JPH06292015A - Thresholding device and reader and facsimile equipment provided with the device - Google Patents

Abstract

(57) [Abstract] [Purpose] To improve the image quality by maintaining the continuity of the contour portion of the image at the time of binarizing the image with almost no adverse effects. The image area determination unit 12 includes multi-valued image data 15a that is a pixel of interest and multi-valued image data 15 that is a pixel on the previous line.
From b and the multi-valued image data 3a which is the pixel of the rear line,
The contour portion of the image in the pixel of interest is determined. According to the determination result 12a, the state probability determination unit 13 determines that the binarized data 6a of the previous pixel, the binarized data 17a of the pixel of the previous line, and the multi-valued image data of the rear pixel when the pixel of interest is located in the contour portion. Depending on each state of 5a, it is determined whether the continuity is in the main scanning direction or the sub scanning direction. The threshold selection unit 11 determines whether the first
~ A desired threshold value is selected from the third threshold values.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention uses multi-valued image data, which is obtained by scanning an image and sampling it on a pixel-by-pixel basis, and which expresses the gray scale of the image. Regarding a binarization device for converting into digitized data, for example,
Facsimile device, image scanner device or image generating device (X-ray CT [Computerized Tomography] device, MRI [Magnetic Radio iso
tope] device, ultrasonic diagnostic device, etc.) and the like.

[0002]

2. Description of the Related Art In a conventional facsimile machine, for example,
The image (transmission document) is scanned by a line sensor (CCD sensor, contact sensor, etc.), sampled in pixel units,
The gray scale of the image is represented by multi-valued image data. After that, the obtained multi-valued image data is subjected to a binarization process to be converted into white or black binarized data according to the density level, and an image is displayed. Furthermore, MH, which is a standard encoding method for facsimile machines, is applied to binarized data.
(Modified Huffman) coding or MR (Modified Relative Element Address)
Designate: Modified read) Data is compressed by encoding and transmitted.

However, even if the original image has the same density, the "dirt" of the transmission original, the image properties (line thickness and inclination, density, correlation with surrounding pixels, etc.) or MTF (Modureation) of the line sensor. The obtained multi-valued image data has various values due to the characteristics of the transfer function. Therefore, if the multi-valued image data is directly subjected to the binarization process, the reproducibility of the obtained image will be significantly deteriorated.

Under these circumstances, generally, the image reproducibility is improved by performing a process of correcting the signal level of the multi-valued image data before performing the binarization process.

FIG. 13 is a block diagram showing the structure of a conventional binarizing device. In FIG. 13, 1 is an input terminal for inputting the density level of an image read from a line sensor (not shown), 2 is a sampling of the density level of the image in pixel units, and multivalued image data (multivalued digital data) is obtained. An A / D conversion unit for conversion, 3 a shading correction unit, 4 a γ correction unit, 5 an MTF correction unit, 6 a binarization unit that performs binarization by comparing the magnitude of multi-valued image data with a threshold value, 7
Is an output terminal for outputting binarized data, 8 and 10 are first and third threshold value adding portions for giving first and third threshold values, respectively, and 11 is a first and third threshold value. A threshold value selecting unit that selects a predetermined threshold value from the threshold values output from the adding unit.

That is, for example, as shown in FIG. 13, in the shading correction section 3, correction processing is performed in accordance with the output characteristic (sensor distortion) of the line sensor so that the output from the line sensor (not shown) becomes uniform. And a shading correction process that performs a background density correction process of the read document, a γ correction process that performs a contrast adjustment of grayscale in a γ correction unit 4, and an MTF correction unit 5 that matches an image feature. In the character / line image area, edge enhancement processing is performed on a pixel to be binarized (hereinafter referred to as a pixel of interest), and in the halftone dot / photo image area, MTF correction processing for smoothing is performed and then binarization. In the part 6, in the character / line drawing area, a threshold value t having a predetermined value
Binarization processing for determining white or black is performed by comparing h and the density level of the target pixel to improve the reproducibility of the image.

FIG. 14 is an explanatory diagram for explaining the binarization processing in the binarization apparatus of FIG. FIG. 14A shows a read document and also shows a reading position by the line sensor as a line of interest. FIG. 14B shows the relationship between the density level and the threshold value in the binarization process on the line of interest. FIG. 14C shows a binary output image obtained by the binarization processing.

As shown in FIG. 14, even if the above-mentioned γ correction section 4 adjusts the contrast of the light and shade gradation, the density level is likely to be unstable in the part where the density level of the image largely changes. If it is unstable near the threshold value, the resulting binarized data will flutter (that is, the binary image will be discontinuous), causing the deterioration of the image quality (occurrence of isolated black dots / white dots, In the following, it is referred to as a notch), and further, there is a problem in that the amount of transmission data increases due to the deterioration of the coding efficiency when performing data compression, resulting in an increase in transmission time.

Therefore, in the prior art, the threshold value is not fixed at the time of binarization, but, for example, Japanese Patent Laid-Open No.
As described in JP-A-3-233673,
By giving a hysteresis characteristic having a predetermined fluctuation width that makes it easy to hold the value of the binarized data processed in the previous pixel, and performing binarization by comparing the magnitude with the threshold value (so-called By performing the hysteresis binarization process), the notch is prevented from occurring and the image quality is improved.

FIG. 15 is an explanatory diagram for explaining the binarization processing in such a conventional binarization apparatus. Figure 15
(A) shows a read original and also shows the reading position by the line sensor as a line of interest. FIG. 15B shows the relationship between the density level and the threshold value in the binarization process on the line of interest. FIG. 15C shows a binary output image obtained by the binarization processing.

That is, for example, as shown in FIG. 15B, if the binarized data of the previous pixel is white, by selecting thl as the threshold value for the pixel of interest, the binarized data of the pixel of interest is selected. If the binarized data of the previous pixel is black, conversely, if thh is selected as the threshold value for the pixel of interest, the binarized data of the pixel of interest is likely to be black. By performing such hysteresis binarization processing, even if the density level is unstable in the vicinity of the threshold value, the generated binarized data may fluctuate (that is, discontinuity of the binary image) to some extent. This can be prevented, and further image quality can be improved.

[0012]

However, when converting the multi-valued image data representing the gray scale of the image into the binary data of white or black corresponding to the density level, the binary value of the previous pixel is converted. In the hysteresis binarization processing in which the threshold value is determined by the digitized data, there are the following problems.

That is, when the hysteresis binarization process is similarly performed for all pixels, the density level near the threshold value becomes unstable in a portion where the density level of the image largely changes in pixel units. Although it is possible to prevent the fluttering of the obtained binary data (that is, discontinuity of the binary image) to some extent, when the density level of the read document is small (for example, a thin line or a high spatial frequency due to the MTF characteristic). When the output level of the line sensor is low due to the range component, or in the area where the change in density level is small such as in the photographic image area), the change in density level is absorbed by the hysteresis characteristics, and the thin line breaks, "Collapse",
There is an adverse effect that "blurring", "tail tailing", "corner chipping", etc. occur, so that the hysteresis binarization process is hardly used in practice.

The object of the present invention has been made in view of the above-mentioned problems of the prior art. The object of the present invention is to make the density level unstable when binarizing multi-valued image data. Prevents discontinuity of the obtained binary image as much as possible, and at the same time cuts fine lines, “crushes”, “blurrs”, “tails”,
Almost no adverse effects such as "corner chipping" occur 2
Provide a quantizer to improve the quality of reproduced images, and further perform encoding by MH encoding, which is an encoding method that determines the code length based on the statistical properties of information sources such as run length. In this case, it is to prevent the amount of transmission data from unnecessarily increasing due to the reduction of the compression rate.

