JPH07295527A - Processor for image signal binarization processing and method therefor - Google Patents

Abstract

PURPOSE:To provide a processor and method for image signal binarization processing which can process a multilevel image signal fast and obtain a binary image signal consisting of sufficient gradations. CONSTITUTION:A multi-valuing circuit 10 converts a 1st multi-level image signal I (x, y) with resolution R1 and N1 gradations into a 2nd multi-level image signal E (x, y) with resolution R2 and N2 gradations and then a 2nd multi-level circuit 12 converts the 2nd multi-level image signal E (x, y) into a binary image signal B (x, y) with resolution R3 and N3 gradations. Here, R1, R2, and R3 are so set that R1<=R2<R3; and N1, N2, and N3 are so set that N1>N2>N3=2.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION The present invention uses an error diffusion method and a density pattern method or a systematic dither method to process an image signal by dividing the resolution and the number of gradations into two steps.
The present invention relates to an image signal binarization processing apparatus and method for obtaining a value image signal.

[0002]

2. Description of the Related Art When a gradation image is reproduced by using a display device or a printer device capable of only binary display, a gradation image signal which is a multi-valued image signal is composed of "0" and "1" signals.
A process of converting into a value image signal is performed. In this case, various methods have been conventionally proposed as a binarization processing method.

For example, a multi-valued image signal is made to correspond to a sub-matrix composed of ON / OFF signals of M × N dots according to its gradation, and each multi-valued image signal has an area of dots which are ON in the sub-matrix. The density pattern method that reproduces at a rate, M × N pixels as one unit for gradation reproduction,
A systematic dither method for binarizing a dither matrix composed of a plurality of threshold signals of M × N dots corresponding to N pixels and comparing the dither matrix with the M × N pixels to obtain a multi-valued image signal When obtaining a binary image signal by comparing with a fixed threshold value signal, there is an error diffusion method or the like in which an error caused by binarization is distributed and added to a multi-valued image signal of neighboring pixels to perform binarization.

[0004]

By the way, in order to reproduce a gradation image with a high gradation using a binary image signal, it is necessary to set the resolution of the binary image signal to a large value.

In this case, if the density pattern method or the systematic dither method is used for binarization, high speed processing is possible, but between the periodic component contained in the multi-valued image signal and the sub-matrix or dither matrix. Interference easily occurs. As a result, there is a problem that moire and rosette patterns are generated in the binary image.

On the other hand, the error diffusion method is used for binarization.
If this is to be realized by software on a RIP (Raster Image Processor), it will take a considerable amount of time for calculation processing. Note that the calculation time in this case becomes long depending on the square of the resolution of the binary image. Further, if the hardware (image processing circuit) is used, the circuit size becomes large because the amount of buffer memory depends on the resolution.

The present invention provides an image signal binarization processing apparatus and method capable of solving the above-mentioned problems, processing a multi-valued image signal at high speed, and obtaining a binary image signal having sufficient gradation. The purpose is to do.

[0008]

[Means for Solving the Problems]
Therefore, the present invention comprises a processing circuit based on the error diffusion method.
And resolution R₁, Number of gradations N₁The first multi-valued image signal of
Resolution R₂, Number of gradations N ₂(R₂≧ R₁2 <N₂<
N₁) First conversion means for converting into a second multi-valued image signal,
The processing circuit based on the density pattern method,
2 Multi-valued image signal₃, Number of gradations N₃(R₃> R
₂, N₃= 2) second conversion means for converting into a binary image signal
And are provided.

The present invention is also based on an error diffusion method.
Composed of a circuit, resolution R₁, Number of gradations N₁First multi-value of
Image signal with resolution R₂, Number of gradations N₂(R₂≧ R₁Two
<N ₂<N₁) First conversion for converting into a second multi-valued image signal
And a processing circuit based on the ordered dither method.
The second multi-valued image signal is converted into a resolution R₃, Number of gradations N₃
(R₃> R₂, N₃= 2) conversion into a binary image signal
And 2 conversion means.

