JPH11328389A - Device and method for image processing - Google Patents

Abstract

PROBLEM TO BE SOLVED: To simplify the hardware and to convert the gradations of multivalued image data with high picture quality by performing the gradation conversion by adding a random numbers value to the inputted multivalued image data, dividing the addition result by a specific value, and outputting only the quotient. SOLUTION: A sign inversion and data hold part 711 converts the sign of random numbers R inputted from a random numbers generation part 710 and outputs hold random numbers to a selector 712. The selector 712 switches and outputs the random numbers R inputted from the random numbers generation part 710 and the hold random numbers from the sign inversion and data hold part 711. An addition part 707 performs an adding process between an inputted image signal D and an addition quantity control part 713. A hysteresis controlled value calculation part 708 calculates the controlled value of hysteresis with the signal from the addition part 707 and outputs it to a threshold value calculation part 705. A division part 709 divides the inputted image signal DR by a constant 17 and outputs only its quotient. Thus, a gradation converting process part 604 performs binarizing processing.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing apparatus and method for converting input multivalued image data into gradations and outputting the converted data.

[0002]

2. Description of the Related Art Conventionally, an error diffusion method (hereinafter referred to as "ED") and an average density storage method (hereinafter referred to as "MD") are generally known as image processing methods for performing halftone expression. These are intended to express a halftone macroscopically by performing area gradation expression using a small number of gradations.

[0003] However, these processes are performed only when the input image (8 bi
Since processing is performed for t), the processing load was large. Therefore, 8 bits of the input image are converted to 4 bits by pre-processing,
U.S. Patents USP 5,394,250 and US
It is proposed in P5,436,736.

[0004]

However, the above-mentioned proposal is to improve a unique texture which is a problem in a specific density region when processed by ED or MD even if the processing load can be reduced. Could not.

Therefore, as shown in FIG. 1, there has been proposed a method of adding a predetermined random number to an input image at a stage prior to the t-value processing to improve the texture of the input image. The details are described below.

In FIG. 1, reference numeral 101 denotes a random number generation unit, which is a processing unit for generating a random number R. FIG. 2 is a block diagram illustrating a configuration of the random number generation unit 101. FIG. 3 shows the random number generation in a programming language C. Here, the description will be given with reference to FIG.

First, in initialization, p [ii] :( 0 ≦ ii ≦ 2
Write "0" to the register of 5), and set "1" only to the register of p [12]. Then, before outputting a random value, p [0] = (p [25] ＾ p [24] ＾ p
[23] After performing the calculation of p [22] & 1), a random number value of -17 to 17 is generated by the following calculation.

Random number = (1-2 * p [22]) * (((p
[15] * 64 + p [16] * 32 + p [17] * 16 + p [18] *
8 + p [19] * 4 + p [20] * 2 + p [21]) * 17) / 1
28) Here, the reason for generating a random number of ± 17 or more and finally dividing by 128 is to eliminate a bias in the random number generation rate.

In the above equation, random numbers from -17 to 17 are generated.
It goes without saying that a similar effect can be obtained even if the change is made to generate up to 15 random numbers.

Random number = (1-2 * p [2]) * (p [18]
* 8 + p [19] * 4 + p [20] * 2 + p [21] The random number value tends to increase the texture improvement effect when the absolute value of the random number value is increased, and decreases when the absolute value is decreased.

However, if the random number value is too large, it goes without saying that the image becomes rougher than the effect of improving the texture, leading to a reduction in image quality. 7 to 17 in absolute value
A degree value is feasible.

Returning to FIG. 1, the random number value output from the random number generation unit 101 is input to the sign inversion and data holding unit 102 and the selector 103.

The sign inversion and data holding unit 102 inverts the sign of the input random number once to obtain “p × X + p /
For 2 ″ (arbitrary even value of p ≧ 2, X: address value in the main scanning direction), the random number value is held and output.

On the other hand, the selector 103 is configured to switch the output between the input random number value and the value of the sign inversion and data holding unit 102 according to the image position signal.

Here, the above-mentioned pixel position signal is “p ×
A value of the random number generation unit 101 is output for each pixel of X "(an arbitrary even value of p ≧ 2, X: an address value in the main scanning direction),
A control signal for inverting the sign and outputting the value of the data holding unit 102 for each pixel of “p × X + p / 2”.
Generated from the sync signal. At other pixel positions,
It is configured to output a value of “0”.

When the input multi-level signal D is other than "0" or "255", the adding section 104 adds the random number generated by the above processing to the input multi-level signal D.
At this time, although not shown, a limiter is set at a minimum of 0 and a maximum of 255.

Next, the divider 105 divides the input image signal by "17" to separate the quotient from the remainder. Here, the quotient (0 to 15) is input to the adder 107, and the remainder (0 to 16) is input to the comparator 106.

