JPS60236363A - Halftone display method - Google Patents

Abstract

PURPOSE:To display satisfactorily a linear part by discriminating the density distribution of input picture elements for each input matrix and then selecting a proper one of set threshold value matrixes. CONSTITUTION:Plural threshold matrixes are used for conversion of input dot matrix into an output dot matrix. A P value deciding circuit 26 obtains the average gradation of each input picture element with an input dot matrix. Then the prescribed value is added to said average value to obtain the comparison gradation P. This gradation P is compared with the gradation of each input picture by comparators 45-1 and 45-2. Based on these comparison results, the distribution of density is discriminated. Then the optimum threshold value matrix is selected for each input dot matrix, and a gradation distribution is decided for each output dot of an output dot matrix.

Description

DETAILED DESCRIPTION OF THE INVENTION "Field of Industrial Application" The present invention relates to a halftone expression method for digitally expressing halftones.

"Prior Art" For example, in many facsimile machines, image information on a transmitted document is decomposed into pixels and an analog image signal representing the brightness of each pixel is created. This analog image signal is binarized at a predetermined threshold level, and digital image signals corresponding to so-called white and black of the transmitted original are created. On the recording device side, an image is recorded by an array of recording pixels (hereinafter referred to as output dots) corresponding to this digital image signal. In such a recording method, since each pixel is recorded based on a binarized image signal, it is naturally impossible to express halftones.

On the other hand, for example, in a recording apparatus using a thermal head, a halftone expression method has been proposed that can express several levels of halftones. In this halftone expression method,
A recording-side device receives image information representing halftones (hereinafter referred to as halftone information). This thermal head is then controlled in multiple stages to change the amount of thermal energy generated, thereby changing the density or dot diameter of each output dot in thermal recording or thermal transfer recording to express intermediate tones. However, in such halftone expression methods, the density and diameter of the recording pixels are subtly affected by temperature conditions such as the temperature of the substrate of the thermal head and the temperature of the recording paper, and there is a high possibility that density unevenness will occur.

Therefore, if a stable image is to be obtained, only a very small number of gradations can be expressed.

Therefore, a halftone expression method has been proposed that can stably record or display halftone information on a display using a plurality of output dots. These methods are broadly classified into dither methods and density pattern methods. In these methods, an output dot matrix consisting of a predetermined number of output dots is used to express gradations using this as the minimum expression unit.

FIG. 1 shows an example of an output dot matrix 12 in which output dots 11 are arranged in a 2×2 matrix, or in the form of a square. In this specification, a manual dot matrix is defined as a concept corresponding to the output dot matrix. FIG. 2 shows a manually generated dot matrix 14 corresponding to the output dot matrix shown in FIG. Positionally, they each correspond to the output dots 11 in FIG.

In the dithering method for expressing halftones, a threshold level is determined for each input pixel constituting the dot matrix, and the corresponding gradation level of the output dot is determined by this threshold level. In the density pattern method, a pattern corresponding to the gradation of each input dot matrix is determined in advance, and the gradation of the output dot matrix is expressed by this pattern. As described above, in the conventional halftone expression method, the output dot matrix is determined without considering the arrangement of input pixels in the input dot matrix. For this reason, when attempting to record or display halftone information including line segments, such as Chinese characters with a large number of strokes, there is a problem in that the contours of these line segments are distorted and their reproducibility is poor.

For example, suppose there is halftone information as shown in FIG.

Here, the 8 frames are the human dot matrix 1 shown in Figure 2.
It represents 4. The four numerical values shown in the human dot matrix 14 correspond to the four human pixels 1 shown in FIG.
5, and these are divided into 9 levels of gradation (0 to 8
). For example, in the human dot matrix in the upper right corner of Figure 3, three relatively high density input pixels with a gradation level of ``7'' and one human input pixel with a relatively low density with a gradation level of ``work'' are arranged. ing.

