JP3341174B2 - Binarization processing method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to the conversion of analog graphic image data and analog character image data into single-bit binary
The present invention relates to a binarization processing method for converting image data into image data.

[0002]

2. Description of the Related Art Generally, a binary (1,0) image (binary image) for obtaining a binary (1,0) image from a multivalued image (analog image, analog image) such as a grayscale image or a color image. As the binarization processing method, there are a simple binarization processing method and a floating binarization processing method.

The former simple binarization processing method is a method of performing a binarization by setting a threshold value (threshold level) (fixed to a constant value). For example, FIG. 13 shows the binarization with the threshold set to 96/255, and FIG. 14 shows the binarization with the threshold set to 64/255. When the size is large, crushing is likely to occur as shown in FIGS. 12 and 13, and this crushing is particularly remarkable in a character having a large number of strokes. Conversely, when the threshold value is small, blurring tends to occur as shown in FIG. 14, and this blurring is particularly noticeable in characters having a small number of strokes.

In order to solve such a problem, 2
Adjustment of the threshold value at the time of the binarization process can cope to some extent, but crushing and blurring are in a trade-off relationship, and in order to determine the optimal threshold value, the shading volume, contrast (contrast, A light and dark volume is required.

On the other hand, the latter floating binarization processing method is a method in which the threshold value is sequentially changed from the brightness (pixel density) around the target pixel to be binarized, and the binarization is performed.
As shown in FIG. 15, good binarization processing can be performed.

[0006] However, in this floating binarization processing method, the gradation (where gradation is a contrast of the degree of shading consisting of tone, black and white, and halftone) as the level of brightness in the visual sense. ) Because it is a faithful representation of
If the background of the character is colored rather than a white background, the background is also faithfully binarized, resulting in a binarized image as shown in FIG. In FIG. 16, the binarized background portion is indicated by hatching for convenience of illustration.

When an image as shown in FIG. 16 (an image in which the background is binarized) is used, for example, as base data for character recognition, the background portion shown by hatching in FIG. 16 is merely a noise component. Therefore, it becomes a factor to lower the recognition rate.

In order to correct the above-mentioned background binarization,
By adjusting the threshold value by the simple binarization processing method,
The binarized background can be cut to some extent. In this case, however, a volume such as a light and shade volume and a contrast volume is required. Can not be.

In the above-mentioned floating binary processing method,
When the color of the character is light, the contrast between the background and the character is too weak (the gradation difference is too small), so that the floating binary feature cannot be fully utilized, and as shown in FIG. There was a problem that caused.

On the other hand, as a method of removing background data after the above-mentioned binarization processing, an isolated point removal method is known. In this isolated point removing method, as shown in FIG. 18, for example, a block of 3 × 3 (9 pixels) is created from image data after binarization processing, and black data (FIG.
, Black data is shown by hatching for convenience of illustration), the black data at the center is removed as an isolated point, and 3 × 3 (9 pixels) other than the center is removed.
If there is black data in the periphery of the block of
This is a method of prohibiting removal because there is a possibility that the data is continuous outside the frame.

However, in this isolated point removing method,
In this method, a memory for storing image data after binarization processing is required, and a 3 × 3 (9 pixel) block is formed on all image data, an isolated point is detected, and the isolated point is removed. Therefore, there is a problem that the circuit scale becomes large in the hardware structure, and the processing time becomes long in the software processing.

In the conventional various conventional methods described above, when a changeover switch for selecting a grayscale volume, a contrast volume, and a binarization process is provided, the user must operate the volume and the changeover switch. The appropriate selection of the changeover switch cannot be an easy operation for the user. For this reason, erroneous selections are often made, and it is necessary to retake images when such erroneous selections are made. In addition, when used for character recognition, etc., it is impossible to determine whether or not an appropriate image is obtained until a recognition result is obtained, and it is necessary to display an image in order to determine whether or not image data is appropriate. Therefore, it is difficult for the user to determine whether or not the image is an appropriate image, even if the image is displayed on the display device, in addition to the complexity of the device. There are various practical problems.