The random number generation unit 109 generates 0 to 1 generated by a method basically similar to that of the random number generation unit 101 described above.
6 is output. The details thereof are the same as those of the random number generation unit 101, and thus the description thereof is omitted, but FIG. 4 shows the random number generation of the random number generation unit 109 in the programming language C.

Next, in the comparator 106, the random number value (0 to 16) input from the random number generation unit 109 and the division unit 1
Compare with the remainder value (0-16) entered from 05,
Only when the remainder is larger than the random value, a value of “1” is output, and otherwise, “0” is output. This output value is added with the quotient value (0 to 15) from the division unit 105 by the addition unit 107, and a 4-bit signal is output to a t-value processing unit described later.

After the input signal is converted into a 4-bit signal by the above processing, a t (t ≦ 4) value conversion process is performed by the t-value conversion processing unit 108 and output.

However, in the case of the method shown in FIG. 1 described above, the addition unit 104 and the addition unit 107 add a random number twice to the input image, which not only increases the hardware scale but also increases the image quality. There is a problem that it also leads to a decrease in

SUMMARY OF THE INVENTION The present invention has been made to solve the above problems, and provides an image processing apparatus and method which can simplify hardware and perform high-quality gradation conversion of multi-valued image data. The purpose is to:

[0023]

According to the present invention, there is provided an image processing apparatus for converting input multi-valued image data into gradations and outputting the converted data. An addition unit that adds the random number value to input multi-valued image data, and a gradation conversion unit that divides an addition result by the addition unit by a predetermined value, outputs only a quotient, and performs gradation conversion. It is characterized by having.

According to another aspect of the present invention, there is provided an image processing method for converting input multi-valued image data into gradations and outputting the same, the method comprising the steps of: And a tone conversion step of dividing the addition result in the addition step by a predetermined value, outputting only a quotient, and performing tone conversion. And

[0025]

Embodiments of the present invention will be described below in detail with reference to the drawings. In this embodiment, a case where the present invention is applied to a binarization process of a color copying machine will be described as an example.

FIG. 5 is a block diagram showing a schematic configuration of a color copying machine according to the embodiment. In the figure, reference numeral 509 denotes an image reading unit, which includes a lens 501, a CCD sensor 502, and an analog signal processing unit 503. Here, the original 5 imaged on the CCD sensor 502 via the lens 501
Image information of R (Re)
d), G (Green), and B (Blue) are photoelectrically converted into analog electric signals. The converted image signal is input to an analog signal processing unit 503, where the image signal is sampled and held for each of R, G, and B colors, corrected for a dark level, and the like, and then subjected to analog / digital (A / D) conversion. You. The digitally converted full-color image signal is 50
4 is input to the image processing unit.

The image processing unit 504 performs correction processing necessary for the image reading system such as shading correction, color correction, and Υ correction, smoothing processing, edge enhancement, other processing, processing, and the like. Is output.

The printer unit 505 is composed of an exposure control unit (not shown) composed of a laser or the like, an image forming unit (not shown), a transfer paper transport control unit (not shown), and the like. An image is recorded on the transfer paper by a signal.

Reference numeral 510 denotes a CPU circuit, and the CPU 50
6, an image reading unit 509, an image processing unit 504, and a printer unit 505.
And the like to control the sequence of the apparatus in a comprehensive manner.

Image processing unit Next, the image processing unit 504 will be described in detail. FIG. 6 is a block diagram illustrating a configuration of the image processing unit 504 according to the embodiment. In the figure, 601 is a shading correction unit, 602 is a gradation correction unit, 603 is a color / monochrome conversion unit, and 604 is a gradation conversion processing unit.

In the above configuration, first, the digital image signal output from the analog signal processing unit 503 is input to the shading correction unit 601. The shading correction unit 601 corrects variations in sensors that read the original and light distribution characteristics of the original illumination lamp. next,
The corrected image signal is input to the tone correction unit 602 to convert the luminance signal into density data, and density image data is created. The image signal converted into the density data is input to the color / monochrome conversion unit 603, and is output as monochrome data. The data output from the color / monochrome conversion unit 603 is input to the gradation conversion processing unit 604, and error diffusion processing (ED processing) or average density storage processing (MD processing) is performed as pseudo halftone expression.

Next, a gradation conversion processing section 604 according to the present invention.
Will be described in detail.

FIG. 7 is a block diagram showing a detailed configuration of the gradation conversion processing unit 604 in the embodiment. In the embodiment, a binary MD method that enables texture control will be described as an example.

A random number, which will be described later, is added to the image signal D from the color / monochrome conversion unit 603, and a signal DR ′ obtained by performing a division process by a constant “17”, which will be described later, and error data generated by the binarization process. E is input to the error correction unit 702. Then, an error correction, which will be described later, is performed, and the image signal is output to the binarizing unit 701 as an image signal DE.