[0015]

In order to achieve the above object, according to the present invention, multi-valued image data converted into digital data is used to determine image area characteristics, particularly the contour of an image in which the density level sharply changes. Image area determination means for determining a part,
State probability determining means for determining the state probability that the value of the binarized data for the target pixel will be white or black from the state of the surrounding pixels, and the binarized data will be black based on the result of the state probability determining means. First threshold value thh for facilitating, 2
Threshold value selecting means for selecting one of a second threshold value th and a third threshold value thl for facilitating whitening of the binarized data when the state probability of the binarized data is indefinite; And a binarizing means for comparing the threshold value selected by the threshold value selecting means with image data represented by multi-valued image data and determining white / black depending on the magnitude. It is characterized by

[0016]

In the present invention, the multi-valued image data converted into digital data is an image in which the density level, which is a portion where the notch is likely to occur and is prominent in the character / line drawing area, is sharply changed by the image area determination means. The contour portion of the is determined and extracted. The state probability determining means determines a state probability that the value of the binarized data for the target pixel will be white or black from the state of the surrounding pixels. According to the above result, if the region where the pixel of interest is located is not the contour portion of the image, the threshold value selecting means and the binarizing means select the second threshold value th to perform binarization. In addition, when it is determined that the pixel of interest is located in the contour portion of the image, the first threshold value for each of the state probability black and white is determined according to the state probability after binarization determined from the states of the surrounding pixels. Value t
By selecting hh and the third threshold thl,
The binarized value makes it easier to maintain continuity from surrounding pixels. This prevents the occurrence of only specific notches that are likely to occur in the contour part of the image as much as possible, and
It is possible to obtain a high-quality binary image with almost no adverse effects such as line breakage, “crush”, “blurring”, or “tail trailing”.

[0017]

Embodiments of the present invention will now be described in detail with reference to the drawings. FIG. 1 is a block diagram showing the configuration of a binarizing device as a first embodiment of the present invention.

In this embodiment, a line sensor (CCD sensor, contact sensor, etc.) is used for one-dimensional scanning (main scanning direction), and the sensor position is sequentially moved (sub-scanning direction) to change the density level of an image. Line-sequential reading,
The read density level is sampled in pixel units, and the density gradation of the image is represented by multi-valued image data (the high density gradation level is relatively assigned to white and the low density gradation level is assigned to black). It is a binarization device that performs binarization processing on multi-valued image data and converts it into binarized data of white or black according to a density level.

In FIG. 1, 1 is an input terminal for inputting the density level of an image read from a line sensor (not shown), 2 is a sampling of the density level of the image in pixel units, and multivalued image data (multivalued digital A / D conversion unit for converting to (data), 3 is a shading correction unit, 4 is a γ correction unit, 5 is an MTF correction unit, and 6 is a binary value that is binarized by comparing the multivalued image data with a threshold value. The conversion part, 7 is 2
Output terminals for outputting digitized data, 8 to 10 are first to third threshold value adding sections for giving first to third threshold values, and 11 is a first to third threshold value adding section. A threshold value selecting unit for selecting a predetermined threshold value out of the threshold values output from the image area determining unit 12 for determining image characteristics of a pixel (pixel of interest) to be binarized; Is a state probability determination unit that determines the state probability of the binarized data of the pixel of interest. Further, 15 is a first storage medium unit that stores at least two lines of multivalued image data after shading correction, 14 is a delay unit that delays MTF-corrected multivalued image data by the processing of one pixel, and 16 is a previous unit. A delay unit that delays the binarized data of the pixel (the pixel immediately before the pixel of interest) by the processing of one pixel, and 17 is a second storage medium unit that stores the binarized data for at least one line.

Next, the operation of this embodiment will be described with reference to FIG. In the A / D converter 2, the density level of the image read from the line sensor is sampled in pixel units, converted into multi-valued image data 2a, and output.

In the shading correction section 3, the variation range of the density level is adjusted by the correction processing such that the output from the line sensor becomes uniform according to the sensor characteristic or the sensor distortion (device characteristic) or the background density correction processing of the read document. The optimization process for optimizing and preventing deviation of the density level is performed and output.

In the first storage medium unit 15, the multi-valued image data 3a processed by the shading correction unit 3 is stored and held for two lines on a pixel-by-pixel basis, and the pixels of the previous line as seen from the read pixel position of the line sensor are stored. The multi-valued image data 15a and the multi-valued image data 15b of the pixels in the line before the previous line are output.

The γ correction section 4 adjusts the contrast of the multi-valued image data 15a of the pixels on the previous line read from the first storage medium section 15 and outputs the result.

In the MTF correction unit 5, the multi-valued image data 4a processed by the γ correction unit 4 is processed even if the density level change is small at the contour portion of the image due to the MTF characteristic of the line sensor. Edge emphasis processing for emphasizing the density level change amount of the contour portion is performed and output.

The delay unit 14 delays the multi-valued image data 5a processed by the MTF correction unit 5 by one pixel.

The binarization unit 6 compares the output 14a, which is multi-valued image data of the pixel of interest, from the delay unit 14 with a threshold value 11a selected by a threshold value selection unit 11 described later. . As a result, when the multi-valued image data 14a representing the density level is larger than the threshold value 11a, it is judged to be white, and in other cases, it is judged to be black, and the binary data 6a is represented by representing it as 1-bit digital data. Generate and output.

The second storage medium section 17 stores and holds one line of the binarized data 6a for each pixel, and stores the binarized data 17a of the pixel on the preceding line as seen from the pixel position of the binarized data 6a. Output.

In the image area determination section 12, the multi-valued image data 15a which is the density level of the target pixel, the multi-valued image data 15b which is the density level of the pixel of the front line among the peripheral pixels of the target pixel, and the rear line. From the multi-valued image data 3a which is the density level of the pixel of
In particular, it is determined whether or not the pixel of interest is located in the contour portion (edge portion) of the image, and the determination result 12a is output.

In the state probability judging section 13, the image area judging section 12
If the pixel of interest is located in the contour portion in accordance with the determination result 12a from the
The binarized data 16a, the binarized data 17 of the pixels of the previous line
a and the respective states of the multi-valued image data 5a of the post-pixels which have not been binarized yet, are referred to so that the continuity between the peripheral pixels and the target pixel is in either the main scanning direction or the sub-scanning direction. It is determined whether or not it has. Thereby, after the binarization process, the binarized data of the pixel of interest that should hold the continuity in the determined direction is predicted. According to this prediction result, the state probability of the binarized data of the pixel of interest is set to white or black, or undefined if the binarized data of the pixel of interest cannot be predicted. If the pixel of interest is not located in the contour portion, the binarized data of the pixel of interest is unpredictable, and the state probability of the binarized data of the pixel of interest is determined to be indefinite. As described above, the state probabilities of the binarized data for the target pixel are set in three ways, and the determination result 13a is determined.
Is output.

The first to third threshold value adding sections 8 to 10 respectively have a first threshold value thh and a second threshold value t.
It holds h and the third threshold value thl. here,
The second threshold value th is statistically determined from, for example, the density distribution of each pixel, and when the threshold value is used for the binarization processing of multi-valued image data, the obtained binary image is almost It is assumed that the multi-valued image is displayed faithfully. in this case,
It is assumed that the first threshold value thh is a value obtained by increasing the threshold value by a predetermined value Δdh with respect to the second threshold value th.
Black is emphasized during binarization. Further, the third threshold value thl is a value obtained by lowering the threshold value by a predetermined value Δdl with respect to the second threshold value th, and white is emphasized during binarization.

