JPH0691614B2 - Image signal binarization device - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image signal binarizing device for binarizing image signals sequentially output from a pixel decomposition type scanner into black and white pixels.

[0002]

2. Description of the Related Art In order to decompose an image on an original, such as a laminated original, a mixed document, a color original, or the like, in which the background color or density is partially changed into pixels, and binarize it, It is necessary to change the threshold for binarization according to the background color and density of the document.

Therefore, conventionally, the density level signal of each pixel is divided at a predetermined ratio using a dividing resistance, the divided signal is low-pass filtered to form a so-called floating threshold value, and the density level signal is obtained by this floating threshold value. A device for binarizing is used.

[0004]

However, in such a conventional apparatus, since the time constant element is used for the means for low-pass filtering, the density of the background such as the boundary between the originals is drastically changed. In the part, there was a problem that it was not possible to sufficiently follow this change in density and the threshold value could not be set with high accuracy.

The present invention eliminates the inconvenience of such a conventional apparatus, and appropriately adjusts the image signal to 2 even if the background density of the document changes.
It is an object of the present invention to provide an image signal binarization device that can be digitized.

[0006]

SUMMARY OF THE INVENTION The present invention is a pixel signal binarizing apparatus for binarizing image signals sequentially output from a pixel decomposition type scanner into black and white pixels. First average value calculating means for sequentially calculating an average value of the densities of a plurality of reference pixels continuous to the pixel while moving the target pixel in the scanning direction of the scanner, and an average value of the densities of the target pixel and the reference pixel. Are sequentially calculated while moving the target pixel in the direction opposite to the scanning direction of the scanner, the first average value calculated by the first average value calculation means, and the second average value calculation means. The difference calculating means for sequentially calculating the difference from the second average value calculated by the average value calculating means for each pixel of interest, and the predetermined value based on the change in the difference within a predetermined area centered on a specific pixel. Within the area, run from the above specific pixel Selection means for selecting the first average value at the portion on the start position side and the second average value at the portion on the scanning end position side from the pixel of interest, and a change threshold value is calculated based on the output of the selection means. A change threshold calculation means and a threshold calculation means for calculating a threshold value based on the first average value and the second average value are provided, and the threshold value is corrected by the change threshold value.

Further, when the difference of the target pixel has a maximum value within a predetermined area centered on the target pixel, the selecting means selects a portion of the predetermined area closer to the scanning start position than the target pixel. Then, the first average value may be selected so that the second average value is selected at a portion closer to the scanning end position than the pixel of interest.

Alternatively, the selecting means pays attention to the variation of the absolute value of the difference calculated by the difference calculating means, and when the absolute value of the difference having a magnitude equal to or larger than a certain value continues continuously, the continuous Detect the largest of the absolute values of the above differences,
Within the predetermined area centered on the pixel corresponding to the difference having the maximum absolute value, the first average value is calculated at a portion closer to the scanning start position than the corresponding pixel and closer to the scanning end position than the target pixel. It is possible to use a part that selects the second average value. Further, the first average value calculation means and the second average value calculation means perform weighted average value calculation in which a weighting coefficient that decreases exponentially according to the distance from the pixel of interest is assigned to each reference pixel. Use what you do.

[0009]

Therefore, since the threshold value is calculated based on the forward average value, the backward average value of the pixel densities, and the difference between these average values at each pixel, the image of the portion where the density changes abruptly is calculated. The signal can also be appropriately binarized.

[0010]

Embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

FIG. 1 is a block diagram showing an embodiment of the present invention.

In the figure, a density signal (image signal) Px for each pixel applied from a scanner (not shown) is temporarily stored in the line memory 1 for a plurality of scanning lines.

The density signal Px stored in the line memory 1
Is added to the forward direction average value calculation unit 2 in the order corresponding to the scanning direction of the scanner for each scanning line, and is also added to the backward direction average value calculation unit 3 in the order corresponding to the anti-scanning direction.

The forward average value calculation unit 2 calculates the forward average value Mf of the densities around each pixel by the following equation (I) based on the sequentially added density signals Px.

[0015]

[Equation 1]

However, S ₀ is the density of the pixel for which the forward average value is to be calculated, m ₁ is the weighting coefficient, and Mf ′ is the forward average value obtained for the immediately preceding pixel.

The formula (I) will be described below.