The binarization section 701 receives the error-corrected image signal DE, a binarized slice value S described later, and an average density value m described later, and receives the image signal DE and the binarized slice value S. To obtain a binary output N. Further, the binarization error data E is calculated by subtracting the image signal DE from the average density value m.

The binarization result delay section 703 receives the binarized binary output N, delays the data by a predetermined number of lines, and converts the delayed data into a plurality of line binarization results Nmn and B *.
ij is sent to the average density calculator 704 and the threshold calculator 705.

The average density calculating section 704 receives the delayed multi-line binarization result Nmn, performs a product-sum operation with a preset coefficient, calculates an average density value m, and outputs an average density value m.
06 and an average density value m to the binarization unit 701.

The threshold value calculation unit 705 calculates the binarization result B * ij of the plurality of lines of the above-described binarization result delay unit 703, the input multi-value data D, and the output T of the hysteresis control amount calculation unit 708. A threshold control amount in an arbitrary area is calculated in accordance with the input B * ij signal which is a past binarization situation (pattern), and the calculated threshold control amount is sent to the addition unit 706 as a binarized slice value S '. I do.

The addition section 706 receives the signal of the average density value m of the average density calculation section 704 and the binary slice value S ′ of the threshold value calculation section 705, performs addition processing, and outputs the result. Output to the binarization section 701 as a binarized slice value.

The random number generating section 710 generates an m-sequence random number R of “−17 to 17” by a method described later, and outputs it to the selector 712 and the sign inversion and data holding section 711.

The sign inversion and data holding unit 711 inverts the sign of the random number R input from the random number generation unit 710 and holds data only for a certain number of pixels, which will be described later.
The stored random number is output to 12.

The selector 712 switches and outputs the random number R input from the random number generator 710 and the random number held from the sign inversion and data holding unit 711 based on the pixel position signal at the timing described later.

The addition amount control unit 713 controls a random number output value according to the value of the image signal D by using a method described later.

The addition section 707 performs an addition process between the input image signal D and the addition amount control section 713. The hysteresis control amount calculation unit 708 calculates a hysteresis control amount by a method described later based on a signal from the addition unit 707, and outputs the control amount to the threshold value calculation unit 705. The division unit 709 divides the input image signal DR by a constant 17 and outputs only the quotient. At this time, all remainders have been truncated.

With the above configuration, the gradation conversion processing section 604
Performs a binarization process.

Next, each processing unit of the gradation conversion processing unit 604 will be described in detail. First, the random number generation unit 710 generates a random number with a configuration similar to that of FIG. 1 described above.

First, in initialization, p [ii] :( 0 ≦ ii ≦ 2
Write "0" to the register of 5), and set "1" only to the register of p [12]. Then, before outputting a random value, p [0] = (p [25] ＾ p [24] ＾ p
[23] After performing the calculation of p [22] & 1), a random number value of -17 to 17 is generated by the following calculation.

Random number = (1-2 * p [22]) * (((p
[15] * 64 + p [16] * 32 + p [17] * 16 + p [18] *
8 + p [19] * 4 + p [20] * 2 + p [21]) * 17/12
8) In the embodiment, random numbers from -17 to 17 are used. However, the random number generation unit may be changed to generate random numbers from -15 to 15 by the following calculation.

Random number = (1-2 * p [2]) * (p [18]
* 8 + p [19] * 4 + p [20] * 2 + p [21] Here, what is important is that the maximum value of the generated random numbers (17 in the embodiment) is divided by a division unit 709 described later (the embodiment). In this case, more than 1/2 of 17) (fraction cut off)
That is what you need to do.

In the embodiment, since the number to be divided by the division unit 706 is 17, 17/2 = 8 (decimal part truncation), and the maximum random number generation value of the random number generation unit 710 is set to a value of 8 or more. (In this embodiment, 17 is set to 8 or more). If the number to be divided by the division unit 706 is 7, 7/2 = 3 (decimal point truncation), and the maximum random number generation value of the random number generation unit 310 needs to be 3 or more. Needless to say.

As described above, random number generation is performed for all pixels.

Next, the random number inversion and data holding section 711 calculates the random number value of the random number generation section 710 generated at the pixel position of “p × X” (even number of p ≧ 2, X: address value in the main scanning direction). , Only the sign is inverted, and the output is held for “p / 2” pixels. For example, when the value of p is “2”, a random number generated at a pixel position of “p × X”, that is, a pixel position of “0, 2, 4, 6, 8, 10, 12, 14,. Are temporarily held, and the pixel positions are “1, 3,
5, 7, 9, 11, 13, 15,... ”, The sign of the random number value is inverted and output. Of course, it is not necessary to keep the random number values of all the pixels. A configuration in which each pixel is refreshed is sufficient.