According to the present invention, the first multi-valued image signal having a resolution R ₁ and a gradation number N ₁ is converted into a resolution R based on the error diffusion method.
₂ , _second number of gradations N ₂ (R ₂ ≧ R ₁ , 2 <N ₂ <N ₁ )
Based on the first step of converting into a multi-valued image signal and the density pattern method, the second multi-valued image signal is converted into 2 with resolution R ₃ and gradation number N ₃ (R ₃ > R ₂ , N ₃ = 2). And a second step of converting into a value image signal.

Further, the present invention is based on the error diffusion method,
The first multi-valued image signal having the resolution R ₁ and the gradation number N ₁ is converted into the _second multi-valued image signal having the resolution R ₂ and the gradation number N ₂ (R ₂ ≧ R ₁ , 2 <N ₂ <N ₁ ). Based on the first step of converting and the systematic dither method, the second multi-valued image signal is converted into a resolution R ₃ ,
The second step of converting into a binary image signal having the number of gradations N ₃ (R ₃ > R ₂ , N ₃ = 2).

[0012]

According to the image signal binarization processing apparatus and method of the present invention, a multi-valued image signal having an intermediate resolution and an intermediate gradation number is generated based on the error diffusion method, and then the density pattern method or the systematic method is used. Based on the dither method, a binary image signal capable of expressing a desired gradation is generated from the multi-valued image signal. In this case, the resolution of the multi-valued image signal generated in the middle is set to be small, so that the calculation processing by the error diffusion method can be speeded up, and the multi-valued image signal and the binary image signal can be separated from each other. By reducing the difference in the resolutions between the two, it is possible to avoid the generation of moire and rosette patterns by the density pattern method or the systematic dither method.

[0013]

1 shows the outline of an image signal binarizing apparatus and method according to the present invention.

In this case, the first multi-valued image signal I (x, y) having the resolution R ₁ and the number of gradations N _{1 is} converted into the multi-valued circuit 10 (first
By a conversion means) to a second multi-valued image signal E (x, y) having a resolution R ₂ and a gradation number N _2, and then the second multi-valued image signal E (x, y)
The multi-valued image signal E (x, y) is converted into a binary image signal B (x, y) having a resolution R ₃ and a gradation number N ₃ by the binarization circuit 12 (second conversion means). Note that the resolutions R ₁ , R ₂ ,
R ₃ is set in a relationship of R ₁ ≦ R ₂ <R ₃ , and the number of gradations N ₁ , N ₂ , N ₃ is set in a relationship of N ₁ > N ₂ > N ₃ = 2. . Further, (x, y) represents an array of pixels configured by each image signal.

As shown in FIG. 2, the multi-value quantization circuit 10 has a solution.
Image R₁, Number of gradations N₁The first multi-valued image signal I (x,
y) is interpolated to convert the magnification and the resolution R
₂(R ₁≤ R₂), Number of gradations N₁First multi-valued image signal consisting of
A magnification conversion circuit 13 for generating a signal I '(x, y) and a resolution R
₂, Number of gradations N₁The first multi-valued image signal I '(x,
y) is the resolution R₂, Number of gradations N₂(N₁> N₂) Consists of
Error diffusion processing circuit for converting into a second multi-valued image signal E (x, y)
14 and. In addition, the magnification conversion circuit 13
If the magnification is set to 1, then R₁= R₂Becomes