Now, assuming that each output dot of the output dot matrix has three levels of gradation that can be expressed, we will first try to express intermediate tones using Bayer type dither, which is a typical dither method. FIG. 4 shows the threshold values used in this case. A threshold level is determined corresponding to each of the four human pixels. Here, if the gradation level of the human pixel is assumed to be ■, and the gradation level of the output dot is 0, 1.2, then a/
These have the following relationship with the threshold level expressed in fractional form of b. However, here, a dot of gradation level 2 refers to a printing dot that almost covers one pixel area, and a dot of gradation level 1 refers to a printing dot whose diameter is considerably smaller than this printing dot. Furthermore, dots at gradation level 0 are dots that are not printed.

I want a ・・・・・・O a ≦rL<b ・・・・・・l IL ≧b ・・・・・・2 Figure 5 shows the result of ternarization using this dither pattern. It is something. In the figure, the larger circle 16 indicates the output dot at gradation level 2, and the smaller circle 17 indicates the output dot at gradation level 1. Comparing it with the halftone information in Figure 3, we see that.

In Figure 3, the gradation level “1” is human/nodomatori/
Nox exists in a rectangular shape, but it can be seen that this is considerably unclear in Figure 5.

Next, the halftone information shown in FIG. 3 will be reproduced using the conventional density no-turn method. In FIG. 3, the sum of the gradation levels of the human-powered pixels in each human-powered dot matrix 14 is "'4".
”, Pa 16” or Pa 22”. In this density pattern method, the gradation level is set for the manual dot matrix 7 x 14 (Figure 6a) where the sum of the gradation levels is “4”. An output dot matrix 12 (see figure b) in which one level 1 output dot is placed in the upper left corner is used. Also, a human dot matrix 14 (see figure b) in which the sum of the gradation levels is "16" is used. For FIGS. 7a-d), one output dot of gradation level 2 is placed in the upper left corner, and two output dots of gradation level 1 are placed in the right half of the output dot matrix 12 ( e) in the same figure.Furthermore, for the human dot matrix (a-Ld in Fig. 8) in which the sum of the gradation levels is "'22", the output corner and throat of gradation level 2 are at the upper left corner. An output dot matrix 12 (e in the same figure) is used, in which output dots of gradation level 1 are arranged in the corners and lower right corner, and in the rest. Fig. 9 shows the result of ternarization using this density pattern method. It is something that

When compared with the halftone information in FIG. 3, it can be seen that the rectangular linear portions are blurred.

"Problems to be Solved by the Invention" In view of the above circumstances, an object of the present invention is to provide a halftone expression method that can satisfactorily reproduce linear portions included in halftone information.

``Means for Solving the Problems'' In the halftone expression method of the present invention, multiple IJ hooks (threshold values for converting from an input dot matrix to an output dot matrix) are prepared, and the density in the manual dot matrix is Let the optimal threshold matrix be selected to approximate the distribution. Then, the optimum threshold matrix is used for each manual dot matrix to determine the gradation level of each output dot of the output dot matrix.

As a result, the density distribution within the output dot matrix is approximated to that of the manually generated dot matrix, so that linear portions can be expressed satisfactorily.

〔Example〕

The present invention will be described in detail with reference to Examples below.

FIG. 10 shows the same halftone information as FIG. 3, which is used as processing data in this embodiment, expressed in coordinates (i, j). FIG. 11 shows four types of threshold matrices (1) to (2) for converting a manual dot matrix having such a 2.times.2 matrix structure into an output dot matrix.

The first threshold matrix ■ is the same as the threshold matrix shown in Figure 4, and has a density pattern in which the top left of the matrix is the highest density position, and the density decreases in the order of bottom right, top right, and bottom left. It will be expressed. Therefore, for a human-made dot matrix 14 with a relatively low density as shown in FIG. become. The second threshold matrix (2) has the highest density at the top right of the matrix, and expresses a density pattern similar to that of the first threshold (ma) rotated 90 degrees to the right. The third threshold matrix ■ has the highest density position at the bottom right of the matrix, and expresses a density pattern as if the first threshold matrix ■ had been rotated 180 degrees. 4th threshold value Ma)
The IJx (■) has the highest density at the lower left of the matrix, and expresses a density pattern similar to that of the first threshold value (Ma) IJx (■) rotated 270 degrees to the right.