On the other hand, the selector 712 is configured to switch and output the random number value of the random number generation unit 710 generated for each pixel and the random number value from the sign inversion and data holding unit 711 according to the pixel position signal. I have.

This pixel position signal is “p × X + p /
Only at the pixel position of 2 ″ (even number of p ≧ 2, X: address value in the main scanning direction), sign inversion and data holding unit 711
Is selected, and otherwise, the random number value from the random number generator 710 is selected.

The addition amount control unit 713 determines that the large random number value in the above-described random number generation unit 710 is larger than 1/2 (8 in the embodiment) divided by the division unit 709 (17 in the embodiment). In addition, an important point is that the output control according to the input multi-level signal D is performed only for the large random number.

FIG. 8 shows the addition amount control of the addition amount control unit 713 in a program language C. What is important here is a method of setting a constant SL for dividing the output RD of the selector 712 by a constant SL.

The constant SL is determined so that the result of RD / SL is １ ／ of the number divided by the divider 709. That is, in the embodiment, since the number divided by the division unit 709 is 17, 17/2 = 8 (decimal part truncation), and RD
/ SL = 8, 17 / SL = 8 (fraction truncated) to S
It is necessary to set the L value to “2”.

When the input multi-level signal D is equal to or smaller than N1 (for example, 16), the minimum necessary random number is added by the calculation of "P1 = RD / SL". Here, the minimum necessary random number means that the amplitude of the random number is set to “−8 to 8” in order to divide by “17” in the division unit 709. That is, assuming that the amplitude of the random number is α (here, 16), the number divided by the division unit 709 is (α +
1) (here, 17).

When the input multi-level signal D is larger than N1 and N
2 (for example, 32) or less, "P1 = (RD-R
D / SL) * (D-N1) / (N2-N1) + RD / S
By the calculation of L ″, the amplitude-controlled random numbers are added. The point here is that the amplitude is controlled only for the portion that is equal to or more than the minimum necessary random number. Is "(RD-TD / SL) *" in the above equation.
(D-N1) / (N2-N1) ".

Similarly, when the input multilevel signal D is N3 (for example, 2
01) and smaller than N4 (for example, 233), "P1 = (RD-RD / SL) * (N4-D) /
By the calculation of (N4−N3) + RD / SL ”, a random number whose amplitude is controlled in the same manner as in the above-described processing is added. (RD-RD / SL) * (N4-D) /
(N4-N3) ".

When the input multi-level signal D is equal to or larger than N4, only the minimum necessary random numbers are added by the calculation of "P1 = RD / SL". Further, when the input multi-level signal D is out of the above range, all the input random numbers RD are output from the addition amount control unit 713 as the random numbers P1.

The adder 707 performs the process of adding the random number P1, the input multi-level signal D, and the constant 8 described above. The addition of the constant 8 is also an important point here. This is because, since the constant to be divided by the divider 709 described later is “17”, the remainder after division becomes 16 at the maximum, the amplitude of the random number to be added needs to be an even number of 16 or more. The constant 8 is obtained from the calculation. This is added as a bias component.

Of course, if the number to be divided by the divider 709 is "5", the constant to be added by the adder 707 is "2". Although not shown, a limiter is set so that the addition result falls within the range of “0” and “255”. The signal from the adder 707 is input to the divider 709 and the hysteresis control amount calculator 708.

The division unit 709 performs the operation of dividing by the constant 17 as described above. At this time, the output signal is only the quotient after the division, and all the remainders are discarded. That is, in the embodiment, it is not necessary to use a comparator for comparing the “remainder of division” and the “random number” used in the conventional example, and it is possible to perform processing of converting the signal into an 8-bit signal only by the quotient of the division processing. Needless to say, the 4-bit image quality is higher than that of the conventional example. The output signal DR ′ from the division unit 709 is input to an error correction unit 702 described later, and an error correction process is performed.

Next, a binarization method for performing texture control will be described. As described above, the error correction unit 702
Is used to input the image signal DR and the binarized error data E, calculate the image signal DE obtained by performing error correction on the image signal DR ′, and output the image signal DE to the binarizing unit 701, as shown in FIG. It is configured as follows.

The input binarization error data E is halved by the division circuit 901. The result is branched into two systems, one of which is input to a subtraction circuit 902 and the other is input to a line buffer 903. Subtraction circuit 9
02, the difference EB between the binarization error data E and E / 2 (=
EE / 2) is calculated, and the result is input to the addition circuit 904. At this time, although not shown, possible values of the binarized error data E are changed to “−6 to +
6 ".