Generally, there is a correlation between the density of the pixel of interest to be binarized and the density of a pixel (hereinafter referred to as a reference pixel) located in the vicinity of this pixel of interest, and
The closer the reference pixel is to the target pixel, the stronger the relationship with the target pixel, and the stronger the correlation in density.

Therefore, the weighted average value calculated by assigning the weighting coefficient, which decreases exponentially as the distance from the pixel of interest increases to each reference pixel according to the distance, is the density of pixels near the pixel of interest. The distribution, that is, the background density is reflected.

Therefore, by performing a predetermined calculation (for example, conversion by a linear function) on the weighted average value, a threshold value reflecting the background density can be obtained.

Now, assuming that the density of the pixel of interest is S ₀ , the density and the weighting coefficient of the reference pixel that are separated from the pixel of interest by a distance (i × r) (i is an integer, r is the width of one pixel) are respectively set.

[0022]

[Equation 2]

When (0 <m1 <1), the weighted average value Mf of the reference pixels including the target pixel is given by the following equation (II).

[0024]

[Equation 3]

Transforming this equation (II), the following equation (II
I) is obtained.

[0026]

[Equation 4]

Here,

[0028]

[Equation 5]

Therefore, the above formula (III) is as follows.

[0030]

[Equation 6]

In this way, the above formula (I) was obtained.

The reverse direction average value calculation unit 3 is similar to the forward direction average value calculation unit 2, and the reverse direction average value M of the densities around each pixel.
r is calculated by the following equation (IV).

[0033]

[Equation 7]

However, S ₀ is the density of the pixel for which the backward average value is to be calculated, m ₂ is the weighting coefficient, and Mr ′ is the backward average value obtained for the immediately preceding pixel.

The forward direction average value Mf calculated by the forward direction average value calculation unit 2 and the backward direction average value Mr calculated by the backward direction average value calculation unit 3 are different from each other by the average value calculation unit 4 for each scanning line segment. It is added to the calculation unit 5 and the average value selection unit 6.

The average value calculation unit 4 calculates the average value Mm of the forward average value Mf and the backward average value Mr for each pixel,
This average value Mm is added to the threshold value calculation unit 7.

The threshold value calculation unit 7 executes the following equation (V) based on the added average value Mm to calculate the threshold value TH of each pixel and adds it to the threshold value correction unit 8.

[0038]

[Equation 8]

However, k ₁ and k ₂ are constants.

Now, consider the case where the density signal Px changes as shown in FIG. In the figure, the direction from left to right is the scanning direction.

As described above, the forward average value computing unit 2 uses the equation (I), and the backward average value computing unit 3 uses the equation (I).
V) calculates the forward average value Mf and the backward average value Mr, respectively.

Since the expressions (I) and (IV) both take the form of a recurrence expression, there is a great advantage that the structure of the arithmetic circuit can be realized very simply, while the density state of each pixel is directly changed. Since it does not appear, it has a delay characteristic as shown in the figure.

That is, the forward average value Mf is the density signal P
The backward average value Mr is delayed with respect to the change of x in the scanning direction.
Is delayed with respect to the change of the density signal Px in the anti-scanning direction.

Therefore, considering the scanning direction,
The rising point P1 and the falling point P of the density signal Px
In the case of 2, the difference between the two average values is remarkably large as compared with the state in the vicinity thereof.

Therefore, the difference between the forward average value Mf and the backward average value Mr is calculated for each pixel, and the pixel having the largest difference in the front and rear (A + B) range around each pixel is searched for,
The forward average value Mf is selected for the portion up to the pixel that is separated from the pixel by the distance A in the anti-scanning direction. On the other hand, the backward average value Mr is selected in the portion up to the pixel which is separated from the pixel by the distance B in the scanning direction.

By doing so, it is possible to obtain an average value that more accurately represents the average of the density signal Px at the rising portion and the falling portion of the density signal Px and in the vicinity thereof.

The distances A and B are distances corresponding to the delay times of the backward average value calculation unit 3 and the forward direction average value calculation unit 2, respectively.

The selection of such an average value is realized by the difference calculator 5 and the average calculator 6.

The difference calculator 5 calculates the difference Sm between the forward average value Mf and the backward average value Mr for each pixel and outputs it to the average value calculator 6.

The average value calculation unit 6 compares the difference Sm in the range (A + B) before and after that pixel to find the pixel having the maximum difference Sm, and as described above, the forward and backward portions of the pixel. The average value Mf and the reverse direction average value Mr are selected and added to the change threshold calculation unit 9.