As a prior art for solving the above problems,
There is an automatic shading correction method. This automatic shading correction method is
Store all image data to be valued in memory, and create a frequency distribution table for each density (brightness) of the image data,
As shown in FIG. 19, the densities are detected from the highest density (pure black) and the lightest density (pure white) to a few percent of the total number, for example, from pure white and pure black to reach 2% each. The dynamic range (dynamic range, operating range) is expanded (amplified) as dark and lightest density,
This is a method for performing a correction calculation for contrast adjustment.

However, there are the following problems in this automatic shading correction method. That is, since a memory for temporarily storing image data and a pre-scan are required, real-time processing becomes impossible, and a multiplier, a CPU, and the like are required for calculations for expanding a dynamic range, a hardware configuration and a gate are required. There is a problem that the number of circuits increases and the cost increases.

[0015]

SUMMARY OF THE INVENTION The present invention does not require the use of a contrast volume, a contrast volume, etc. for determining an optimum threshold value, and regardless of the presence or absence of a background and the contrast of a character. It is an object of the present invention to provide a binarization processing method which can binarize only characters and graphics (chart) with a simple circuit configuration without crushing and blurring.

[0016]

SUMMARY OF THE INVENTION The present invention provides a binary image
The original image data to be converted to data, 該画
A density detector for detecting the presence of image data in each gradation area in the image data; and a target image from each gradation area of the image data.
Image characteristics divided by defining the density distribution characteristics
Mode is set, and the density is detected by the concentration detection unit.
A determination unit that determines the density distribution of the target image from the gradation range of the image data obtained, and classifies the image data into the respective modes based on the determination; and a mode determined by the determination unit.
Cuts the data in the predetermined gradation range of the image data according to
The remaining image data to the original data range.
That data correction unit and a mode based on determined in the determination unit
Correction data of the data correction unit selected by Zui or,
A binarization processing unit for performing binarization processing on normal data, and a binarization processing method for converting analog image data into binarized image data.

[0017]

According to the present invention, the above-mentioned density detecting section detects the presence of image data in each gradation area, and the determination section determines the target image from the gradation area of the image data detected by the above-mentioned density detecting section. In addition to determining the density distribution, a case classification is performed for each mode, and the data correction unit determines the image data according to the mode.
After cutting the data in the specified gradation range, the remaining image data
Data is expanded to the original data range and corrected. Accordingly
The above-described binarization processing unit selects based on the mode.
Serial data correction of the correction data or normal data,
According to the determined mode, for example, floating 2
The analog image data is converted into binary image data by performing a binarization process by a binarization process or a simple binarization process.

[0018]

As a result, there is no need for a contrast volume, a contrast volume, or a changeover switch for determining an optimum threshold value as in the prior art. Irrespective of the above, there is an effect that binarization can be performed with a simple circuit configuration without crushing or blurring only characters and figures used as target.

[0019]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the present invention will be described below in detail with reference to the drawings. FIG. 1 shows a binarization processing device used in a binarization processing method. A 7-bit A / D converter 11 converts analog pixel data into 7-bit digital data. Following the above 7-bit A / D converter 11, the density detector 2
2 (see FIG. 5), a density detection circuit 12 including a counter circuit 34 (see FIG. 6) and a determination unit (see FIG. 9), and a data correction unit 13.

In this embodiment, the dynamic range of the pixel data is divided into eight gradation ranges "0" shown in FIG.
It is classified into “1”, “2”, “3”, “4”, “5”, “6”, and “7”, and as shown in FIG. , The dark part 3/8 is set in a very dark gradation area B (area B), and the light part 3/8 is set in a relatively light gradation area C (area C). The presence of pixel data in the adjustment areas A, B, and C is detected in real time, and the density distribution of the target image is determined from the gradation areas A, B, and C where the pixel data is detected. Are configured so that the cases are classified into the modes 1 to 4 according to the determination conditions shown in Table 1.