The addition circuit 904 includes a line buffer 90 for one line of a plurality of bits (three bits in the embodiment).
The sum of EA delayed by one line by 3 and EB from the subtraction circuit is calculated and output to the addition circuit 905. The adding circuit 905 calculates the sum of EA + EB and the image signal D
The sum with R ′ is calculated and output as an image signal DE. That is, as shown in FIG. 10, the error correction unit 702 calculates the binarization error EA when the pixel A on one line is binarized with respect to the target pixel “*” and the pixel B one pixel before. A process of adding the value of the binarization error EB at the time of binarization to the value of the target pixel is performed.

Next, the binarizing section 701 inputs the above-mentioned image signal DE, a binarized slice value S which will be described later, and an average density calculation value m which will be described later, and compares them to obtain a binary signal. It outputs a value output N and binarization error data E, and is configured as shown in FIG.

The input image signal DE is branched into two systems, one of which is input to a comparison circuit 1101 and the other is input to a subtraction circuit 1102. Comparison circuit 1101
Then, the value of the image signal DE is compared with the value of the binary slice value S, and the binary output N is output as follows.

When DE> S, N = 1 when DE ≦ S. N = 0. In the subtraction circuit 1102, the value of the image signal DE is subtracted from the average density calculation value m, and the binarization error data E Output as

E = m-DE At this time, as described above, although not shown, limiter processing is performed so that the value of E falls within the range of "-6 to +6".

Next, the binarization result delay section 703 receives the binary output N from the binarization section 701, delays the signal by a predetermined number of lines, and converts the binary result Nmn, B * ij of a plurality of lines. The data is sent to the average density calculation unit 704 and the threshold value calculation unit 705 as shown in FIG.

First, the input binary output N is 1 bit 1
The data is sent from the line buffer 1201 for the line to the line buffer 1202, and the data is delayed for each line. At the same time, the delay of one pixel is successively performed by the delay 1203 to the delay 1208, which is a delay circuit for one pixel. Then, the output of the delay 1206, the delay 1
The outputs of 207 are output as N14 and N15, respectively.

The binarized data delayed by one line by the line buffer 1201 is delayed by the delay 1209 to the delay 1214, and the delay 1209 is delayed by the delay 1209.
213 is output as N25 from N21. The binarized data further delayed by one line by the line buffer 1202 is output from the delay 1215 to the delay 1
Delayed by 220 and delayed 1215 to delay 121
9 is output as N35 from N31.

At the same time, the outputs of the delays 1206 to 1208 are output as B10, B20 and B30, respectively. The binarized data delayed by one line by the line buffer 1201 is output as B32 to B02 and Bi12 to Bi32, respectively, after being delayed. Further, the binarized data further delayed by one line by the line buffer 1202 is
They are output as B31 to B01 and Bi11 to Bi31, respectively.

That is, the average density calculation section 704 performs delay processing of a plurality of lines and a plurality of pixels of data obtained by binarizing a two-dimensional image, and obtains a multi-line binarization result Nmn as shown in FIG. Is input to the average density calculation unit 704 in an appropriate state.

Next, the average density calculation section 704 receives the multi-line binarization result Nmn, performs a product-sum operation on the preset coefficient and the delayed binarization result, and
01 and the adder 706 output data m used, and are configured as shown in FIG.

That is, the multiplication circuit 1401 inputs the binary data N15 and the coefficient M15, and outputs the result of multiplication of the two. In addition, in the multiplication circuit 1402, the binarized data N
14 and a coefficient M14, and outputs the result of multiplication of the two. Similarly, the above operation is performed by each of the multiplication circuits 1403 to 1412, and the multiplication results are all added by the addition circuit 1413. The result is output as the average density calculation value m. FIG. 15 is a diagram illustrating an example of coefficients when performing the process of calculating the average density.

Next, the hysteresis control amount calculating section 708
Changes the value of the constant ALF (= 32) according to the input signal DR and outputs it as the S ′ signal. This is for adjusting the amount of hysteresis in an arbitrary density region. That is, this enables texture control in an arbitrary density region.

FIG. 16 shows the process of calculating the hysteresis control amount in a program language C. When the input signal DR is equal to or less than the constant LR1 (= 16), processing is performed to set 11 to “0”, and the input signal D
R is greater than constant LR1 and constant LR2 (= 48)
In the following case, 11 is obtained by the following equation.

Ll = ((DR-LR1) * (ALF * 256 / (LR2-LR1))) / 256; By this operation, as the value of the input signal DR increases from the constant LR1 to the constant LR2, the value of 11 Gradually approaches from 0 to the constant ALF (= 32).

On the other hand, if the input signal DR is larger than the constant LR2 and equal to or smaller than the constant LR3 (= 233),
Is output as a constant ALF. Also, the input signal D
R is larger than the constant LR3 and the constant LR4 (= 25
5) In the following cases, 11 is obtained by the following equation.