In the threshold value selecting unit 11, the first to third threshold value adding units 8 to 10 are selected according to the result 13a of the state probability of the binarized data of the pixel of interest by the state probability determining unit 13.
A desired threshold value is selected from the first to third threshold values. In this case, considering the respective characteristics of the first to third threshold values, if the state probability according to the determination result 13a is black, the first threshold value thh is set, and if the state probability is white, the third threshold value thh is set. If the state probability is indefinite, the threshold value thl is selected and the second threshold value th is output.

Next, the image area determination operation of the image area determination unit 12 and the state probability determination operation of the state probability determination unit 13 will be described in more detail with reference to FIGS. 2 to 8 and Table 1.

FIG. 2 is an explanatory diagram showing an example of a target pixel P and peripheral pixels which are target areas when performing binarization.

It is shown that three pixels are referred to in the main scanning direction and three pixels in the sub scanning direction centering on the pixel of interest P, and the image information is referred to as peripheral pixels. Here, it is assumed that the main scanning direction and the sub-scanning direction are as shown, binarization is sequentially performed in the main scanning direction, and the line sensor moves in the sub-scanning direction from top to bottom.

The pixel on the line of interest shown in the figure is the output 15a of the first storage medium unit 15, and the pixel on the previous line is the first line.
In the output 15b of the storage medium unit 15, the pixels in the subsequent line correspond to the multi-valued image data 3a.

First, the operation of the image area determination section 12 will be described with reference to FIG. As shown in FIG. 3, in the peripheral pixels of the target pixel P, two pixels Pl and Pru sandwiching the pixel Pu located above the target pixel P in the previous line and a pixel located below the target pixel P in the subsequent line. 2 pixels P across Pd
Focusing on the density levels of the four pixels of ld and Prd, the difference value between the density level of the pixel of interest P and the density level of the pixel of interest P is calculated according to the following equations 1 to 4.

Two pixels Plu and Pr located in the main scanning direction
The difference value Δ1 (absolute value) between the average density level of u and the density level of the target pixel P is calculated.

[0038]

[Expression 1] Δ1 = | (Plu + Pru) / 2−P | Difference value Δ2 of the density level between the target pixel P and the average value of the density levels of the two pixels Pld and Prd located in the main scanning direction.
Calculate (absolute value).

[0039]

Δ2 = | (Pld + Prd) / 2−P | The difference value Δ3 between the density levels of the target pixel P and the average density level of the two pixels Plu and Pld located in the sub-scanning direction.
Calculate (absolute value).

[0040]

Δ3 = | (Plu + Pld) / 2−P | The difference value Δ4 between the density levels of the target pixel P and the average value of the density levels of the two pixels Pru and Prd located in the sub-scanning direction.
Calculate (absolute value).

[0041]

## EQU4 ## Δ4 = | (Pru + Prd) / 2−P | It is possible to specify that the portion where the change in the density level for each pixel is relatively sharp is the contour portion of the image. It is possible to detect parts. Therefore, the optimum value dth for separating the contour part and the other parts is statistically determined from the distribution of difference values between pixels obtained for all pixels, and is set as the fourth threshold value dth. It is assumed that the difference between pixels and the fourth threshold value dth are compared to determine whether or not it is the contour portion of the image. As a result, if there is a difference value larger than the fourth threshold value dth among the difference values Δ1 to Δ4, it is determined that the pixel of interest P is located at the contour portion of the image. Is.

FIG. 4 shows a specific example of the image area determination section 12 that performs the above processing. In FIG. 4, 18 to 22 are delay sections for delaying one pixel processing, 23 to 26 are averaging sections for obtaining an average density level of multi-valued image data (density levels) of 2 pixels, and 27 to 30 are of the target pixel P. A difference value calculation unit that obtains a difference value between the density level and the average density level of the two pixels, 3
Reference numeral 6 denotes an addition unit for adding the above-mentioned fourth threshold value dth, which serves as a reference for judgment when it is judged whether or not it is the contour portion of the image, and 31 to 34 are difference values with the fourth threshold value dth. A comparing unit for performing a magnitude comparison with the difference value obtained by the calculating units 27 to 30, and a determining unit 35 for determining whether or not the pixel of interest P is located at the contour portion of the image according to the results of the comparing units 31 to 34. Is.

The operation of the image area determination unit 12 shown in FIG. 4 will be described. Here, the output 15a of the first storage medium unit 15 corresponding to the multi-valued image data of the pixel of the line of interest, the first storage medium unit 15 corresponding to the multi-valued image data of the pixel of the previous line
15b (Plu), and the multi-valued image data 3a (Pld) corresponding to the multi-valued image data of the pixels on the subsequent line.

The two-stage delay units 18 and 19 delay the multi-valued image data 15b by two pixel processing, and output the multi-valued image data 19a corresponding to the pixel Pru.

The multi-valued image data 3b is delayed by the two-pixel processing by the two-stage delay units 20 and 21, and the multi-valued image data 21a corresponding to the pixel Prd is output.

The delay unit 22 delays the multivalued image data 15a by one pixel processing, and outputs the multivalued image data 22a corresponding to the pixel of interest P.

By the averaging unit 23, in the multi-valued image data,
2 pixels Pl (15
The average value 23a of b) and Pru (19a) is calculated and output.

Similarly, by the averaging unit 26, in the multi-valued image data, two pixels Pld located in the main scanning direction of the subsequent line are displayed.
The average value 26a of (3a) and Prd (21a) is calculated and output. Further, the averaging unit 24 causes the multi-valued image data to have two pixels Plu (15b) and P which are located in the sub-scanning direction.
The average value 24a of ld (3a) is calculated and output.

By the averaging unit 25, in the multi-valued image data,
Two pixels Pru (19a) and Prd located in the sub-scanning direction
The average value 25a of (21a) is calculated and output.

In the difference value calculating units 27-30, the average density levels 23a-26 obtained by the averaging units 23-26 are calculated.
The difference values (absolute values) between a and the density level 22a of the target pixel P are calculated as Δ1, Δ3, Δ4, and Δ2, respectively.

The comparators 31 to 34 compare the fourth threshold value dth with the difference values Δ1 to 4 obtained by the difference value calculators 27 to 30 to determine whether they are large or not. The result is output.

According to the comparison results 31a to 34a of the comparing units 31 to 34, the judging unit 35 determines that the pixel of interest P is determined when the difference values Δ1 to 4 are larger than the fourth threshold value dth. Determined that it is located in the outline of the image,
The determination result 12a is output.

Next, the operation of the state probability determining section 13 will be described in detail with reference to FIGS. 5 and 6 and Table 1.

According to the result 12a of the image area determination unit 12, when it is determined that the pixel of interest is located in the contour portion, the following processing is performed.

Table 1 shows the processing contents of the state probability judging section 13.

[0056]

[Table 1]

As the image information of the peripheral pixels, as shown in FIG. 5, for the pixel of interest P, the binarized data Pl (16a) of the preceding pixel and the multivalued image data Pr of the succeeding pixel in the main scanning direction.
(5a), and further, the binarized data P of 3 pixels in the previous line
The states of lu, Pu, and Pru (17a) are referred to.