[0021]

[Table 1]

The above-mentioned division into the modes 1 to 4 is executed as follows. That is, the gradation area A is true (YE
S), when the gradation range B satisfies both true (YES), mode 1, the gradation range A is true (YES), and the gradation range B is false (NO).
When both are satisfied, the mode 2, the gradation area A is false (NO),
When the gradation range C satisfies both true (YES), mode 3,
When the gradation area A satisfies both false (NO) and the gradation area C satisfies both false (NO), the above-mentioned gradation areas A,
The state of the image is decoded (decoded) from B and C.

The above-mentioned data correction unit 13 has a 7-bit A / D
The pixel data that has been converted from analog to digital by the converter 11 is constantly corrected to three types of correction data in accordance with the data correction contents in Table 1 described above.

That is, in modes 1 and 4, a normal mode is used in which no correction is performed. In mode 2, the darker half of the pixel data value is cut (to zero) and the remaining brighter half is removed. Data correction for doubling the data value of 2 to return to the original range was executed. In mode 3, the brighter 3/4 of the pixel data value was cut (to the maximum brightness value) and remained. Data correction is performed to return the data of the darker quarter to four times the original range.

An example corresponding to the above-mentioned modes 2 and 3 is shown in FIGS. FIG. 3 shows a data correction algorithm (algorithm) corresponding to mode 2, in which the darker half of the image data of 0 to 127 points, that is, 6
Anything less than 4 points is regarded as zero, and the remaining brighter half
The range of the data is doubled and returned to the original range. In binarization, the correction data is converted into 5 bits and used. That is, the 7-bit data D ₇ , D ₆ , D ₅ , D
_{4, D 3,} D 2, using the data _D 6 to D ₂ in the case of converting _{D 1} to 5 bits, this _time, at the time of _D 7 = 0 and all the 5-bit data _D 6 to D ₂ zero .

Similarly, FIG. 4 shows an algorithm for data correction corresponding to mode 3, in which the brighter 3/4 of the image data of 0 to 127 points, that is, the data of 31 points or more is cut (to 127 points). The remaining dark 1/4 data is expanded to 4 times the range and returned to the original range. In binarization, the correction data is converted into 5 bits and used. That is, the 7-bit data D ₇ ,
When D ₆ , D ₅ , D ₄ , D ₃ , D ₂ , and D ₁ are converted to 5 bits, when D ₇ = D ₆ = 0, the data D _{5 to} D ₁ are 5 bit data, and the 7 bit data D _7, of the D _6, and all 5-bit data 127 points at the time of either one nor "1". In the normal mode data correction algorithm, when converting 7-bit data to 5-bit data, D ₇ to D ₁ are ignored ignoring D ₂ and D _{1.
It is} assumed that 5 bits of ₃ are converted into 5-bit data.

In FIG. 1, a 5-bit conversion circuit 14 is connected to the next stage of the data correction unit 13 described above. This 5-bit conversion circuit 14 is a binary conversion circuit 18 at the subsequent stage.
Is a circuit for converting 7-bit data to 5-bit data in order to perform processing with 5 bits. In other words, the above-described data correction unit 13 has data having four times the gradation of the data to be subjected to the binarization processing. In other words, in response to performing the binarization processing using 5-bit data,
By providing ⁷⁻⁵ = 2 ² = 4 times gradation, correction data is calculated from data of 4 times gradation, and data is corrected without impairing gradation. .

A data selection circuit 15 is connected to the next stage of the 5-bit conversion circuit 14 described above. The data selection circuit 15 is a circuit that selects one of three types of pixel data, i.e., normal, only the upper half, and only the lower quarter, according to the mode determination signal m from the density detection circuit 12. To the next stage of the data selection circuit 15, a floating binarization threshold value setting circuit 16 and a binarization circuit 18 are connected.