In this way, the four types of threshold matrices express different density patterns. For this reason, in the present invention, the density distribution of each manual dot matrix is determined, and the optimum threshold value matrix is selected to determine the gradation distribution of each output dot in the output dot matrix. In order to determine this density distribution, in this embodiment, we first calculate the average tone of each input pixel in the input dot matrix,
A predetermined value is added to this to calculate the comparative gradation P.

Next, this comparison level mP is compared with the gradation level of each manually-powered pixel, and the distribution is determined from the results of these comparisons.

This will be explained in detail next.

Now, suppose that the gradation of each human pixel of the input dot matrix is expressed as follows.

At this time, a comparison gradation P, which is obtained by adding a positive number α to the average value of the gradation of the input dot matrix, is expressed by the following equation.

The determination result of the magnitude relationship between the comparative gradation P determined in this way and the gradation akL of each input pixel (where 1=l or 2) is set as shown in the following equation.

(+) When akL≦P, the judgment result is “0” (II
) When akt>P, the determination result is "1" Table 1 below shows the distribution of the determination results for each human pixel and the type of threshold matrix to be selected.

Table 1 Here, the determination result and the threshold matrix have the following relationship.

(i) Judgment result "1"' is only one among four input pixels
If there is one, the 7 corresponding to this judgment result “1”
The threshold matrix is selected so that the output dot of the highest tone level is located on the 2+x.

For example, if only the determination result for gradation a2□ is 1''',
3 at the lowest value (“l”, “5”) at this corresponding position
The threshold matrix (■) for valorization will be selected.

(11) If there are two determination results of “1” among the four input pixels, a threshold matrix is selected that performs ternarization at the first or second lowest threshold level at the corresponding position. Ru. For example, floor ++l! Ia
When the determination result of 21 and a22 is "1", the threshold matrix ■ corresponding to the two sets of threshold levels ``'1'', ``5'' and ``3'', and ``'7'' is selected. Ru. Further, when the determination result for the gradations all and a2+ is 1°, the threshold matrix ① corresponding to two similar sets is selected.

(iii) If there are three determination results “1” among the four human-powered pixels, 1 to 1 are displayed at the corresponding positions.
A threshold matrix is selected that performs ternarization at the third lowest threshold level. For example, three gradations a12,
When the determination results of a21 and a22 are "1"', the threshold value (ma) IJx (2) is selected in the same way.

(iv) If the determination results for the four input pixels are all equal, the threshold matrix ■ is selected. Of course, other threshold matrices (1), (2), or (C) may be selected.

FIG. 12 shows threshold matrices corresponding to each input dot matrix for the halftone information shown in FIG. 10. However, in this case, the number α of pressures is set to 1. When this number α is increased to 1.1.1.2..., the value of the comparison gradation P that is compared with the gradation akL increases, and the density distribution is determined based on the higher gradation. will be held.

As an example, FIGS. 13 to 16 show the input dot matrix 14 in the case where the sum of the gradations of the input pixels is "16".
This represents the relationship between the output dot matrix 12 and the output dot matrix 12. First, the average gradation P for these manual dot matrices 14 is as follows.

= 5 Therefore, the judgment result is 0" and 2 '1'", and the first
For the input dot matrix (IJ wax 4 in FIG. 3), the threshold matrix ■ is selected. Further, for the human-powered dot matrices shown in FIGS. 14 to 16, threshold matrices (2), (2), and (2) are selected, respectively.

As a result, for the input dot matrix 14 in FIG.
The threshold level is “1”, “5” is “3”
, "7°'," and both determine the output dot of gradation level 2. Also, gradation level "1" (D gradation a 2+
s a a. is compared with the threshold level "4", 8" or "2'', par 6", and the output dot of gradation level 0 is determined for both of them. As a result, the output dot matrix 12 in this case is as shown in the figure. In the case of Figs. 14 to 16, ternarization is performed in the same way, and the input dot matrix 14 (a in the figure) becomes the output dot matrix 12 (b in the figure). Comparing with FIG. 7, it can be seen that each output dot matrix represents a density distribution corresponding to the human dot matrix.