First, the binarized data Pl of the preceding pixel,
From the image information of 2 pixels of the multi-valued image data Pr of the rear pixel,
A main scanning direction state determination is performed as to whether or not the pixel of interest P is located in a contour portion (a contour portion generated in a direction perpendicular to the main scanning direction) in the main scanning direction.

For example, as shown in Table 1, when the binarized data Pl of the front pixel is white and the multi-valued image data Pr of the rear pixel is equal to or smaller than the third threshold value thl, Indicates that the binarized data of the rear pixel is surely black, and therefore the pixel of interest P is determined to be located in the contour portion in the main scanning direction (the contour portion occurring in the direction perpendicular to the main scanning direction) ( 6 (c) and 6 (d)). On the contrary, the third threshold value thl
If it is larger, it is determined that the pixel of interest P is located in the non-contour portion in the main scanning direction.

Similarly, when the binarized data Pl of the front pixel is black and the multi-valued image data Pr of the back pixel is larger than the first threshold value thh, the binarization of the back pixel is performed. Since the data is surely white, it is determined that the pixel of interest P is located in the contour portion in the main scanning direction (the contour portion occurring in the direction perpendicular to the main scanning direction) (FIG. 6A,
(B)). Otherwise, the pixel of interest P is determined to be located in the non-contour portion in the main scanning direction.

According to the above determination result, when it is determined that the target pixel P is located in the non-contour portion in the main scanning direction, the target pixel P is generated in the sub scanning direction in the contour portion (direction perpendicular to the sub scanning direction). Since it can be determined that the binarized data is located in the contour portion, the binarized data of the target pixel is predicted so that the binarized data can easily maintain continuity in the main scanning direction. That is, 2 of the target pixel
The state probability of the binarized data is determined by assigning the binarized data of the pixel Pl.

On the contrary, when it is determined that the pixel of interest P is located in the contour portion in the main scanning direction (the contour portion generated in the direction perpendicular to the main scanning direction), the binary data of the pixel of interest P is scanned in the main scanning direction. It is not always possible to specify from the binary data of the direction. Therefore, the binarized data of the target pixel P is two-dimensionally predicted from the image information in the sub-scanning direction.

In this case, as shown in FIG. 5B, the binarized data P of three consecutive pixels in the preceding line which has been binarized.
As the states of lu, Pu, and Pru (previous line run state), a state in which white and black are mixed (for example, FIG. 6A to FIG.
(D)) and the state where all three pixels have the same value (white run, black run) (for example, FIGS. 6E to 6L) are roughly classified.

As a result, when the preceding line run state is mixed (discontinuous), the pixel of interest P is located at the contour portion in the main scanning direction (the contour portion occurring in the direction perpendicular to the main scanning direction). Then, it is determined, and the binarized data of the target pixel is predicted so that the binarized data can easily maintain continuity in the sub-scanning direction. That is, as the state probability of the pixel of interest, the pixel P
It is determined by assigning u binary data.

If all three pixels have the same value (continuous), it can be almost determined that the pixel of interest P is located at a corner (corner). In such a corner, the state probability of the pixel of interest P is determined to be indefinite without considering the continuity from the preceding pixel Pl or the preceding line pixel Pu.

Further, the determination result 12a of the image area determination unit 12
Then, when it is determined that the pixel of interest is located in the non-contour portion, the state probability of the pixel of interest P is determined to be indefinite.

By the processing of the state probability determining unit 13 described above, the state probability of the pixel of interest P is determined to be white, black or indeterminate, and the determination result 13a is output.

FIG. 7 shows a state probability determining unit 1 for performing the above processing.
3 shows a specific example. In FIG. 7, reference numeral 37 designates a selection section for selecting from the first threshold value thh and the second threshold value thl in accordance with the binarized data 16a of the preceding pixel, and 38 designates the multi-valued image data 5a of the succeeding pixel. And a threshold value 39, a reference numeral 39 denotes a contour portion in the main scanning direction (a contour portion generated in a direction perpendicular to the main scanning direction) in accordance with the comparison result and the binarized data 16a of the previous pixel. ), The main-scanning-direction contour portion determining unit, 4
0 and 41 are delay units that delay the binarized data of the pixels of the previous line by one pixel processing, 42 is a previous line run detection unit that detects black run / white run of the three pixels of the previous line, and 43 is the above state probability. It is a state probability determination unit that performs determination processing.

Next, the operation of the state probability judging section 13 of FIG. 7 will be described. In the selection unit 37, the binarized data 1 of the pixel Pl
In accordance with 6a, the first threshold value thh for facilitating the binarized data to become black if black is selected, and conversely, the third threshold value thl for facilitating the binary data to become white if white is selected. Output.

The comparison section 38 compares the multi-valued image data 5a of the pixel Pr after the MTF correction processing with the threshold value output 37a from the selection section 37 to obtain a comparison result 38a.

In the main-scanning-direction contour determination unit 39, according to the comparison result 38a, as the main-scanning-direction contour determination, when the binarized data of the pixel Pl is black and the multivalued image of the pixel Pr is obtained. If the data 5a is larger than the threshold value output 37a, it is determined that the pixel of interest P is located in the contour portion in the main scanning direction (the contour portion occurring in the direction perpendicular to the main scanning direction).
On the other hand, when it is small, it is determined that it is located in the non-contour portion in the main scanning direction. When the binarized data of the pixel Pl is white and the image data 5a of the pixel Pr is smaller than the threshold output 37a, the pixel of interest P is in the main scanning direction in the contour portion (main scanning direction). Contours that occur in the direction perpendicular to the direction)
On the contrary, when it is large, it is determined to be located in the non-contour portion in the main scanning direction.

In the previous line run detection unit 42, as the previous line run determination in the sub-scanning direction, second binary data corresponding to three consecutive pixels Plu, Pu, Pru located in the previous line with respect to the pixel of interest P is detected. Output 17 of the storage medium unit 17 of
a, referring to the outputs 40a and 41a of the delay units 40 and 41,
If all three pixels have the same value, it is determined to be continuous, and if not, it is determined to be discontinuous (mixed), and the result 42a is output.

In accordance with the above result, the state probability determining unit 43 determines the state probability (white, black or indefinite) of the binarized data of the target pixel P from the determination result 12a of the image area determining unit 12 and the above state probability determining process. Determined based on the method.

FIG. 8 is an explanatory diagram for explaining the binarization processing in this embodiment. FIG. 8A shows the original document to be read, and shows the reading position by the line sensor as the line of interest. FIG. 8B shows the relationship between the density level and the threshold value in the binarization process on the line of interest. Figure 8
(C) shows a binary output image obtained by the binarization process.

As shown in FIG. 8, according to the present embodiment, the density level of the image data obtained by reading the document by the sensor is "dirt" of the document, the nature of the image (line thickness and inclination,
Even if it has various values depending on the density, the correlation with surrounding pixels, etc., or the MTF characteristics of the sensor, by adaptively correcting the signal level before performing the binarization processing, the image reproducibility is improved. It is possible to improve.

In the image area determination, in a multi-valued image such as a photograph, the change in the density level in pixel units is gentle,
The density level of the pixel of interest is almost equal to the density level of the surrounding pixels and is small, but at the boundary of the image in the binary image, the property that the density level changes sharply is used. Is. In other words, it is obtained from the multi-valued image data read from the line sensor when the outline portion of the image in the character / line drawing area where the density level sharply changes (that is, when only the simple binarization processing based on the fixed threshold value is performed. In a binary image, it is a portion where a notch is likely to occur and is conspicuous.), And it is possible to easily extract the high frequency component of the spatial frequency and the halftone dot where the change of the density level of the image is relatively gentle. It is possible to effectively determine a photographic image area or the like as a non-contour portion.

