JPH02285771A - Picture processing unit - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to an image processing apparatus such as a facsimile machine or a digital copying machine, and particularly to an image processing apparatus that performs pseudo-halftone processing on multivalued data read from a document and reproduces it in binary form. It relates to a processing device.

[Prior Art] Conventionally, as pseudo halftone processing methods of this type, there are (1) error diffusion method and (2) average density approximation method (Japanese Patent Application Laid-Open No. 1043693/1983).

In the former method, the multi-value pixel data of the pixel of interest is processed with a fixed threshold value to be binarized, and the error between the binarization level and the multi-value pixel data of the pixel of interest is distributed to pixels in the vicinity of the pixel of interest at a predetermined distribution rate. This is a binarization method that performs addition. The latter uses already binarized binary data in the vicinity of the pixel of interest, and when the pixel of interest is binarized into black or white, the weighting of both is determined by the neighborhood average value.
This is a method of binarizing the pixel of interest by selecting one of the average values that is closer to the level of the pixel of interest.

The former requires two-dimensional calculations on multivalued data, and has the disadvantage that it cannot be made inexpensive due to the large amount of processing required. texture occurs, and the actual number of reproducible gradations is extremely inferior to the former.

[Problems to be Solved by the Invention] Therefore, a new binarization method is being considered in which an error diffusion method is added to the above average density approximation method.

However, this method has the disadvantage that pixel-sized notches occur at the edge portions of characters, line drawings, etc., making clear binarization impossible.

An object of the present invention is to eliminate the above-mentioned drawbacks, and to have a high-quality binary image reproduction ability with a small amount of processing and a simple configuration.
In particular, it is an object of the present invention to provide an image processing device that eliminates notches at edge portions.

[Means for Solving the Problem] In order to solve this problem, the image processing device of the present invention has the following features:
a first binarization means that binarizes input image data using an average value based on binarized data in the vicinity of a pixel of interest that has already been binarized; and a first binarization means that binarizes input image data using a constant threshold value. The image forming apparatus includes a second binarization means, and a selection means for selecting one of the first and second binarization means according to whether the input image data is an edge portion or not.

Here, the first binarization means includes an error diffusion means for diffusing a binarization error.

Further, the selection means determines an edge portion based on the binarized data.

Further, the average value and the selection signal of the selection means are stored in a storage means using the binary data near the pixel of interest as an address.

[Operation] In such a configuration, the first binarization means binarizes the input image data by an average value based on the binarized data near the pixel of interest that has already been binarized, and the input image data is binarized by a certain threshold value. The second binarization means for binarizing the image data is selected depending on whether the input image data is an edge part or not, thereby eliminating notches at edge parts of characters, line drawings, etc.

[Embodiment] FIG. 1A is a block diagram of an image processing apparatus of this embodiment.

In the figure, 100 is an image data input section, and an image receiving section 1
01 or the image reading unit 102 reads a document, digitizes it, and inputs multivalued data 500 that has been subjected to shading correction etc. in the image preprocessing unit 103. 200 is the first 2
The value conversion section includes an average density calculation processing section 200a and an error diffusion processing section 200b in an integrated form. 2o1 is a second binarization unit that performs binarization using a constant threshold value T0. Reference numeral 202 denotes a binarization selection unit which selects the first binarization unit 20 and the second binarization unit 201 depending on whether the input data is a character or an edge portion of a diagram. The binarized data 510 selected by the binarization selection unit 202 is transmitted by the image data output unit 300 to the image transmission unit 301, or to the image display unit 3o29 and the image printing unit 30.
3 is output.

The principle of the binarization method of the first binarization unit 200 in this embodiment will be described.

In FIG. 2A, f (i, j) indicates the multilevel density (value between 0 and 1) of the input image at the pixel position of interest that is currently being binarized, and pixel positions above the broken line are already binary. After the pixel is binarized, the same processing is sequentially performed on f (i+1.j). Similarly, binarized image B
(i, j) indicates the density after binarization (taken as a value of 0 or 1).