Here, the above-mentioned floating binarization threshold setting circuit 16 has a peripheral density calculator 19 for calculating the peripheral density of the pixel to be binarized, and a threshold calculator for calculating the floating binarization threshold. 20 and a floating binary threshold is set corresponding to the above-described modes 1, 2 and 3. Further, the simple binarization threshold value setting circuit 17 provides a threshold value which is deviated to a darker direction, for example, the threshold value = 3 corresponding to the above-described mode 4.
Set H. Specifically, the threshold value = 3H means that the value of 5-bit data is set to 0 when the value of the 5-bit data is 0 to 2 points.
The point indicates that it is binarized to 1.

A threshold selection circuit 21 is connected to the next stage of each of the threshold setting circuits 16 and 17 described above. This threshold value selection circuit 21 uses a floating binarization threshold value corresponding to mode 1, mode 2, mode 3 according to the mode determination signal m from the density detection circuit 12, or to a darker one corresponding to mode 4. Select whether to use a biased simple binarization threshold. Then, the binarizing circuit 18 connected to the next stage of the threshold value selecting circuit 21
The binarization process is executed according to the threshold value selected in.

FIG. 5 shows a specific circuit configuration of the density detecting section 22 constituting a part of the density detecting circuit 12 shown in FIG.
A total of six NOT circuits 23 to 28 as a NOT circuit;
A total of two OR circuits 29 and 30 as an OR circuit;
A total of three NAND circuits 31 as a NAND circuit,
32, 33, and the upper three bits of signals AD5, AD6, A / D converted by the 7-bit A / D converter 11.
A logical operation is performed based on the AD7 input, and an area B signal in which data darker than the output of the NAND circuit 31 exists
This is a decoding circuit that obtains an area C signal in which data that is considerably brighter than the output of the NAND circuit 32 exists and an area A signal in which data that is much brighter than the output of the NAND circuit 33 exists (in FIGS. Area A,
The signals B and C correspond to the signals with bars in the figure).

FIG. 6 shows a counter circuit 3 interposed between the density detecting section 22 shown in FIG. 5 and the judging section 47 shown in FIG.
The counter circuit 34 includes three AND circuits 35, 36, and 37 as a NAND circuit, and a total of nine AND circuits.
When the signals shown in FIG. 5, that is, the area A signal, the area B signal, and the area C signal are respectively confirmed four times, the existence of the area signal is determined. It is configured to perform density detection that is not easily affected by noise or the like.

When the area A signal containing very bright data is confirmed four times, the condition confirmation signal CONA of the area A is confirmed from the flip-flop circuit 40 and the area B signal containing dark data is confirmed four times. In this case, when the condition determination signal CONB for the area B is confirmed from the flip-flop circuit 43 four times, and when the area C signal having relatively bright data exists is confirmed four times, the condition determination signal CONC for the area C is determined from the flip-flop circuit 46. Are output.

However, flip-flop circuits 38, 39, 40 for outputting the condition determination signal CONA of area A
And flip-flop circuits 44, 45 and 46 for outputting the condition determination signal CONC of the area C clear the respective clear terminals with the CLR1 signal, and for outputting the condition determination signal CONB of the area B. Reference numerals 41, 42 and 43 clear the respective clear terminals with the CLR2 signal. The timing of each signal of these CLRs 1 and 2 is as shown in FIG. 20, for example, "tree" shown in FIG.
When one character is scanned for one line, the CLR1 signal is output for each line of pixel data, and the CLR2 signal is output for one line.
Output once per scan.

This is thought to be due to the fact that there is a lot of bright part (background) data.
By clearing 8 to 40 and 44 to 46, the influence of dirt and the like is reduced, and there are some lines in which the dark part (character) data is small. Therefore, detection is continuously performed for one scan to obtain certainty.

7 and 8 show the sampling range of the target pixel. As shown in FIG. 7, when the number of pixels in one line is 126, only the central 64 pixels are sampled.
When the reading width is 8 mm as shown in FIG.
Only mm is the search range. That is, the above-described density detection circuit 12, specifically, the density detection unit 22, does not perform the density detection for all the pixels in the main scan, but performs the density detection for only the pixels in the central portion. By not targeting the part and the end part, density detection that is not easily affected by noise or background change is performed.