FIG. 17 shows the halftone reproduction state as a result of performing ternarization on the halftone information shown in FIG. 10 using the respective selected threshold matrices. When compared with FIGS. 5 and 9, it can be seen that the line segments are more clearly reproduced.

The above description uses the threshold matrices 1 to 2 shown in FIG. 11, but it is also possible to use other types of matrices as the threshold matrices. For example, the first
The threshold matrices 1 to 2 in FIG. 8 are modifications of the Bayer type threshold matrix shown in FIG. FIG. 19 shows the threshold matrix ① to ① shown in FIG. 11 replaced with four threshold levels D and -D, and these threshold levels D1 to D have the following relationship.

0≦Dz<D2<D3<D4≦n-1-
(3) Here, the integer n is the number of human pixels arranged on one side of the matrix, and in this example, n=
It becomes 2. When the threshold matrices ① to ② of Fig. 18 are expressed using this notation, the result is as shown in Fig. 20. In FIGS. 19 and 20, it can be seen that the highest threshold level D4 and the next highest threshold level D3 are interchanged in a sash-like manner.

FIG. 21 shows another format of threshold matrices 1 to 0 when such a notation is used.

This threshold matrix ① to 0 is called a spiral type. When output dots are created using this threshold matrix ①~0, the deterioration in resolution is further reduced compared to the former two methods, and although the gradation expression is slightly inferior, it is possible to obtain better results in terms of reproducibility of line drawings.

The method of selecting the threshold matrices ① to ＠ for the determination results is the same as the principle shown above. For example, az
>p and a, 2>p, a2t≦p and a22≦p
If so, the threshold value Ma) IJx is selected (Fig. 22).Also, a++>p and a2.>p, and a1□≦
If p and a22≦p, the threshold matrix 12 is selected (FIG. 23). Furthermore a+2>p1a2+>pS
When a22>p and a-th<p, the threshold matrix [phase] is selected (FIG. 24). The same applies below.

The 2×2 matrix has been described above.

Generally, the gradation of an nxn input dot matrix can be expressed as follows as a generalization of equation (1).

(margin below) At this time, each output dot has a value of m (m is a positive integer)
It is assumed that the gradation level can be changed by . The threshold matrix in this case can be expressed as shown in FIG.

Here, it is assumed that each of the m-1 threshold levels in parentheses has the following relationship.

0≦Thz <Tkt2 <...<Tk,,. −
1゜・・・・・・(6) Threshold level group enclosed in parentheses (T□l
l Tkt2 Setting..., threshold level Tk, q (1≦q≦m-
1) assumes that each parenthesis has a different value. For example, the following relationship exists.

Tzl< Tl21 < T131-- < T,,-
+T + 12 < 71゜2 < Tl32 ...kuT
ゎゎ2T ll[l1l-11<Tl2++n-+1
<Tl3cm-B...<Tn,, (m-
(2) In such a case, the present invention first calculates the comparison gradation P.

Then, the magnitude relationship between each floor glWakt and the comparison tone P is determined, and based on this, the threshold matrix shown in equation (5) is rearranged. At this time, the threshold level group enclosed in parentheses (Thz + Tkt2゜...
..., the arrangement is changed within the nxn matrix using these threshold level groups as a unit without changing Tkt(m-1]) itself.

Based on the threshold matrix thus set, m-value conversion is performed for each human pixel of the input dot matrix. In other words, it is as follows.

(!) When akt < Tht, gradation level 0 is output.

(i!) k [When Q-11≦akt<TkLq, output the gradation level q.

(!ii) When Tkt fm-ll < akt, output the gradation level (m-1).