R(x, y) is a standardized weight mask for calculating the average density, and an example will be shown in which 13 surrounding pixels including the pixel of interest are referred to.This method is similar to the average density approximation method described above. Similarly, when the pixel of interest is binarized as black or white, the average density in the vicinity m+(i, j),
Mo(1,j) is calculated using the following formula.

mt (f, j) ”Σ Σ R(X, y) X
B (i+x, jay) −・・■yyuki−2x±−
Matashi, B (i, l = Matashi, B (i, j) = 0 On the other hand, using the average value m ts + m l, the pixel of interest f (i, j) is t (t, j) + E (i, j) > 1/2
When [mo (i, j) J, (i, j) ] = me (i, j) + '8 (0,0)B (i,
j ) =1 H, (i+ 1.j)x'8[f(i,j)+E(i,
j)-mt(i,j)]Ez(j,i+1)-Et(i
+1. j) f(t, j)+E(i, j)≦Tl1o (
i, j) + 'When R(0,0) B (i, j
) = O Et (i+ l, j) s+l8[f (i, j
) + E (i, j) -mo (i, j)]E
* (j, i + 1)・Et (i+ 1. j)
...■Here, R(0,0) is the average calculation weight for the pixel of interest, and E(i, l・Et
(i, j) + E2 (i, j).

In the binarization based on the above equation 0, the binarization error E(i,
j) is the sum of the error that occurs when two pixels, the pixel f (i-1, j) one pixel before and the pixel f (i, j-1) one line before, are binarized. , that is, 2 of the two pixels
At the time of value conversion, the pixel f (t, j-1) is mt
(i, j-1) or +110 (IIJ-1), f
(i-1,j) is mt (i-1,j) or mo(i-
1, j), errors occur during binarization in each case.

The binarization method of the first binarization section 200 of this embodiment is different from the conventional average density approximation method and greatly improves the halftone reproduction ability. In addition, in this binarization process, the input image data f(i, j) is binarized while being corrected by the binarization error E(i, j), so the error is extremely small compared to the conventional error diffusion method. It is possible to express gradation with the same high quality with a reduced amount of calculation.

However, in this binarization method (hereinafter referred to as the average density preservation method), resolution information,
In particular, extremely noisy notches occur in the resolution information at the edges of characters and diagrams.

In addition, granular dots occur in the back shadows of characters. Therefore, the binarization method of the second binarization unit 201 of this embodiment, that is, the binarization method using a constant threshold T0, is based on the discrimination result by the pattern matching method on the binarized image, which will be described below. ,
Selectively applied to the edge area of the character.

Figures 4A to 4H show typical examples of surrounding binary image patterns when the pixel of interest is thought to be located at a location that should be binarized using a certain threshold, such as the edge of a character area. * in Figure 0 indicates that either O or l may be used, and Figures 4A to 4D
The figure and FIGS. 4E to 4H show back patterns in which O and 1 are reversed, respectively.

The same pattern has the following two characteristics. ■ Characters and line drawings are often a collection of patterns of 0 or 1 consecutive in either of two orthogonal directions.

(2) Halftone areas are often a collection of isolated points that are inverted with respect to surrounding pixel values.

This is an example of a pattern selected with an eye on Although it is possible to achieve more accurate discrimination if the same region is further expanded, good results were obtained with only the eight patterns of this example, as described below.

The binarization unit of this embodiment will be explained using the detailed block diagram of the binarization unit shown in FIG. 1B.

14 is an average value calculation ROM, which delays the flip-flops 2 to 13 shown and the binarized data by two lines.
Due to i Fo 1, when the binarized data is simultaneously input to the address terminal while maintaining the positional relationship shown in FIG. 2A with respect to the pixel of interest, the above-mentioned average density value m0 is output.

If a conversion table is written in advance in the average value calculation ROM 14 by normalizing the calculation results using the weighting coefficients shown in FIG. 2C to the 8-bit input multi-value data level, the average value ffLo can be easily output. . In addition, in a similar manner, one bit of the output is used to output a discrimination signal sel indicating whether or not the patterns match the patterns shown in FIGS. 4A to 4H.

The average value m0 is input to a binarization error calculation unit 15 composed of a comparator 15a and a subtractor 15b, and the input multivalued data is binarized and binarized after being corrected by a binarization error, which will be described later. An error value is calculated. This binarized data is input to one input terminal of the multiplexer 16 as binarized data by the first binarization section 20 .

The error values E+ and Ex according to formula (2) are input to correction adders 19 and 20, respectively, and first, in the adder 20, the input multi-value data f is converted into a binarized error E.

After delaying the correction value f+Ez by one line, the adder 19 adds correction for the error Ex, and the error due to binarization is corrected. The same correction value f+(εr
+Ea) is held with a delay of one clock at D-F/F l 8, and is input to the aforementioned binarization error calculation portion 15 and the binarization comparator 17 of the second binarization section 21.