FIG. 9 shows a case where the density distribution of the target image is determined from the gradation areas A, B, and C in which the image data is detected, and the cases are classified into mode 1, mode 2, mode 3, and mode 4. 4 shows a specific circuit configuration of the determination unit 47, and the determination unit 47 includes a total of four NANDs as a NAND circuit.
Circuits 48, 49, 50, 51 and one as an AND circuit
Each of the output signals from the counter circuit 34 shown in FIG. 6 is used as an input signal of this determination unit 47, and the respective condition determination signals CONA, CONB, CO
A logical operation is performed based on the input of the NC and each condition determination NOT signal (a signal in which a bar is added to the condition determination signal in FIG. 9), and the respective modes are classified.

In this embodiment, each of the gate circuits 48 to
52, NOTMODE4, NOTM
ODE3, NOTMODE2, NOTMODE1, NO
A mode signal of TMODE23 (however, each mode signal is indicated by a bar in FIG. 9) is obtained. This determination condition is as shown in [Table 1] described above.

FIG. 10 shows the data correction unit 13 shown in FIG.
FIG. 10 shows a specific circuit configuration of the 5-bit conversion circuit 14 and the data selection circuit 15, and FIG.
3 is a total of five AND circuits 53 to 5 as AND circuits
7 and a total of six OR circuits 58 to 6 as OR circuits.
3 and the above-mentioned AND circuit block is used for the upper half correction corresponding to the mode 2 and the above-mentioned OR circuit block is used for the lower quarter correction corresponding to the mode 3.

In FIG. 10, the subsequent 5-bit conversion circuit 1
4 and the data selection circuit 15 include two selector circuits 6
4, 65, and two selector circuits 66, 67,
7 bits are converted into 5-bit data, and 5-bit data selected (selected) according to the mode is output. That is, in these two circuits 13, 14, and 15, the 7-bit A / D converter 11 performs A / D conversion.
Bit signals AD1, AD2, AD3, AD4, AD5
After correcting AD6 and AD7 to normal, upper half, and lower quarter, data DATA0, DATA1, DATA2, DATA3, which are bit-converted into 5-bit data
It is configured to output DATA4.

The above-described selector circuits 64, 65, 6
6, 67 logic outputs are high at the SEL terminal.
Is input, the output of the Y terminal is the same as that of the B terminal, and when a low (Low) signal is input to the SEL terminal, the output of the Y terminal is the same as that of the A terminal.

A processing method for binarizing analog image data using the binarization processing device configured as described above will be described in detail below. The analog image data is converted into 7-bit digital data by the 7-bit A / D converter 11 shown in FIG. 1, and the converted 7-bit digital data is output to the next-stage density detection circuit 12 and data correction unit 13.

In the density detecting section 22 (see FIG. 5) of the density detecting circuit 12, the upper three bits of the signals AD5, A
D6 and AD7 are logically operated, and among the gradation areas classified into a total of eight gradation areas shown in FIG. 2, a very bright gradation area A, a very dark gradation area B, and a relatively bright gradation area. It is detected which data of the tuning range C is present, and an area A signal, an area B signal, and an area C signal are taken out from the output stage of the density detecting section 22 in accordance with the presence.

Each of the above-mentioned area signals is output to a counter circuit 34 shown in FIG. 6. Only when the area signal is confirmed four times, the condition determination signal CONA, which determines the existence of the area signal, is output. CONB, CO
Output NC.

Each of these condition determination signals CONA, CON
B, CONC, and each condition determination NOT signal are output to the determination unit 47 (see FIG. 9) at the next stage, and the determination unit 47 classifies the density distribution into modes 1 to 4.

The data correction section 13 corrects data corresponding to Modes 1 to 4. That is, in the mode 1 and the mode 4, the correction is not performed in the normal mode. In the mode 2, the darker half of the pixel data value is cut (to zero), and the remaining brighter half data is removed. Data correction for doubling the value to return to the original range is executed.
/ 4 is cut (to the brightest pixel value) and the data value of the remaining darker quarter is quadrupled to return to the original range.