Further, when the area determination is performed, since the determination is made based on the difference value between the average of the density levels of the two pixels located in the predetermined direction and the target pixel, the detection capability of the diagonal line contour portion is suppressed while It is possible to effectively extract only the contour portion generated in the direction perpendicular to the scanning direction or the sub-scanning direction. Further, it is possible to prevent erroneous detection of the contour portion of the image due to noise included in the density level of each pixel, and to prevent erroneous detection of the contour portion in the halftone image as much as possible.

As described above, in the case where the pixel of interest P is located in the contour portion, the state of the binarized data or multivalued image data of the pixel which has already been binarized is referred to as the image information of the peripheral pixels. As a result, it is possible to easily predict the binary data that should have the continuity with the peripheral pixels. According to this prediction result, white or black as the state probability of the binarized data for the target pixel P,
Alternatively, if unpredictable, it is possible to easily determine which of the above three states is indefinite.

According to the state probability of the pixel of interest P obtained as described above, a threshold value for binarization is adaptively selected from a plurality of threshold values. If the state probability is black, black is emphasized. It is possible to perform binarization by selecting a threshold value to be used, a threshold value that emphasizes white if the state probability is white, and an ordinary threshold value if the state probability is indefinite.

As a result, even if the contour portion of the image in the character / line drawing area in which the density level changes steeply and unstable in which the image quality deterioration cannot be avoided by the notch generation in the simple binarization processing, Smoothing can be performed while maintaining the continuity of the digitized data as much as possible, and a binary image with few notches can be obtained.

On the other hand, in the area where the spatial frequency is high even in the character / line drawing area, or in the area where the density level of each pixel is small, such as the halftone dot / photo image area, it is fixed. Since the simple binarization process is performed by using the threshold value, it is possible to prevent the resulting binary image from having adverse effects such as thin line cutting, “crushing”, “blurring”, or “tail trailing”.

Further, even in the contour portion of the image, since it becomes possible to extract the corner portion of the image from the state of the peripheral pixels, as shown in FIG. It is possible to prevent adverse effects such as “tail buzz” from occurring. This makes it possible to easily obtain a high-quality binary image.

Here, in the present embodiment, as the method for detecting the contour portion of the image, only the algorithm for setting the target area to the peripheral pixel 3 × 3 area and extracting it two-dimensionally is shown. It is not limited to this. Also, as the contour detection method, one method has been shown for the detection method by the difference value calculation, but needless to say, the present invention is not limited to this and can be performed by another detection method by the difference value calculation or image pattern recognition. Yes.

In the present embodiment, only the case where the target area is set to be 2 × 3 area and defined by 2 lines when the state probability is determined is shown, but of course, the target area is expanded and the feature of the image is determined. As a result, it is possible to further improve the probability confirmation accuracy.

Further, in the present embodiment, only the case where the state probability is set to three states and the processing is performed is shown. However, by further subdividing the state probability and performing the processing of changing the threshold value more adaptively, It is possible to achieve higher image quality.

Also, only the case where three consecutive pixels of the previous line are focused on is shown when making a decision on the corners of an image, but of course this is not the only case, and decision is made by adaptively changing the reference pixel. It is needless to say that the accuracy can be increased and the pattern recognition of the image can be performed.

Further, as the smoothing method for maintaining the continuity with respect to the contour area of the image, only the method of setting the processing direction and performing the hysteresis binarization processing is shown.
Needless to say, other smoothing methods, for example, adaptive correction processing of multi-valued image data before binarization, data conversion processing after binarization, or the like can be performed.

In the present embodiment, the reading density level correction processing method is limited, but the method is not limited to this, for example, another method such as AGC (Auto Gain Controller).
It goes without saying that the same effect can be obtained even if the correction processing of the density level is performed by the above.

Further, in the present embodiment, only the case where the processing system having the hardware configuration is used is shown, but of course, even if the processing system according to the present invention is realized by the software processing, the same effect can be obtained. Needless to say.

FIG. 9 is a block diagram showing the configuration of a binarizing device as a second embodiment of the present invention. The present embodiment is an application of the binarizing apparatus of the present invention to an image transmitting apparatus (for example, a facsimile apparatus) that reads and encodes a multi-valued image such as a character / line drawing or a photograph and stores or transmits it.

In FIG. 9, reference numeral 44 denotes a line sensor, which is a reading sensor for one-dimensionally scanning and reading a transmission original,
12 'indicates whether the pixel of interest is a pixel in a binary image area such as a character or line drawing, a pixel located in the outline of a character or line drawing, or a multi-valued image area such as a photograph, based on the density level of surrounding pixels. Image area discrimination, and at the same time, in the case of a multi-valued image area, the degree of multi-valued image area property (degree indicating the closeness of a binary image area or a multi-valued image area) is determined. 45, 46 are first and second γ correction units for converting the density level to a predetermined level according to the level, and 47 is a density level of the pixel of interest based on the density levels of the pixel of interest and peripheral pixels. Is a smoothing unit for smoothing the image, 48 is an edge emphasizing unit for emphasizing the edge portion of the line drawing based on the density levels of the target pixel and peripheral pixels, and 49 is a density level for the smoothing unit 47 and the edge emphasizing unit 48. According to the determination result 12b of the image area determination unit 12, mixing at a predetermined ratio Mixing unit that, 50 is a delay unit.

Reference numeral 51 denotes a multi-valued image area, in which a multi-valued image having M gradations is re-quantized into N gradations (where M and N are integers, N ≦ M), and a quantization error at that time is re-quantized to peripheral pixels A halftone processing unit for performing N-value error diffusion processing for diffusing the image data into an average density level before and after quantization, 6 is the same as that shown in FIG.
A binarization unit that binarizes white (maximum density level) and black (minimum density level) by binarization processing, and 52 is either the output of the halftone processing unit 51 or the binarization unit 6 as binarized data. A selection unit that outputs the image according to the determination result of the image area determination unit 12;
Reference numeral 3 is a size conversion unit for reducing / enlarging the read image, 54 is an encoding unit for compressing the binary data by MH encoding or MR encoding, and 55 is the encoded data, which is sent to the line 56. It is a modem. Other than that, the same components as those in FIG. 1 are denoted by the same reference numerals.

The operation of this embodiment will be described in detail with reference to FIG. The reading sensor 44 reads the transmission document line by line, samples the read density level (analog value) pixel by pixel, and outputs the sampled density level. In the A / D converter 2,
The sampled read density level is quantized, represented by multivalued image data (multivalued digital data), and supplied to the shading correction unit 3.

The shading correction unit 3 corrects the sensitivity variations of the pixels of the reading sensor 44 and the unevenness of the light source for illuminating the original. Therefore, the shading correction unit 3 stores in advance multi-valued image data (referred to as a shading waveform) for one line, which is obtained by reading the white reference document with the reading sensor 44 and quantized the shading waveform. In addition, the quantized value (density level) of the read pixel is corrected corresponding to the position on the line of the document (on the main scanning). This shading correction result is stored in the first storage medium unit 15,
The first and second γ correction units 45 and 46 and the image area determination unit 12
Is supplied to.

The first storage medium unit 15 stores the multivalued image data 3a after shading correction for at least two lines, and the multivalued image data 15a of the pixel on the previous line when viewed from the read pixel position of the read sensor 44 and The multi-valued image data 15b of the pixels on the line before the previous line is output.