FIG. 26 shows the main parts of a recording device for realizing the halftone expression method in a 2×2 matrix described above. Image data 22 is supplied to an input buffer multiplexer 21 of this device. The image data 22, for example, as partially shown in FIG. 3, is data in a format in which halftone information consisting of a large number of input dot matrices is manually read out serially line by line (raster) by pixel. The input buffer multiplexer 21 sequentially distributes the image data 22 line by line to a manual buffer 23 that is composed of line buffers for N lines (N is an integer of 4 or more). The input buffer 23 is composed of a line buffer for four lines or more, because it processes the data corresponding to the input dot matrix using the image data 22 for two lines, and also uses separate memories for data writing and reading. This is to improve the processing speed by performing this in the area.

First and second input multiplexers 24 to 1.2
4-2 sequentially selects the line buffers in the manual buffer 23 and reads out the already written data in parallel. The two adjacent lines of image data 25 to L 25-2 thus obtained are supplied to a P value calculation circuit 26, where the average gradation P is calculated.

FIG. 27 specifically shows this P value calculation circuit. One image data 25-1 is sent to the multiplexer 26
-1 is alternately supplied to the first or second serial-to-parallel converter 27-1, 27-2 for each pixel. The other image data 25-2 is similarly allocated to the first or second serial-to-parallel converter 28-1, 28-2 by another multiplexer 26-2. two first serial-to-parallel converters 17-1, 18-1
Then, gradation data 2911 and 2921 for one manual pixel each delayed by one crotter are supplied to the adder 31. At this time, the other two second serial-to-parallel converters 17-2 and 18-2 each supply gradation data 29 and 2.2922 for one pixel manually to the adder 31. That is, the adder 31 inputs the human-powered dot matrix (7)4-') (7) human-powered pixel gradation a++, a shown in equation (1).
+2, data indicating a21sa22 will be supplied.

The adder 31 receives these gradations a++, a12, and a2+.

The sum of a22 is determined and the addition result 32 is input to the buffer 33. The buffer 33 is constituted by a shift register, and by shifting the contents thereof, an average value is determined and average value data 34 is created.

The average value data 34 is input to an adder 35 and added to a positive number α supplied from an α value buffer 36. In this way, the comparison gradation P shown in equation (2) is calculated. The comparison gradation data 37 representing the comparison gradation P is supplied to the first to fourth comparison circuits 38-1 to 38-4, where the four gradation data 2911.
A comparison is made with 2912.2921.2922. The determination results 39-1 to 39-4 each become a signal "o" when akl≦P, and become a signal "1" otherwise.

The 4-bit judgment result 39 is used as address information of the threshold table ROM 41 shown in FIG.
The selection of the threshold matrix is done according to the table. Accordingly, the threshold table ROM 41 stores threshold level data 4211.42, 2.422 for each input pixel.
Read 0.4222. Threshold level data showing these threshold levels 42°1.421
2.42°1.4222 is the corresponding threshold buffer 43-
1 to 43-4.

Two threshold multiplexers 44-1 and 44-2 are provided after the threshold buffers 43-1 to 43-4 to select the threshold level of the input pixel for each line. The first threshold multiplexer 44-1
Threshold level data are alternately selected from the threshold buffers 43-L 43-2, and the first comparison circuit 45-1
pass this on to. The first comparison circuit 45-1 compares the delayed image data 47-1 obtained by delaying the image data 25-1 with the first delay circuit 46-1 at this threshold level to determine the gradation level of the output dot. . The gradation level data 48-1 representing one of the three gradation levels from 0"' to 2" for each output dot is sent to two output buffers 51-1.
52-1 in line units. The same applies to the second threshold multiplexer 44-2, which alternately selects threshold level data from the third and fourth threshold buffers 43-3 and 43-4, and selects the threshold level data alternately from the third and fourth threshold buffers 43-3 and 43-4.
The delayed image data 47-2 delayed in step 2 is compared with the second comparison circuit 45-2.

The thus created gradation level data 48-2 is written line by line to the other two output buffers 51-2 and 52-2 via the output write multiplexer 49-2.