The correction data (including normal data) is bit-converted from 7 bits to 5 bits by a 5-bit conversion circuit 14 in the next stage, and a data selection circuit 15 in the next stage is a density detection circuit 12, specifically, According to the mode determination signal m from the determination unit 47, only the normal, upper half,
One is selected from three types of correction data of only the lower 1/4.

That is, as shown in Table 1 above, in the case of mode 1 and mode 4, normal data is selected,
In the case of mode 2, the upper 1/2 data is selected, and in the case of mode 3, the lower 1/4 data is selected.

In this manner, the above-described data selection circuit 15
The threshold value is set by each of the threshold value setting circuits 16 and 17 in the next stage based on the corrected image data selected in the above.
In mode 1, mode 2 and mode 3, the floating binarization threshold setting circuit 16 sets the threshold for floating binarization, and in mode 4, the simple binarization threshold setting circuit 17
As a result, a threshold value for simple binarization that is deviated toward the dark side, for example, a threshold value = 3H is set.

Further, in the threshold selection circuit 21 at the next stage,
Thresholds corresponding to modes 1 to 4, ie, a floating binarization threshold and a simple binarization threshold, are selected by mode determination signal m. A binarization process is performed by the binarization circuit 18 of the stage.

In short, there is no correction for image data (corresponding to mode 1) having both opposite gray scale areas of a very bright gray scale area A and a very dark gray scale area B shown in FIG. After the floating binarization threshold value is selected under the condition, the binarized image data having the very bright gradation area A and not having the very dark gradation area B shown in FIG. (Corresponding to mode 2), since the image data is deviated to the bright side, the darker half of the image data, that is, the darker half of each pixel value is cut (reduced to zero) and remains. Bright one 1
/ 2 is doubled, data is returned to the original range, and after the floating binarization threshold value is selected, binarization is performed and there is no very bright gradation area A shown in FIG. And image data having a bright gradation range C (corresponding to mode 3)
, Since the image data is biased to one side (for example, when the background of the character graphic is colored instead of a white background), the brighter 3/4 of the image data is cut (to the maximum brightness value). The data is returned to the original range by quadrupling the remaining dark 1/4, and after the floating binarization threshold is selected, it is binarized and very bright as shown in FIG. For image data that does not have both the gradation area A and the light gradation area C (corresponding to mode 4), the image data is shifted to the dark side, so that the image data is simply After setting the binarization threshold value to a threshold value that is biased toward darker, binarization is performed.

As described above, the target is used without any need for the density volume, contrast volume, changeover switch, and the like, which are found in the conventional method, and irrespective of the presence or absence of the background and the density of the characters and figures. There is an effect that binarization can be performed with a simple circuit configuration without crushing or blurring only characters and figures. Specifically, circuits other than the binarization circuit can be configured with about 100 gates.

Further, as described in claim 3, when the presence of image data corresponding to each mode is counted by the counter circuit 34 and the existence of the image data is determined when the number of pixels exceeds a predetermined value, noise and the like are affected. It is possible to perform concentration detection that is difficult to be performed.

[0054] As further shown in claim 4, in a concentration detection unit 22, when the main scanning head portion and the end portion concentration detected for only the pixels in the central portion except for being affected by changes in the noise or background Difficult density detection can be performed.

[0055] Furthermore a concentration detection unit 22 as shown in claim 5, 6, the above-described A, B, when configured to detect the presence of data in three or more gradation region and C, comparing a decode circuit It can be easily configured.

[0056] Also as shown in claim 7, 8, when performing binarization processing by 5-bit data at least twice (example of data to be binarized in the data correcting section 13 described above, the data If the correction unit 13 includes data having a gradation characteristic of 7 bits (4 times), data correction can be performed without deteriorating the gradation characteristic.