In the first and second γ correction units 45 and 46,
FIG. 10 shows the density levels of the multi-valued image data 15a of the pixels of the previous line, the multi-valued image data 15b of the pixels of the immediately preceding line, and the multi-valued image data 3a after shading correction, which are read from the first storage medium unit 15. It is corrected according to the input / output density characteristics as shown in. The figure shows three types of correction characteristics. This is done by taking the density characteristics of the multi-valued image into consideration of the sharpness of the image after recording and the recording characteristics. The correction results of the first and second γ correction units 45 and 46 are supplied to the smoothing unit 47 and the edge enhancement unit 48, respectively.

The smoothing unit 47 smoothes the multivalued image data corrected by the first γ correction unit 45.
That is, as shown in FIG. 11A, the weighting factors a, b, c, d, e, f, g, h, for the density levels of the pixel of interest (hatched pixel) and its surrounding pixels. Multiply i and determine the density level of the pixel of interest as the sum thereof. The weighting coefficient is given a positive value with respect to the pixel of interest and the peripheral pixels, and the weighting coefficient is distributed so that the sum becomes 1. As a result, changes in the density level of each pixel are smoothed. The degree of smoothing depends on how the weighting factors are given.

The area of the peripheral pixels in the smoothing section 47 is not limited to that shown in FIG. 11A, and in the case of a multi-valued image consisting of halftone dots, the peripheral pixels covering the cycle of the halftone dots. The area must be used to smooth so as to detect the average density within the halftone dot period. Here, for convenience,
In order to show the principle, a 3 × 3 pixel area as shown in the figure is used.

In the edge emphasis unit 48, the second γ correction unit 4
Edge enhancement is performed on the multi-valued image data corrected in step 5. That is, as shown in FIG. 11B, the weighting factors a, b, c, d, e, f, g, h, for the density levels of the pixel of interest (hatched pixel) and its surrounding pixels. i
And the density level of the pixel of interest is determined as the sum. The weighting factor gives a positive value (e) to the pixel of interest and negative values (a, b, c, d, f,
g, h, i) are given and distributed so that the sum of the weighting factors becomes 1. As a result, when the difference in density level between the peripheral pixel and the target pixel is large, the difference is further emphasized, and as a result, the image boundary portion such as a line drawing is emphasized.

If the difference in density level between the target pixel and the average value of the density levels between the two pixels described in the first embodiment is larger than the predetermined value dth, the image area determination unit 12 'determines that the contour portion of the image is It is a binary image area portion, and if it is less than that, it is determined to belong to the multivalued image area portion.

The area determination is performed based on the difference in density level between the target pixel and the average density level between two pixels.
Change the density level change of black in the halftone dot image and white between halftone dots to 2
This is to avoid erroneous determination as a value image area.

In addition, when the image area determination unit 12 'determines that the image area belongs to the multi-valued image area portion, the degree of multi-valued image area property (binary image area, multi-valued image area) is determined from the degree of density level difference. 12b) is also determined at the same time.

Although the case where the multi-valued image data after the shading correction processing is used as the multi-valued image data for calculating the density level is shown here, the multi-valued image data after the smoothing processing or the edge emphasis processing is used. The image area discrimination accuracy may be improved by using the value image data (for example, by using the multivalued image data that has been smoothed in consideration of the halftone dot period).

The image area determination is not limited to the above method, and various determination methods can be used. For example, Japanese Patent Laid-Open No. 2-292956 and Japanese Patent Publication No.
-62355, paying attention to the halftone dot periodicity, first detecting the halftone dot area, and then, Japanese Patent Laid-Open No. 58-11.
As described in Japanese Patent No. 5975, the difference between the density level change amount and the change frequency of the line drawing and the photographic image (excluding the halftone dot image) is used to distinguish between the line drawing and the photographic image. It is also possible to discriminate between the binary image area and the multivalued image area such as a photograph (including a halftone image).

In the mixing unit 49, the multivalued image data 47a smoothed by the smoothing unit 47 and the multivalued image data 4 in which the image boundary portion such as a line drawing is emphasized by the edge emphasizing unit 48.
8a and weighting factors α and β (where α + β = determined by the degree 12b of multivalued image area characteristic from the image area determination unit 12).
1, α ≧ 0, β ≧ 0), respectively, and the density level of the pixel of interest is determined as the sum thereof.

As a result, the edge-enhanced value in the binary image area and the smoothed value in the halftone area are
Each is determined as the density level of the pixel of interest. Further, in the intermediate area between the binary image area and the multi-valued image area, the degree of multi-valued image area property obtained by the image area determination unit 12 is 12
If the weighting factor is close to the binary image area according to b, α ≦ β
In addition, if it is close to the multi-valued image area, α ≧ β is set to determine the density level of the pixel of interest. Then, the calculated multi-valued image data is given to the halftone processing unit 51.

The halftone processing section 51 requantizes the multivalued image data represented by M gradations obtained from the mixing section 49 into N gradations (where M and N are integers and N ≦ M). Then, N-value error diffusion processing is performed in which the quantization error at that time is diffused to peripheral pixels and the average density levels before and after quantization are matched.
In this case, the degree 12b of the multivalued image area characteristic obtained by the image area determination unit 12 determines the degree of diffusion of the quantization error to the peripheral pixels. The binarized data 51a calculated in this way is given to the selection unit 52.

The halftone processing for displaying the multi-valued image area by the binarized data may be represented by the dither method.

The delay section 50 delays the multivalued image data 48a corrected by the edge enhancing section 48.

In the binarizing section 6, the threshold value selecting section 11 selects the multivalued image data of the pixel of interest based on the output 50a from the delay section 50 and the processing shown in the first embodiment. The magnitude comparison with the threshold value 11a is performed. As a result,
The multilevel image data 50a representing the density level is the threshold value 11
If it is larger than a, it is judged to be white, and if not, it is judged to be black, and represented by 1-bit digital data to generate and output the binarized data 6a.

The selection unit 52 includes halftone processing units 51 and 2.
In the case where the pixel of interest is located in the multi-valued image region according to the degree of multi-valued image region property 12b obtained by the image region determination unit 12 in the binarized data for each pixel of interest obtained by the binarization unit 6, , And outputs the result of the halftone processing unit 51 and the result of the binarization unit 6 in the case of a binary image area.

In the size conversion section 53, the binarized data obtained by the selection section 52 is subjected to an interpolation process, a thinning process, a logical operation process, etc., to enlarge / reduce the size of the read document, and to perform a moving process ( Rotational movement, parallel movement, reverse movement, etc.), smoothing processing, etc.

In the encoding unit 54, the binary data processed by the size conversion unit 53 is data-compressed by MH encoding which is adopted as an international standard system for one-dimensional encoding of G3 facsimile. According to the run frequency of the binarized data, this coding method assigns short code words to run lengths with high occurrence frequency and long code words to run lengths with low occurrence frequency. By allocating, the data is compressed. The compressed encoded data is modulated by the modem unit 55,
It is output from the line 56.

As described above, according to the present embodiment, the density level of the image data obtained by reading the original with the sensor is such that the "dirt" of the original, the nature of the image (line thickness and inclination, density, peripheral pixels). It is possible to improve image reproducibility by adaptively correcting the signal level before performing the binarization process even when the values have various values depending on the correlation (or the like) or the MTF characteristics of the sensor. Become.

