximated by the configuration of FIG. Equation 2 is approximated by adding the coefficient of equation (1) to a power of two. For example, the coefficient 0.3 in Equation 1 approximates 0.25 + 0.0625. By doing so, the calculation of 0.3 * R can be easily performed by adding the data obtained by shifting the input data by 2 bits and the data shifted by 4 bits.

Next, FIG. 4 will be described. R, G, B input data are set in 501, 502, and 503, respectively.
Numerals 519 and 520 denote circuits for shifting bits of input data by a bit shift unit. 514, 515, 516, 517, 518, 521, 52
3, 527 is an adder, and 524, 525, 526 are subtractors. In the finally calculated data, Y is output to 528, I is output to 529, and Q is output to 530.

Next, the data flow will be described. The calculation method is Y, I, Q
Since the same calculation is performed, Y is described here as an example, and I and Q are omitted. Equation 2 As the calculation of the term, a two-bit shift is performed at 504 and a four-bit shift is performed at 507, and the result is added by an adder 517. Also As the calculation of the term, the input G data is shifted by 1 bit at 509 and shifted by 4 bits at 510, and the result and the output of the adder 517 are added by the adder 522. The output of the adder 522 is the result of the addition of the R and G terms in the calculation of Y in Equation 2. Next, the input data 503 of B is shifted by 3 bits and input to the adder 523. Consequently adder 5
The output 528 of 23 becomes Y in Equation 2.

According to the above embodiment of the present invention, means for separating the color separation data (RGB) sent from the scanner into chromaticity data and lightness data, means for generating saturation data from chromaticity data,
A method of comparing brightness data with a threshold value to determine a bright pixel if brightness data is large, a means of determining a dark pixel if brightness data is small, a color pixel if saturation data is large by comparing saturation data with a threshold, and a monochrome pixel if brightness data is small. Means for determining, means for increasing the number of black and white pixels by one when the input pixel is determined to be a black and white pixel, means for correcting the determination when the input pixel is determined to be a color pixel, means for correcting the determination, If there are dark pixels and white pixels around the input pixels, the number of black and white pixels is increased by one, and if not, means for increasing the number of color pixels by one. By comparing the number of color pixels with a threshold value, a means for discriminating a color document from a color document if the number of color pixels is large, and a means for discriminating a black and white document if the number of color pixels is small is provided so as to discriminate between a color document and a black and white document. is there.

However, the threshold value for comparing the saturation data with the threshold value must be changed between the input pixel and the surrounding pixels.

That is, means for separating each component signal of the input color signal into a lightness signal and a chromaticity signal, means for generating a saturation signal from the chromaticity signal, means for comparing the saturation signal with a threshold value α, Means for comparing with a threshold value β, means for comparing a lightness signal with a threshold value γ, means for increasing the number of black-and-white pixels by 1 if the saturation signal is smaller than α,
As a result of comparing the saturation signal of the pixel of interest with the threshold value, if the saturation signal is equal to or greater than α, the result of comparing the saturation signal with the threshold value β of the surrounding pixels indicates that the saturation signal is less than β and the threshold value is the brightness signal. As a result of comparing the value γ, it is determined that the brightness signal is smaller than γ, and if a pixel satisfying this condition is included in the surrounding pixels,
Means for increasing the number of black and white pixels by 1 and increasing the number of color pixels by 1 if not included, means for comparing the number of color pixels with a threshold value δ, and comparing the number of color pixels with the threshold value δ, the number of color pixels> If δ, by providing a means for determining that a document is a color document, the document can be determined without erroneous determination even when the accuracy of the scanner is not good or when the background of the document has a light color. As a result, it is possible to perform a process according to the type of the original, thereby improving print quality and reducing communication costs.

The image input means is not limited to the scanner, and may be an interface such as a host computer, a still video camera, or a video camera.

Further, the number of peripheral pixels to be referred to is not limited to eight. Further, a majority decision process may be performed instead of the OR process.

In addition, in addition to separating into (Y, I, Q), (L ^* ,
a ^* , b ^* ), (L ^* , u ^* , v ^* ), (Y, u, v)
The determination may be performed separately.

〔The invention's effect〕

As described above, according to the present invention, it is determined whether a target pixel in image information according to an image is chromatic or achromatic, and reference to peripheral pixels of the target pixel is performed to correct the determination result. When the target pixel is determined to be achromatic, the computer performs a process of controlling not to perform the process on all the pixels in the peripheral pixels, so that the target pixel in the image information corresponding to the image is The determination as to whether the image is chromatic or achromatic can be made with high accuracy and with a reduced processing load.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a color / monochrome original automatic discriminating apparatus according to a first embodiment of the present invention, FIG. 2 is a block diagram of a YIQ conversion section, FIG. 3 is a block diagram of a pixel discriminating section, and FIG. FIG. 5 is a block diagram of the YIQ conversion unit of the second embodiment, FIG. 6 is a diagram showing the distribution of I and Q when a monochrome document is read, and FIG. 7 is a color document. FIG. 8 is a diagram showing the state of color misregistration, FIG. 9 is a diagram showing the overall flow of color / monochrome document discrimination, and FIG. 10 is a diagram for explaining pixel correction processing. It is. 101: Scanner 102: YIQ conversion unit 103: Pixel discrimination unit 104: Document discrimination unit 105: Document discrimination signal

──────────────────────────────────────────────────続 き Continued on front page (58) Field surveyed (Int.Cl. ⁷ , DB name) H04N 1/46-1/62 H04N 1/40-1/409

Claims (1)

(57) [Claims] 1. An image processing method for performing processing by a computer, comprising: a first determination step for causing the computer to determine whether a target pixel in image information corresponding to an image is chromatic or achromatic. A second determination step of causing the computer to determine whether or not the peripheral pixel includes a first pixel whose brightness and chromaticity are all equal to or lower than a certain level with reference to a peripheral pixel of the target pixel; A correcting step of causing the computer to correct the result of the first determining step on the pixel of interest when it is determined in the second determining step that the first pixel is included in the peripheral pixels of the pixel of interest; When the target pixel is determined to be achromatic by the first determination step, the reference of the peripheral pixel of the target pixel by the second determination step is performed for all the pixels in the peripheral pixel. Not performed And a control step of controlling the computer as described above.