Ll = ALF-((DR-LR3) * (ALF * 256 / (LR4-LR3))) / 256; This is because as the value of the input signal DR increases from LR3 to a constant LR4, the output 11 , Gradually approaching 0 from the constant ALF. When the input signal DR is LR4
If it is larger, processing is performed to set 11 to 0.

After the above processing, the constant ALFm (=
16) is output as the output signal T. The purpose of this subtraction is to change the signal T of the hysteresis control amount calculation unit 708 from a negative value to a positive value. As a result, it is possible to perform texture control in an arbitrary density region in a wide latitude range.

Next, the threshold value calculation section 705 will be described. FIG. 17 shows that the threshold value calculation processing is performed in the programming language C.
It is shown by.

First, the threshold value calculating section 705 converts the value of the input signal T of the hysteresis control amount calculating section 708 into constants LT1 (= 2), LT2 (= 4), and LT3.
(= 8), divided by LT4 (= 16), and variables A (= T / LT1), B (= T / LT2), (C (=
T / LT3) and D (= T / LT4).

Next, the value of the binarized slice value S ′ is changed to the variable A according to the binarized result arrangement state (pattern) of the output B ′ * ij from the binarized result delay unit 703 by a method described later. ,
It is controlled by B, C, D and constants. FIG. 18 is a diagram showing a binarization result arrangement state (pattern). In this example, the pixel immediately before the pixel of interest is not referred to for high-speed processing. Of course, if a sufficiently high-speed logic can be constructed, it goes without saying that there is no problem even if the pixel immediately before the target pixel is referred to.

Next, processing for controlling the binarized slice value S according to the arrangement (pattern) of the binarized result will be described.

When the binarization state around the target pixel is as follows, the binarization slice value S is forcibly set to the constant 15 of max.
And output. This is to make it difficult to form dots.

B32 == 0 && B22 == 1 && B12 == 0 && B21 == 0 && B11 == 1 && B01 == 0 or Bi12 = 0 && Bi22 == 1 && Bi32 == 0 && B01 == 0 && Bi11 == 1 && Bi21 == 0 Also, two values around the pixel of interest When the binarization status is as follows and the input value data D is less than 31 (31 of 0 to 255), the binarized slice value S is forcibly set to the constant 15 of max.
And output. This is also to make it difficult to forcibly print dots under the above conditions.

B12 == 0 && B02 == 0 && Bi12 == 0 && Bi22 == 0 && Bi32 == 0 && B11 == 0 && B01 == 0 && Bi11 == 1 && Bi21 == 0 && Bi31 == 0 && B20 == 0 On the other hand, under the above conditions, the input multi-value data D Is 31 (0-2
In the case of 31) or more of 55, the binarized slice value S is set to the average density calculation value m and output. This is to prevent texture control from being performed when the past binarization result has a specific arrangement (pattern).
Of course, here, the constant 31 is not a fixed value but a parameter, and can be set to another value such as 48 or 64.

At this time, if the value of 31 is increased, it is needless to say that texture control is more likely to be positively applied, and conversely, if it is smaller, texture control is less likely to be applied.

When the binarization state around the target pixel is as follows, the binarization slice value S is calculated from the average density calculation value m.
The value is set to a value obtained by subtracting the variable A (S = mA) and output.

B02 == 0 && Bi12 == 0 && B11 == 0 && B01 == 1 && Bi11 =
= 1 && Bi21 == 0 && B20 == 0 This is to make it easy to forcibly print dots under the above conditions. Also at this time, the processing is performed without referring to the binarization result immediately before the target pixel.

Similarly, according to the pattern of each binarization result, the value of the binarization slice value S is used by using the internal variables A, B, C, D and constants without referring to the result immediately before the target pixel. To control the binarized slice value S ′. As a result, when the hysteresis control amount calculation value T is positive, the control is performed so that the dot is easily hit, and when the hysteresis control amount calculation value T is negative, the control is performed so that the dot is hard to hit.

When the above-described processing is sequentially performed for each pixel, the output value B ′ * ij of the binarization result delay unit is obtained in an arbitrary density area according to the value of the hysteresis control amount calculation value T. Can be controlled to an arbitrary texture according to the value of.

In the embodiment, control is performed so that a texture of 2 × 2 pixels is obtained. This makes it possible to form an image in which an arbitrary number of dots are collected and stabilized in an area where one pixel is not stable due to the characteristics of the printer.

The binarized slice value S ′ obtained in this way is input to the adding section 706 together with the output m of the average density calculating section 704, and the adding processing is performed. At this time,
When the signal of S ′ is 15, the binarized slice value S is set to 1
5; otherwise, S = S ′ + m is calculated and output. FIG. 19 shows the above-described calculation in the programming language C.

After the binarization slice value S is obtained by the above-described addition processing, the binarization processing is performed by the above-described binarization section 701, and the binary signal is converted into a gradation conversion processing section 604.
And is printed out by the printer unit 505.