The error diffusion processing circuit 14 is, as shown in FIG. 3, a first multi-valued image memory for storing a first multi-valued image signal I '(x, y) having a resolution R ₂ and a gradation number N _1. 16 and the first multi-valued image signal I ′ (x, y) with resolution R ₂ and gradation number N ₂
A multi-valued function generator 18 for converting into a second multi-valued image signal E (x, y), the first multi-valued image signal I ′ (x, y) and the second multi-valued image signal E (x error signal Err (x, y)
Error memory 20 that stores the error signal Err (x-
error diffusion coefficient memory 22 for storing error diffusion coefficient W (k, l) multiplied by k, yl), and said error diffusion coefficient W (k, l)
Random number generator 2 for generating a random number for randomly switching and multiplying the error signal Err (xk, yl)
4 and. In this case, (xk, yl) represents the surrounding pixels of (x, y). Further, the error signal Err (x, y) is calculated by the subtracter 2
6 is obtained as a difference signal between the first multi-valued image signal I ′ (x, y) and the second multi-valued image signal E (x, y). Further, the error signal Err (xk, yl) and the error diffusion coefficient W (k, l) are multiplied in the multiplier 28 and added to the first multi-valued image signal I ′ (x, y) in the adder 30. It

The binarization circuit 12 is composed of a density pattern processing circuit 32 shown in FIG. The density pattern processing circuit 32 stores a second multi-valued image signal E (x, y) having a resolution R ₂ and gradation number N ₂ generated in the error diffusion processing circuit 14 in a second multi-valued image memory. 34 and M ×
A sub-matrix composed of ON / OFF signals of N dots is stored as a density pattern, and the second multi-valued image signal E is stored.
A density pattern memory 36 for converting (x, y) into a binary image signal B (x, y) having a resolution R ₃ and a gradation number N ₃ (= 2).
And a random number generator 38 for giving a random number to the density pattern memory 36.

Next, the operation of the image signal binarization processing apparatus having the above configuration will be described.

First, in the magnification conversion circuit 13 shown in FIG. 2, the first multi-valued image signal I (x, y) of resolution R ₁ is interpolated to obtain the first multi-valued image signal of resolution R ₂ (R ₁ ≤R ₂ ). A value image signal I '(x, y) is generated. Next, in the error diffusion processing circuit 14 shown in FIG. 3, the first multi-valued image signal I ′ (x, y) having the gradation number N ₁ is converted into the second multi-value image signal having the gradation number N ₂ (N ₂ <N ₁ ). It is converted into a value image signal I (x, y).

That is, the first multi-valued image signal I '(x, y)
Is first stored in the first multi-valued image memory 16 and then, in the multi-valued function generator 18, a _second number of gradation levels N 2
It is converted into a multi-valued image signal E (x, y). In this case, as shown in FIG. 5, the multi-valued function generator 18 outputs the gradation number N ₂ at which E (x, y) = f (I ′ (x, y)) (1) It is composed of a multi-valued function f. Note that in FIG. 5, the first multi-valued image signal I ′ (x, y) of N ₁ = 256 is replaced by the second multi-valued image signal E of N ₂ = 9.
The multi-valued function f converted into (x, y) is shown.

The second multi-valued image signal E (x, y) output from the multi-valued function generator 18 is supplied to a subtractor 26, where Err (x, y) = I '(x, y). The error signal Err (x, y) that becomes −E (x, y) (2) is obtained, and the error memory 2
Stored in 0.

Next, the error signal Err (xk, yl) stored in the error memory 20 is multiplied by the error diffusion coefficient W (k, l) stored in the error diffusion coefficient memory 22 in the multiplier 28. By doing so, the diffusion error signal ΔE (x, y) shown in the following equation (3) is obtained. Note that (k, l) is (x, y)
Represents the range of pixels surrounding the pixel specified by.

[0023]

[Equation 1]

The diffusion error signal ΔE (x, y) obtained as described above is added to the first multi-valued image signal I '(x, y) in the adder 30 to obtain the following equation (4). A modified first multi-valued image signal I '(x, y) shown in is obtained.

I ′ (x, y) = I ′ (x, y) + ΔE (x, y) (4) The corrected multi-valued image signal I ′ (x, y) is again multi-valued. It is introduced into the function generator 18. In this way, the error that occurs when the first multi-valued image signal I ′ (x, y) is converted into the second multi-valued image signal E (x, y) with a small number of gradations is Be spread. Note that the error diffusion coefficient W (k, l) stored in the error diffusion coefficient memory 22 is equal to the error signal Err (x
-k, yl) is randomly switched by a random number from the random number generator 24 and multiplied, so that the second multi-valued image signal E (x, y) does not have periodicity due to error diffusion. .