The comparator 17 has a constant threshold value T based on a threshold value that is either fixed in advance or manually variable by the operator.
Executes binarization processing using 0, and the result is sent to multiplexer 1.
6 is input to the other input terminal.

The multiplexer 16 selects one of the binarized data inputs according to the discrimination signal sel output from the average value calculation ROM 14, and applies it to the recording means and inputs it to the binarized data 2-line delay unit 1, for example. te, the latter 2
Used for valuation. Furthermore, when it is determined that it is an edge part of a character part, the output of the comparator 17 is selected and at the same time the output is set to E, E2, 0 in the binarization calculation unit 15, thereby reducing the error to the vicinity of the pixel of interest. stop propagation.

In addition, for color-separated color image data, if this binarization process is applied to the "black" signal, the "black" signal can be more effectively processed.
Only characters can be clearly binarized. Further, the pattern is not limited to the patterns shown in FIGS. 4A to 4H, and by expanding the reference area, the accuracy becomes more accurate.

In addition, in the average value calculation in the first binarization section 200, even if the weight R(0, O) of the pixel of interest is set to zero, processing results that are sufficiently superior to the conventional error diffusion method and average density approximation method are obtained. R
(0,0)=O may also be used.

Next, another embodiment for determining the edge portion of a character will be described using FIGS. 5A to 5E.

FIG. 5A shows the binarized data of 12 pixels, and by comparing with the binarized data of orthogonally adjacent pixels according to each pixel, 1 is given to the same logic and added, and a coefficient "0 to 1" indicating orthogonality is added. 4”
Assign the value of .

For example, taking the pixel a, if a and f have the same logic and a and b are true logic, set "l" to a, and if a, f, and b all have the same logic, set "l" to a. 2", or "0" if both a and f, a and b are true logic, calculate the sum of the assigned linearity coefficients for all 12 pixels, and if it exceeds a predetermined value, it is determined to be a character edge. You may do so.

The above embodiment is shown in FIG. 5B and FIG. 5C, and the example executed for two types of binary image examples is shown in FIG. 5D and FIG. 5E, respectively. ', the total number of orthogonality cases is S = 8, and in the case of binary image example B shown in Figure 5C, it is 5 = 18, and it is clear that the example in Figure 5C is closer to the character edge pattern. shown. Note that if all the 12 pixels referred to at this time are "1" or "0", the above rule is not followed and the area is intentionally determined to be a halftone area. Needless to say, the above determination can be performed by look-up table conversion processing if it is stored in the ROM as a bit result in advance, as in the embodiment described above.

[Effects of the Invention] According to the present invention, it is possible to provide an image processing device that can remove notches at the edges of character portions and obtain high-quality pseudo-halftone images at low cost.

[Brief explanation of drawings]

FIG. 1A is a block diagram showing the configuration of the image processing device of this embodiment, FIG. 1B is a block diagram showing the detailed configuration of the binarization section of this embodiment, and FIGS. 2A to 2C are the blocks of this embodiment. Figure 3 is a diagram showing the mask used in this example, and Figures 4A to 4H are diagrams showing the matrix used for edge discrimination in this example. The figures shown in FIGS. 5A to 5E are diagrams showing other embodiments of edge portion discrimination. In the figure, 100 is an image data input unit, 101 is an image receiving unit, 102 is an image reading unit, 103 is an image preprocessing unit, 200 is a first binarization unit, 200a is an average density Arithmetic processing section, 200b...Error diffusion processing section, 201 is a second binarization section, 202 is a binarization selection section, 300... Image data output section, 301... Image transmission section, 302...
- Image display section, 303... Image printing section. Figure 2B R (1,0) = R (2,0) = 0 Figure 2C Figure 3 Figure 4A Figure 4B Figure 4C Figure 4D Figure 4E Figure 4F Figure 4G Figure 4H Figure 5A Figure 5B Figure 5D Figure 5C Figure 5E

Claims (4)

[Claims] (1) A first step in which input image data is binarized using an average value based on binarized data near the pixel of interest that has already been binarized.
a second binarization means that binarizes the input image data using a fixed threshold; 1. An image processing apparatus comprising: a selection means for selecting one of the binarization means. (2) The image processing apparatus according to claim 1, wherein the first binarization means includes error diffusion means for diffusing a binarization error. (3) The image processing apparatus according to claim 1, wherein the selection means determines an edge portion based on the binarized data. (4) The image processing apparatus according to claim 1, wherein the average value and the selection signal of the selection means are stored in a storage means using binary data near the pixel of interest as an address.