[Modification] Next, a modification of the above-described gradation conversion processing unit will be described. In the modification, the configuration around the random number generation unit shown in FIG. 7 is simplified.

FIG. 20 is a block diagram showing a detailed configuration of the gradation conversion processing section in the modification. The same components as those shown in FIG. 7 are denoted by the same reference numerals and description thereof is omitted.

In FIG. 20, reference numeral 2001 denotes a random number generation unit which generates a positive random number from 0 to 16. The specific configuration is the same as the configuration shown in FIG. That is, “0” is written to the register of p [ii] :( 0 ≦ ii ≦ 25) at initialization, and “1” is set only to the register of p [12].
Then, before outputting a random value, p [0] = (p [25] ＾ p [24] ＾ p [23] ＾ p [22] for each pixel
After performing the calculation of & 1), random numbers of 0 to 16 are generated by the following calculation. Random number = ((p [15] * 64 + p [16] * 32 + p [17] * 16
+ P [18] * 8 + p [19] * 4 + p [20] * 2 + p [2
1] * 16) / 128) The reason why 16 or more random numbers are generated and finally divided by 128 is to eliminate a bias in the random number generation rate.

An adder 2002 performs a process of adding the input random number R and the input multi-value data. The feature of the modification is that, unlike the adding unit 707 shown in FIG. 7, 1/2 of the maximum remainder value 16 divided by the dividing unit 709 is not added as a bias component. This is because the random numbers to be added only take positive values and do not take negative values,
This is because the bias component is no longer needed.

The output value from the adder 2002 is divided by the divider 709 in the same manner as the configuration shown in FIG. 7, and only the quotient value is output. This makes it possible to omit the comparator for comparing the remainder of the result of the division performed with the random number, which has been performed in the conventional example, thereby simplifying the hardware scale.

[0105] Thereafter, 2 of the same processing as in the above-described embodiment is performed.
The binarization processing is performed, and the binary signal is
4, and is printed out by the printer unit 505.

The point in this modification is that the division unit 7
Since the number to be divided by 09 is 17, the random number generation unit 200
That is, the maximum random number value generated in 1 is set to 16 or more (the number to be divided by the division unit 709; 17-1). Of course, if the amount of random numbers to be added is large when the input multi-valued data is other than "0" and "255", the effect of improving the texture is obviously increased. is there. Therefore, it is preferable to set it to 31 or less.

Further, in the above-described embodiment, the sign inversion and data holding unit 711 generates a random number that regularly changes to ± and adds the generated random number to the input multi-value data D. However, the present invention is not limited to this. It is needless to say that the same effect can be obtained by adding the random number of the random number generation unit that changes to ± at random without using the sign inversion and data holding unit 711.

As a result, the scale of the hardware can be reduced. However, since the low-frequency component is lower in the random numbers that regularly change to ± shown in the embodiment,
Image quality tends to be less deteriorated.

Furthermore, in the above-described embodiment, the minimum necessary random number 8 (the absolute value [the value of the division unit 709; 17 /
2] When adding only the fractional part of the operation result), it goes without saying that the addition amount control unit 713 can be omitted.

Then, the dividing unit 709 of the embodiment described above.
When the number to be divided by is an even number (for example, 18), the input multi-valued data may be doubled by 1-bit shift to perform the calculation.

This is because, when dividing by 18, the remainder of the division is up to 17, so the amplitude of the random number to be added needs to be an odd number of 17 or more. This odd number is equally distributed to ±. This is because the bias component is 17/2 = 8.5 and cannot be divided.
In other words, the bias component cannot be set to a value of 8.5 by the adding unit 707.

Therefore, the processing may be performed by doubling the input multi-value data.

The input multi-valued data is doubled, and the number to be divided by the division unit 709 is also doubled to 36 (18 × 2).
The bias component of 7 is calculated as 17 ([the numerical value of the division unit 709; 36
/ 2-1]), it is possible to divide the amplitude of the random number to be added into equal ± 17. As a result, even when the number to be divided by the division unit 709 is an even number, it can be realized with the above-described configuration.

Although the binarization has been described in the above-described embodiments and modifications, the present invention is not limited to this. In other words, it can be applied to pre-processing such as quaternary and octal, and is not limited to the above-mentioned average density preservation method (MD method), but can also be applied to a general error diffusion method (ED method). It goes without saying that you can do it.

Further, it goes without saying that the present invention can be applied not only to monochrome (monochromatic) processing but also to color signals.

As described above, according to the embodiment,
By combining random number addition processing for improving texture and 8-bit to 4-bit conversion processing using random numbers as pre-processing for t-value conversion processing into one, it is possible to reduce the amount of random number addition and obtain a texture similar to the conventional one. Can be improved,
In addition, preprocessing for converting 8 bits of input data into 4 bits becomes possible.