By the image area determination, the character / line drawing area, the outline portion of the image, and the halftone dot /
It is possible to effectively detect the photographic image area, and it is possible to perform a unique binarization process on each area. Therefore, in the character / line-drawing area, it is possible to perform smoothing that keeps the continuity in the contour portion as much as possible, and it is possible to obtain a binary image with few notches. In regions where the frequency is high, simple binarization processing with a fixed threshold is performed, so there are almost no adverse effects such as thin line breaks, "crushing", "blurring", or "tail tailing". In the dot / photographic image area, since the density level is pseudo-displayed as binary data by halftone processing, a high-quality binary image can be obtained.

Further, when compression coding is performed on the binary data by MH coding or the like, even if the run length of the binary data is shortened by the notch, the compression capability is extremely deteriorated. Since the occurrence of notches themselves can be suppressed, it is possible to prevent the amount of transmission data from unnecessarily increasing.

As a modification of this embodiment, as shown in FIG. 12, the binarized data 52a obtained by binarizing the read image data (multi-valued image data) is output from the output terminal 57. By doing so, it is also possible to use it as an image reading device (image scanner device), in which case the obtained binarized data is processed by another data processing device (for example, an image generating device, a data recording / reproducing device or a data recording device). Transceiver device).

In the above embodiments, a line sensor is used as a means for reading an image, and only one-dimensional scanning is shown. However, the present invention is not limited to this and other reading means (X-ray, A device using nuclear magnetic resonance, ultrasonic waves, etc.) may be used.

[0119]

EFFECTS OF THE INVENTION According to the present invention, when binarizing multi-valued image data, most of the adverse effects such as thin line breaks, "crushing", "blurring", "tail trailing" or "corner chipping" occur. A high-quality binary image is obtained by adaptively performing a smoothing process to prevent discontinuity (notch) of the binary image only in a portion where the density level is unstable, such as an outline portion of the image. In addition, it is possible to prevent the transmission data amount from unnecessarily increasing when encoded by MH encoding or the like.

Further, according to the present invention, it is possible to achieve high image quality with a small memory amount and a simple structure.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a binarizing device as a first embodiment of the present invention.

FIG. 2 is an explanatory diagram showing an example of a target pixel P and peripheral pixels which are target regions when performing binarization in the first embodiment.

FIG. 3 is an explanatory diagram showing a target area in an image area determination unit 12 in the first embodiment.

FIG. 4 is a block diagram showing a specific example of the image area determination unit 12 in the first embodiment. .

FIG. 5 is an explanatory diagram showing a target area in the state probability determination unit 13 in the first embodiment.

FIG. 6 is an explanatory diagram for explaining a determination condition in a state probability determination unit 13 in the first embodiment.

FIG. 7 is a block diagram showing a specific example of a state probability determination unit 13 in the first embodiment.

FIG. 8 is an explanatory diagram for explaining a binarization process in the first embodiment.

FIG. 9 is a block diagram showing a binarization device as a second embodiment of the present invention.

FIG. 10 is a characteristic diagram showing input / output density characteristics in the first and second γ correction units 45 and 46 in the second embodiment.

FIG. 11 is an explanatory diagram showing weighting factors used in the smoothing unit 47 and the edge emphasizing unit 48 in the second embodiment.

FIG. 12 is a block diagram showing a modification of the second embodiment.

FIG. 13 is a block diagram showing a configuration of a conventional binarization device.

FIG. 14 is an explanatory diagram for explaining a binarization process in the binarization device of FIG. 13.

FIG. 15 is an explanatory diagram for explaining a binarization process in another conventional binarization device.

[Explanation of symbols]

1 ... Read image density level input terminal, 2 ... A / D converter,
3 ... Shading correction unit, 4 ... γ correction unit, 5 ... MTF
Correction unit, 6 ... Binarization unit, 7 ... Binary data output terminal, 8
-10 ... 1st-3rd threshold value addition part, 11 ... Threshold value selection part, 12 ... Image area determination part, 13 ... State probability determination part, 1
5, 17 ... First and second storage media, 23-26 ... Averaging unit, 27-30 ... Difference value calculation unit, 31-34 ... Comparison unit,
35 ... Judgment part, 36 ... Fourth threshold value addition part, 37 ... Selection part, 38 ... Comparison part, 39 ... Main scanning direction contour part judgment part,
42 ... Previous line run detection unit, 45 ... State probability determination unit, 4
4 ... Read sensor, 45, 46 ... First and second γ correction section,
47 ... Smoothing section, 48 ... Edge enhancement section, 49 ... Mixing section,
51 ... Halftone processing section, 52 ... Selection section, 53 ... Size conversion section, 54 ... Encoding section, 55 ... Modem section, 56 ... Line.

 ─────────────────────────────────────────────────── ─── Continuation of front page (72) Inventor Seiji Tanaka 1410 Inada, Katsuta City, Ibaraki Hitachi, Ltd. Information & Video Media Division

Claims (17)