By repeating the above processing, the second multi-valued image signal E (x, y) having the gradation number N ₂ is generated.
In this case, the error diffusion process can be performed at high speed by keeping the resolution R ₂ low.

Next, the resolution R _2, the second number of gradations N ₂
The multi-valued image signal E (x, y) is a binary image having a resolution R ₃ (R ₃ > R ₂ ) and a gradation number N ₃ (N ₃ = 2) in the density pattern processing circuit 32 shown in FIG. It is converted to the signal B (x, y).

That is, the second multi-valued image signal E (x, y) generated in the error diffusion processing circuit 14 is once stored in the second multi-valued image memory 34 and then supplied to the density pattern memory 36. It In this case, the density pattern memory 36 stores a look-up table composed of a plurality of sets of ON / OFF signals of M × N dots. Therefore, when the second multi-valued image signal E (x, y), the random number signal RN, and the position signals ΔX and ΔY are introduced into the density pattern memory 36, the predetermined address designated by these signals is turned on / off. An OFF signal is selected from the look-up table and output as a binary image signal B (x, y).

Here, the number of gradations of the second multi-valued image signal E (x, y) is N ₂ = 9, and the resolution of the binary image signal B (x, y) is R ₃ =
2R ₂ , the processing in the density pattern processing circuit 32 will be specifically described. In this case, as shown in FIG. 6, the density pattern memory 36 stores ON / O of 2 × 2 dots.
Five sets of lookup tables composed of FF signals are set.

Therefore, when E (x, y) = 0, only the lookup table (100%) in which all of the four dots are OFF signals is selected, and (ΔX, ΔY) = (0,
A binary image signal B (x, y) that becomes 0 for all positions of 0), (1,0), (0,1), and (1,1) is generated. When E (x, y) = 1, 0 or 1 is 50
A lookup table in which all four dots are OFF signals according to the random number signal RN generated with a probability of%,
A lookup table in which 3 dots are OFF signals is selected by 50%, and based on that, the binary image signal B is selected.
(x, y) is generated. In the case of E (x, y) = 2 to 8, the binary image signal B (x, y) is similarly generated.

The density pattern processing circuit 32 described above is used.
Then, the random number generator 38 is used to stochastically generate five sets of look-up tables, so that the binary image signal B (corresponding to the gradation number N ₂ of the second multi-valued image signal E (x, y) is generated. x, y)
However, if the look-up table is set as nine sets of look-up tables consisting of ON / OFF signals of 3 × 3 dots, a desired gradation can be obtained without using the random number generator 38. Can be expressed.

As described above, the second multi-valued image signal E
When the binary image signal B (x, y) is generated from (x, y), the change in resolution is small, so that the generation of moire and rosette patterns can be suppressed appropriately. As a result, the error diffusion processing circuit 14 rapidly performs the intermediate resolution R ₂ and the gradation number N.
Second multivalued image signal E having a ₂ (x, y) to generate, then the second multivalued image signal E (x, y) 2 binary image signal without moire and rosette patterns from B (x, y ) Can be generated.

As another embodiment, instead of using the density pattern method by the density pattern processing circuit 32, a binary image signal B (x, y) is used by using the systematic dither method by the systematic dither processing circuit 40 shown in FIG. ) Can also be generated. In this case, the second multi-valued image signal E (x, y) stored in the second multi-valued image memory 34 is the threshold signals of M × N dots set in the dither matrix memory 44 in the comparator 42. And the comparison result is the binary image signal B
Output as (x, y). The _second number of gradations N ₂ = 5
For the multi-valued image signal E (x, y), the dither matrix (M = N = 2) is set, for example, as shown in FIG. Also in this case, as in the case of the density pattern method, the binary image signal B (x, y) having no moire or rosette pattern can be generated at high speed.