The two lines of gradation data written in this way are sent to the first and second output Huffer multiplexers 5.
3-L Selected by 53-2. Further, the output sending multiplexer 54 provided on the output side alternately selects the output huffer multiplexers 53-L and 53-2 for each line, and generates image data consisting of a data string in the same format as when the original is scanned in a plane. Outputs 55.

Of course, this image data 55 is a data string representing each output dot in three gradation levels.

The image data 55 is supplied to a thermal head drive circuit 57 in the recording section 56, where it is converted into a signal format suitable for driving a thermal head 58 in the recording section 56 as a thermal transfer recording device. The thermal head 58 is of a line recording type, and an ink donor sheet 62 is fed out from a supply roll 61 and a recording paper 63 fed out from a supply tray (not shown) is passed between a pressure roll 59 disposed opposite to each other. They are designed to pass in an overlapping state. The thermal head changes the time width of the applied pulse applied to the unit heating element in two stages according to the gradation level of the output dot, and changes the diameter of the output dot by changing the amount of ink transferred due to the difference in thermal energy. Change in two stages.

As a result, one of the many gradations is selected for each output dot matrix, and the desired image information as a whole is expressed in halftones. Of course, the recording section 56 shown in FIG. 26 may be a device section that performs recording using another recording method, or may be a display section that displays an image using a magnetized latent image or the like.

"Effects of the Invention" As explained above, according to the present invention, the input dot mark) I
We determined the density distribution of human pixels for each J sox and selected an appropriate one from among the preset threshold matrices based on this, so the selection of the matrix prepared as the threshold matrix allows us to express the characteristics of midtones. It can be made to stand. Further, according to the present invention, it is possible to improve the reproducibility of line drawings and to reduce blurring of the contours of figures.

[Brief explanation of the drawing]

Fig. 1 is an arrangement diagram showing an example of the arrangement of output dots in the output dot matrix, Fig. 2 is an arrangement diagram showing an example of the arrangement of input pixels in the input dot matrix, and Fig. 3 is an arrangement diagram showing an example of the arrangement of input pixels in the input dot matrix. An explanatory diagram showing an example. Fig. 4 is an illustration of the arrangement of threshold levels in a Bayer dither matrix. Fig. 5 is an explanatory diagram showing an example of halftone expression using a conventional dither pattern. Fig. 6 is an illustration of the gradation level. Input dot matrix and output dot matrix according to the density pattern method when the sum of gradation levels is "4") An explanatory diagram showing the IJ hook correspondence relationship, Figure 7 shows the same correspondence using the density pattern method when the sum of the gradation levels is "16" FIG. 8 is an explanatory diagram showing a similar correspondence relationship when the sum of gradation levels is 22'. FIG. 9 is an explanatory diagram showing an example of halftone expression by this density pattern method. Figures 10 to 17 are for explaining one embodiment of the present invention. Among these, Figure 10 is an explanatory diagram showing halftone information in correspondence with coordinates, and a diagram for realizing a tone expression method. A block diagram showing an example of the device, FIG. 27 is a block diagram of a P value calculation circuit in this device. 11... Output dot, 12... Output dot matrix, 14...
... Input dot matrix, 15 ... Human pixel, 26 ... P value calculation circuit, 41 ... Threshold table ROM. 45... Comparison circuit. Applicant Fuji Xerox Co., Ltd. Agent Patent Attorney Umeo Yamauchi Figure 12 Figure 13 Figure 15 Figure 17

Claims (1)

[Claims] The number of gradations expressed by the output dots used for recording or display is smaller than the number of gradations expressed by each pixel on the input side corresponding to these output dots, and the predetermined number of output dots are arranged in a predetermined matrix. In a halftone expression method that uses arranged output dot matrices as a unit to express more tones than the tones that can be expressed individually of the output dots, each output in the output dot matrix (output dot matrix) Prepare multiple IJ hooks (threshold values for determining the gradation levels of dots), manually determine the density distribution of input pixels for each dot matrix, select a threshold value matrix from this determination information,
A halftone expression method characterized in that the gradation level of an output dot for each input pixel of the input dot matrix is determined by the selected darkness value matrix, thereby expressing a halftone.