Further, the hardware scale can be reduced, and the image quality can be improved.

The present invention can be applied to a system including a plurality of devices (for example, a host computer, an interface device, a reader, a printer, etc.), but it can be applied to a single device (for example, a copier, a facsimile). Device).

Further, an object of the present invention is to supply a storage medium storing a program code of software for realizing the functions of the above-described embodiments to a system or an apparatus, and to provide a computer (CPU or MP) of the system or apparatus.
It goes without saying that U) can also be achieved by reading and executing the program code stored in the storage medium.

In this case, the program code itself read from the storage medium realizes the function of the above-described embodiment, and the storage medium storing the program code constitutes the present invention.

As a storage medium for supplying the program code, for example, a floppy disk, hard disk, optical disk, magneto-optical disk, CD-ROM, CD
-R, a magnetic tape, a nonvolatile memory card, a ROM, or the like can be used.

When the computer executes the readout program code, not only the functions of the above-described embodiment are realized, but also the OS (Operating System) running on the computer based on the instruction of the program code. ) May perform some or all of the actual processing, and the processing may realize the functions of the above-described embodiments.

Further, after the program code read from the storage medium is written into a memory provided in a function expansion board inserted into the computer or a function expansion unit connected to the computer, based on the instructions of the program code, It goes without saying that the CPU included in the function expansion board or the function expansion unit performs part or all of the actual processing, and the functions of the above-described embodiments are realized by the processing.

[0124]

As described above, according to the present invention,
The hardware can be simplified, and the gradation conversion processing of the multi-valued image data can be performed with high image quality.

[0125]

[Brief description of the drawings]

FIG. 1 is a diagram showing a configuration of a conventional example in a pre-process of a t-value conversion process.

FIG. 2 is a block diagram showing a configuration of a random number generation unit 101 shown in FIG.

FIG. 3 is a diagram illustrating random number generation in a programming language C;

FIG. 4 is a diagram illustrating random number generation by a random number generation unit 109 in a programming language C;

FIG. 5 is a block diagram illustrating a schematic configuration of a color copying machine according to the embodiment.

6 is a block diagram illustrating a configuration of an image processing unit 504 illustrated in FIG.

FIG. 7 is a block diagram showing a detailed configuration of a gradation conversion processing unit 604 shown in FIG.

FIG. 8 illustrates an addition amount control of an addition amount control unit 713 in a programming language C.

FIG. 9 is a diagram showing a detailed configuration of an error correction unit 702.

FIG. 10 is a diagram illustrating a process in which an error correction unit 702 adds a binarization error to a target pixel.

FIG. 11 is a diagram illustrating a detailed configuration of a binarizing unit 701.

FIG. 12 is a diagram illustrating a detailed configuration of a binarization result delay unit 703.

FIG. 13 is a diagram showing a configuration of data used in a binarization result delay unit 703.

FIG. 14 is a diagram illustrating a detailed configuration of an average density calculation unit 704.

FIG. 15 is a diagram illustrating a configuration of coefficients used in an average density calculation unit 704.

FIG. 16 illustrates a process of calculating a hysteresis control amount in a program language C.

FIG. 17 illustrates processing of the threshold value calculation unit 705 in a programming language C.

FIG. 18 is a diagram showing a binarization result arrangement state (pattern).

FIG. 19 illustrates the calculation of a binarized slice value S in a programming language C.

FIG. 20 is a block diagram illustrating a detailed configuration of a gradation conversion processing unit according to a modification.

Claims (7)

[Claims] 1. An image processing apparatus for converting input multi-valued image data into gradations and outputting the same, wherein a random number generating means for generating a random value, and an addition for adding the random value to the input multi-valued image data An image processing apparatus comprising: means for dividing a result of addition by the adding means by a predetermined value, and outputting only a quotient to perform gradation conversion. 2. The image processing apparatus according to claim 1, wherein said random number generating means generates a random number having a predetermined amplitude that changes to ±. 3. The image processing apparatus according to claim 2, wherein said adding means further adds 1/2 of a value divided by said predetermined value as a bias value. 4. The image processing apparatus according to claim 1, wherein said random number generating means generates a random number value having a positive value. 5. The image processing apparatus according to claim 1, wherein when the predetermined value is an even number, the image data and the predetermined value are doubled for the calculation. 6. The image processing apparatus according to claim 1, further comprising a t-value converting means for converting an output value of said gradation converting means into a t-value. 7. An image processing method for converting input multi-valued image data into gradations and outputting the same, wherein a random number generating step of generating a random number value, and adding the random number value to the input multi-valued image data And a gradation conversion step of dividing the addition result in the addition step by a predetermined value and outputting only a quotient to perform gradation conversion.