[Claims] 1. A multi-valued image data obtained by scanning an image and sampling in pixel units is converted into binarized data.
An image for determining a feature of an image in the target pixel from multivalued image data of a pixel to be binarized (hereinafter referred to as a target pixel) and multivalued image data of peripheral pixels with respect to the target pixel in a binarizing device The binarized data of the pixel of interest is predicted based on the region determination means and the states of the binarized data of the peripheral pixel and the multivalued image data, and the state probability of the binarized data of the pixel of interest is predicted according to the prediction result State probability determining means and threshold value selecting means for adaptively selecting a desired threshold value from a plurality of threshold values according to the determination result by the image area determining means and / or the state probability determining means. And the multi-valued image data of the target pixel and the threshold value selected by the threshold value selection means are compared in size, and the multi-valued image data of the target pixel is converted into binarized data according to the comparison result. And a binarizing means for exchanging the data.
Quantizer. 2. The binarizing device according to claim 1, wherein
The binarizing device, wherein the image area determination means determines whether or not the pixel of interest is located at a contour portion of the image, as the determination of the image feature of the pixel of interest. 3. The binarizing device according to claim 2,
The image area determination means determines whether or not the pixel of interest is located at the contour portion of the image generated in a specific direction as the determination of the characteristic of the image of the pixel of interest. apparatus. 4. The binarization apparatus according to claim 2, wherein the image area determination unit determines the peripheral pixel of the peripheral pixel from the multivalued image data of the peripheral pixel and the multivalued image data of the target pixel. Density difference calculating means for obtaining a difference between the density level and the density level of the pixel of interest (hereinafter referred to as density level difference),
Comparison means for comparing the density level difference obtained by the density difference calculation means with a predetermined threshold value, and determination of whether or not the pixel of interest is located at the contour portion according to the comparison result by the comparison means. A binarizing device comprising: 5. The binarizing device according to claim 2, wherein the image area determination unit determines, from the multivalued image data of the peripheral pixels, a predetermined number of the peripheral pixels arranged in a predetermined direction. The density level average value and the density of the pixel of interest are calculated from the average density level calculation means for obtaining the density level average value of the pixel and the calculation result obtained by the average density level calculation means and the multi-valued image data of the pixel of interest. A density difference calculating means for obtaining a difference from the level (hereinafter referred to as a density level difference), a comparing means for comparing the density level difference obtained by the density difference calculating means with a predetermined threshold value, and the comparing means. A binarizing device for deciding whether or not the pixel of interest is located in the contour portion in accordance with the result of the comparison. 6. The binarizing apparatus according to claim 4, wherein the determination unit determines that the pixel of interest is the pixel of interest when the density level difference obtained by the density difference calculation unit is larger than the threshold value. A binarization device, characterized in that it is determined to be located in the contour portion. 7. The binarizing device according to claim 1, wherein
The state probability determining means determines whether or not the pixel of interest is 2 based on the states of the binarized data and multi-valued image data of the peripheral pixels.
Binarized data that retains continuity with the binarized data of the peripheral pixels is predicted as the binarized data, and the state probability of the binarized data of the target pixel is determined according to the prediction result. Characterizing binarization device. 8. The binarizing device according to claim 1, wherein the state probability determining unit includes binarized data of pixels arranged in a predetermined direction across the target pixel among the peripheral pixels, and A contour portion detecting means for detecting whether or not the pixel of interest is located at a contour portion of an image generated in a direction perpendicular to the predetermined direction based on multi-valued image data, and a state of binary data of the peripheral pixels. A state detecting means for detecting
A state probability determining unit that determines a state probability of the binarized data of the pixel of interest based on the detection results of the contour detecting unit and the state detecting unit.
Quantizer. 9. The binarizing apparatus according to claim 8,
In the case where the contour detection unit detects that the pixel of interest is located at the contour of the image generated in the direction perpendicular to the predetermined direction, the state detection unit detects the pixel of interest among the peripheral pixels and the pixel of interest. When it is detected that the binarized data of a predetermined number of pixels that are adjacent to the pixel arranged in the predetermined direction across the target pixel and that are continuously arranged in the predetermined direction have the same value. The binarization apparatus, wherein the state probability determining means determines that the pixel of interest is located at a corner of an image, and makes the state probability of the binarized data of the pixel of interest indefinite. 10. The binarizing device according to claim 8, wherein the contour portion detecting unit detects that the pixel of interest is located at a contour portion of an image generated in a direction perpendicular to the predetermined direction. , The state detecting means is provided with a predetermined number of adjacent pixels, which are adjacent to the pixel of interest and pixels lined up in the predetermined direction with the pixel of interest sandwiched therebetween, among the peripheral pixels. When it is detected that the binarized data of the pixels are not partially the same, the state probability determining unit determines that the pixel of interest is located in the contour portion of the image that occurs only in the direction perpendicular to the predetermined direction, Of the pixels adjacent to the pixel of interest, the two pixels that are located in the same outline as the pixel of interest and have already been binarized.
A binarization device, wherein the binarized data is defined as a state probability of the binarized data of the pixel of interest. 11. The binarizing device according to claim 8, wherein when the contour portion detecting unit detects that the pixel of interest is not located in a contour portion of an image generated in a direction perpendicular to the predetermined direction, The state probability determining means is a pixel that is adjacent to the pixel of interest and is arranged in the predetermined direction together with the pixel of interest, and has already been binarized.
A binarization device, wherein the binarized data is defined as a state probability of the binarized data of the pixel of interest. 12. The binarizing device according to claim 8, 9, 10 or 11, wherein the image area determining unit determines that the target pixel is generated in a specific direction as the determination of the feature of the image in the target pixel. The state probability determining means determines whether the pixel of interest is not located in the contour portion according to the determination result of the image area determining means. If it is determined that the state probability of the binarized data of the target pixel is indefinite, the binarization device. 13. The binarizing device according to claim 1, 8, 9, 10, 11 or 12, wherein the binarized data is a binary value having a maximum density level of a first value and a minimum density level. When a certain second value is taken and the state probability of the binarized data of the pixel of interest judged by the state probability judging means can take the first value, the second value and indefinite, The threshold value selecting unit determines the state probability of the binarized data of the pixel of interest to be the first value from the plurality of threshold values when the state probability determining unit determines that the state probability is indefinite. When the first threshold value which is a value between the second value and the state probability determining means determines the state probability of the binarized data of the pixel of interest as the second value, Selecting a second threshold value that is between the first value and the first threshold value, and When the probability determining means determines that the state probability of the binarized data of the pixel of interest is the first value, the third threshold value is a value between the second value and the first threshold value. A binarization device characterized by selecting a threshold value. 14. The binarizing apparatus according to claim 1, wherein the image is scanned line by line in a main scanning direction and line-sequentially in a sub scanning direction, and the binarized data is binary. In the case where a first value that is the maximum density level and a second value that is the minimum density level can be taken as values, the image area determination means determines the feature of the image in the target pixel as the target pixel. Determines whether or not it is located in the contour portion of an image that occurs perpendicularly to the main scanning direction and / or the sub-scanning direction, and the threshold value selecting means, among the plurality of threshold values,
A first threshold value that is a value between the first value and a second value, and a second threshold value that is a value between the first value and the first threshold value And a third threshold value that is a value between the second value and the first threshold value, and the image area determination unit determines that the pixel of interest is in the main scanning direction and / or
Alternatively, when it is determined that the pixel is located at the contour portion of the image that is generated perpendicularly to the sub-scanning direction, the state probability determining unit is a binary pixel of the peripheral pixels that are arranged in the main scanning direction with the pixel of interest sandwiched therebetween. The converted data and the multi-valued image data are used to detect whether or not the pixel of interest is located in the contour portion of the image generated in the direction perpendicular to the main scanning direction, and the pixel of interest is perpendicular to the main scanning direction. When it is detected that the pixel is not located in the contour portion of the image generated in the above, the state probability determination unit is a pixel which is arranged in the main scanning direction together with the target pixel among the pixels adjacent to the target pixel, The binarized data of the binarized pixel is defined as the state probability of the binarized data of the pixel of interest, and the threshold value selecting means selects the second value or the third value. , The target pixel is in the main scanning direction When the state probability determining unit detects that the pixel is located at the contour portion of the image generated in the vertical direction, the state probability determining unit determines the pixel of interest and the pixel lined up in the predetermined direction with the pixel of interest sandwiched therebetween. When the states of the binarized data of a predetermined number of pixels that are adjacent to each other and are continuously arranged in the predetermined direction are detected, and that the binarized data of the predetermined number of pixels are all the same value The state probability determining means determines the state probability of the binarized data of the target pixel to be indeterminate, assuming that the target pixel is located at a corner of the image, and the threshold value selecting means sets the first value to the first value. When it is detected that the binarized data of the predetermined number of pixels are not partially the same value, the state probability determination means determines that the pixel of interest is an image generated only in a direction perpendicular to the main scanning direction. As noted in the above Among the pixels adjacent to the pixel, a pixel located in the same contour as the target pixel,
The binarized data of a pixel that has already been binarized is defined as the state probability of the binarized data of the pixel of interest, and the threshold selection means selects the second value or the third value. Then, the image area determination unit determines that the target pixel is in the main scanning direction and / or
Alternatively, when it is determined that the state probability of the binarized data of the pixel of interest is indefinite, the state probability determining unit determines that the state probability is not fixed at the contour portion of the image generated perpendicularly to the sub-scanning direction. Is a binarization device, wherein the first value is selected. 15. A reading device for reading an image written on a document and outputting it as data, according to claim 1.
4. A reading device comprising the binarizing device according to any one of 4. 16. A facsimile apparatus for reading an image written on a manuscript and transmitting it as data via a line, according to any one of claims 1 to 14.
A facsimile machine characterized by comprising a digitizing device. 17. A facsimile device comprising a reading device for reading an image written on a manuscript and outputting data, and a transmitting device for inputting the data and transmitting the data via a line, wherein the reading device is a device. A facsimile apparatus using the reading apparatus according to Item 15.