The example in which the image signal binarization processing method of the present invention is realized by an image processing circuit which is hardware has been described above, but the same processing is performed by RIP (Raster Image).
Needless to say, it can be realized by software such as Processor).

[0035]

As described above, according to the image signal binarization processing method and apparatus of the present invention, a sufficient gradation can be expressed and a binary image signal free of moire and rosette patterns can be generated at high speed. can do.

[Brief description of drawings]

FIG. 1 is a schematic configuration explanatory diagram of an image signal binarization processing method and apparatus according to the present invention.

FIG. 2 is a diagram illustrating the configuration of the multi-value quantization circuit shown in FIG.

3 is a configuration block diagram of an error diffusion processing circuit shown in FIG.

FIG. 4 is a configuration block diagram of a density pattern processing circuit.

FIG. 5 is an explanatory diagram of a multi-value quantization function.

6 is a process explanatory diagram of the density pattern processing circuit shown in FIG. 4;

FIG. 7 is a configuration block diagram of a systematic dither processing circuit.

8 is an explanatory diagram of a dither matrix stored in a dither matrix memory shown in FIG.

[Explanation of symbols]

10 ... Multivalued circuit 12 ... Binary circuit 13 ... Magnification conversion circuit 14 ... Error diffusion processing circuit 16 ... First multivalued image memory 18 ... Multivalued function generator 20 ... Error memory 22 ... Error diffusion coefficient memory 24 , 38 ... Random number generator 32 ... Density pattern processing circuit 34 ... Second multi-valued image memory 36 ... Density pattern memory 40 ... Systematic dither processing circuit 42 ... Comparator 44 ... Dither matrix memory

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification number Internal reference number FI Technical indication H04N 1/387 101 1/405 1/40 H04N 1/40 103 B

Claims (4)

[Claims] _1. A first multi-valued image signal having a resolution R ₁ and a gradation number N ₁ is composed of a processing circuit based on an error diffusion method, and a resolution R ₂ and a gradation number N ₂ (R ₂ ≧ R ₁ , 2 <N ₂ <N ₁ )
And a processing circuit based on the density pattern method. The second multi-valued image signal is converted into a resolution R ₃ and a gradation number N ₃ (R ₃ > R).
₂ , N ₃ = 2) second conversion means for converting into a binary image signal, and an image signal binarization processing device, comprising: 2. A first multi-valued image signal having a resolution R ₁ and a gradation number N ₁ which is composed of a processing circuit based on the error diffusion method and which has a resolution R ₂ and a gradation number N ₂ (R ₂ ≧ R ₁ , 2 <N ₂ <N ₁ )
Of the second multi-valued image signal and a processing circuit based on the systematic dither method. The second multi-valued image signal is converted into the resolution R ₃ and the number of gradations N ₃ (R ₃ >). R
₂ , N ₃ = 2) second conversion means for converting into a binary image signal, and an image signal binarization processing device, comprising: 3. A first multi-valued image signal having a resolution R ₁ and a gradation number N ₁ is converted into a resolution R ₂ and a gradation number N based on an error diffusion method.
₂ (R ₂ ≧ R ₁ , 2 <N ₂ <N ₁ ), a first step of converting into a second multi-valued image signal, and the second multi-valued image signal with a resolution R ₃ , based on a density pattern method. A second step of converting into a binary image signal having the number of gradations N ₃ (R ₃ > R ₂ , N ₃ = 2), and a method for binarizing an image signal. 4. A first multi-valued image signal having a resolution R ₁ and a gradation number N ₁ is converted into a resolution R ₂ and a gradation number N based on an error diffusion method.
₂ (R ₂ ≧ R ₁ , 2 <N ₂ <N ₁ ), and a second step of converting the second multi-valued image signal into a resolution R ₃ based on a systematic dither method. And a second step of converting into a binary image signal having a gradation number N ₃ (R ₃ > R ₂ , N ₃ = 2), and an image signal binarizing method.