Further, as described in claim 2 , the density detector 22, the determiner 47, the data corrector 13, the binarization circuit 1
If the processing in step 8 is performed simultaneously with the reading of the image, the processing time of the software processing can be reduced. Note that before the binarization processing of the present embodiment , the removal of the isolated points of the image data may be performed in advance.

[0058] Further, when the separate clear signal CLR1,2 between the bright concentration of the counter and dark levels of the counter according to claim 9 are shown as counter circuit 34 (see FIG. 6), i.e., dark concentration (characters) If the detection is set to one scan, the detection is continuously performed during that period, so that the occurrence of blurring of the character details can be suppressed. If the detection of the light density (background) is performed for each line, the detection range of the light density Is sufficiently wide so that it is clear for each line, and is less susceptible to the effects of background stains and the like, and more accurate density detection is possible.

FIG. 11 shows another embodiment of the binarization processing device used in the binarization processing method. In the embodiment of FIG. 1, the floating binarization threshold value setting circuit 16 is used. In this embodiment, a simple binarization threshold setting circuit 68 is used in place of the circuit 16, and the simple binarization threshold setting circuit 68 uses a threshold substantially in the middle of the dynamic range, for example, a threshold. It is configured to set the value = 10H. Also in the binarization processing method using the binarization processing device configured as described above, the same operation and effect as in the previous embodiment can be obtained, and therefore, in FIG. 11, the same parts as those in FIG. The reference numerals are attached and the detailed description is omitted.

Although the binarization processing devices shown in FIGS. 1 and 11 are all constituted by digital circuits, they may be constituted by analog circuits. That is, instead of using the 7-bit A / D converter shown in FIGS. 1 and 11, analog data is compared by a comparator, and areas A, B, and C are recognized by voltage, and the recognized area signal is compared. Is held and then decoded and mode division is performed, while the analog image signal is data corrected according to the identified mode.

In this case, in the modes 1 and 4, the normal mode in which no correction is performed is used. In the mode 2, the dark half of the image data is subtracted, and the remaining bright half is subtracted.
Data correction is performed to amplify 2 to a double range, and in mode 3, data correction is performed to amplify image data to a quadruple range. Then, the identified data is selected by an analog multiplexer. Then, a binarization process may be performed.

In another embodiment for further performing data correction, data correction is executed by changing the reference voltage of the A / D converter instead of the method of correcting the data itself as in the above embodiments. Is also good.

That is, corresponding to the determined mode,
In mode 1 and mode 4, the reference voltage of the A / D converter is not changed and is used without change. In mode 2, the L reference voltage (refL) is changed to refL + (refH-ref).
L) / 2, in mode 3, the H reference voltage (refH) is changed to refH + (refH-refL) / 4, and an analog multiplexer may be used to select the above-mentioned reference voltage.

In correspondence between the configuration of the present invention and the above-described embodiment, the density detecting section of the present invention corresponds to the density detecting section 22 (see FIG. 5) of the embodiment, and similarly, the determining section determines The binarization processing section corresponds to the binarization circuit 18, the means for counting the presence of image data corresponds to the counter circuit 34, and the data correction section corresponds to
Although it corresponds to the data correction unit 13 (see FIGS. 1 and 10),
The present invention is not limited only to the configuration of the above embodiment.

[Brief description of the drawings]

FIG. 1 is a block diagram of a binarization processing device used in a binarization processing method of the present invention.

FIG. 2 is an explanatory diagram for classifying image data into all eight gradation ranges.

FIG. 3 is an explanatory diagram of upper half correction.

FIG. 4 is an explanatory diagram of lower 1/4 correction.

FIG. 5 is a circuit diagram illustrating a concentration detection unit.

FIG. 6 is a circuit diagram illustrating a counter circuit.

FIG. 7 is an explanatory diagram showing a sampling range of a target pixel.

FIG. 8 is an explanatory diagram showing a search range with respect to a reading width of a target pixel.

FIG. 9 is a circuit diagram illustrating a determination unit.

FIG. 10 is a circuit diagram showing a data correction unit, a 5-bit conversion circuit, and a data selection circuit.

FIG. 11 is a block diagram showing another embodiment of the binarization processing device.

FIG. 12 is an explanatory diagram showing collapse occurrence by a conventional simple binarization processing method.

FIG. 13 is an explanatory diagram showing occurrence of slight collapse by a conventional simple binarization processing method.

FIG. 14 is an explanatory diagram showing occurrence of blurring by a conventional simple binarization processing method.

FIG. 15 shows an example of a proper floating-point binarization processing method.
Explanatory drawing which shows value conversion.

FIG. 16 shows a case where the background is 2 by a conventional floating binarization processing method.
Explanatory drawing which shows the defect which is valued.

FIG. 17 is an explanatory diagram showing occurrence of blurring by a conventional floating binary processing method.

FIG. 18 is an explanatory diagram showing a conventional isolated point removing method.

FIG. 19 is an explanatory diagram showing a conventional automatic shading correction method.

FIG. 20 is a time chart of the CLR1 and CLR2 signals of the counter circuit.

[Explanation of symbols]

 13: Data correction unit 18: Binarization circuit 22: Density detection unit 34: Counter circuit 47: Judgment unit

──────────────────────────────────────────────────続 き Continuation of the front page (56) References JP-A-1-218166 (JP, A) JP-A-1-237142 (JP, A) JP-A-4-159863 (JP, A) JP-A-1 126876 (JP, A) JP-A-62-186667 (JP, A) JP-A-62-107572 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) H04N 1 / 40-1 / 409 H04N 1/46 H04N 1/60

Claims (9)

(57) [Claims] (1)Original image to be converted to binary image data
For image data , In the image dataIn each gradation range
A density detector for detecting the presence of image data;From each gradation range of image data to the feature of density distribution of target image
Set multiple modes with more defined and different image characteristics
With , Of the image data detected by the density detection unit. tone
Determine the density distribution of the target image from the area,Based on the judgment
Image dataJudgment unit that classifies each mode
When,According to the mode determined by the determination unit, the image data
After cutting the data in the specified gradation range, the remaining image data
A data correction unit for expanding data to the original data range,  The mode determined by the above determination unitBased on the above selected
Correction data of data correction section , Or 2 for normal data
Perform value processingA binary processing unit for converting analog image data into binary image data
Treatment method. 2. The apparatus according to claim 1, wherein the density detecting section, the determining section, the data correcting section,
2. The binarization processing method according to claim 1, wherein the processing in the binarization processing unit is performed simultaneously with reading of the image. 3. The binarization processing method according to claim 1, wherein said density detection section counts the presence of image data, and determines the presence of said image data when the number of pixels exceeds a predetermined value. 4. The binarization processing method according to claim 1, wherein the density detection section detects the density only for a pixel at a central portion excluding a leading portion and a trailing portion of the main scanning. 5. The binarization processing method according to claim 1, wherein said density detecting section detects the presence of image data in three or more gradation ranges. 6. The density detecting section includes a bright gradation area, a dark gradation area, and a bright gradation area including the bright gradation area and a dark gradation area.
2. The binarization processing method according to claim 1, wherein the presence of the image data is detected in three gradation regions indicating a gradation region which is brighter than the intermediate gradation region and a slightly brighter gradation region. 7. The data correction unit according to claim 1, wherein the data correction unit performs a binarization process.
Data has a function to correct with twice the gradation.
2. The method according to claim 1, wherein the correction data is calculated from the image data.
Value processing method. 8. The data correction unit according to claim 1, wherein the data correction unit performs a binarization process.
Data with a 4 times higher gradation.
2 according to claim 1, wherein calculating the correction data from the image data
Value processing method. 9. An image of a light density and a dark density in the density detecting section.
The presence of data is counted by separate counters,
The counter is provided for each line of the main scan and for the dark density counter.
Printer clears each scan including sub-scan.
The binarization processing method according to claim 1 .