The change threshold calculation unit 9 calculates the change threshold THv of each pixel based on the average value added by the average value selection unit 6 in the same manner as the threshold calculation unit 7, and outputs this to the threshold correction unit 8.

The threshold correction unit 8 changes the threshold THv for the pixel for which the change threshold THv is output, and for the other pixels, the correction threshold THm (FIG. 2).
(See (b)) and outputs it to the binarization circuit 10.

The binarization circuit 10 has a density signal P for each pixel.
x is compared with the correction threshold value THm to form the binarized image signal SB, which is output to the next-stage device (not shown).

By the way, in addition to the above-mentioned conditions, the following conditions can be used as the conditions for selecting the forward average value Mf and the backward average value Mr.

That is, the difference between the forward average value Mf and the backward average value Mr is calculated for each pixel, and paying attention to the variation of the absolute value of this difference, the difference having a magnitude greater than a certain value is calculated. If there is a portion where the absolute value continues in succession, the maximum absolute value is searched for in the continuous portion, the pixel corresponding to the difference is detected, and the pixel is separated from the pixel by the distance A in the anti-scanning direction. For the portion up to the pixel, the forward average value Mf is selected. On the other hand, the backward average value Mr is selected in the portion up to the pixel which is separated from the pixel by the distance B in the scanning direction.

Further, in this case, the distances A and B are respectively calculated as the backward average value calculation unit 3 and the forward direction average value calculation unit 2.
Is the distance corresponding to the response delay time corresponding to the abrupt change in the scanning direction of the density signal Px of the output of the above, and is set based on the statistical result, or the difference S between the two average values is
Based on m, it is variably set as in the following equations (VI) and (VII).

[0057]

[Equation 9]

However, k ₃ , k ₄ , k ₅ , and k ₆ are constants.

Further, in this case, the average value selection unit 6 pays attention to the variation state of the absolute value of the difference | Sm | in the scanning direction, and the absolute value of the difference having a magnitude equal to or larger than a certain constant value S is continuous. And search for the part that follows. Then, the pixel corresponding to the difference having the maximum absolute value is detected in that portion, and the forward direction average value Mf and the backward direction average value Mr are selected in the portions before and after the pixel as described above. Is added to the change threshold calculation unit 9. However,
The constant value S is a constant statistically determined by the quality of the binarized image.

In the above embodiment, the forward average value Mf
Further, the average value Mm of the reverse direction average values Mr is further calculated, and the threshold value TH is calculated based on the average value Mm. Since there is not much difference between the average values Mr, either one may be used to calculate the threshold value TH.

In general, the density signal does not change much in continuous scanning lines. Therefore, the forward average value and the backward average value may be alternately calculated for each main scanning line.

Furthermore, the average value of the density signals of the pixels may be calculated not only in the scanning line and anti-scanning line directions but also in the third direction, and the state of this average value may be reflected in the threshold value.

The present invention can also be implemented by a microcomputer that realizes each element of the above-described embodiment as a functional block.

[0064]

As described above, according to the present invention,
Since the forward direction average value and the backward direction average value are switched in the portion where the change in the density signal is large, the average value of the density signal can be obtained more reliably. As a result, the threshold value can be made to follow the change of the density signal well, and the effect that the portion where the change of the density signal is large can be appropriately binarized is obtained.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention.

FIG. 2 is a waveform diagram ((a)) illustrating a relationship between a density signal and a forward direction average value and a backward direction average value, and a waveform diagram ((b)) illustrating a density signal and a correction threshold value.

[Explanation of symbols]

 2 forward direction average value calculation unit 3 reverse direction average value calculation unit 4 average value calculation unit 5 difference calculation unit 6 average value calculation unit 7 threshold value calculation unit 8 threshold value correction unit 9 change threshold value calculation unit 10 binarization circuit

Claims (4)

[Claims] 1. An image signal binarizing device for binarizing image signals sequentially output from a pixel decomposition type scanner into black and white pixels, wherein a pixel of interest to be binarized and a plurality of references continuous to the pixel of interest. A first average value calculating means for sequentially calculating an average value of the pixel densities while moving the target pixel in the scanning direction of the scanner; and an average value of the density of the target pixel and the reference pixel for the target pixel of the scanner Second average value calculating means for sequentially calculating while moving in the opposite direction to the scanning direction, first average value calculated by the first average value calculating means, and second average value calculating means for the second average value calculating means. Based on the change of the difference within a predetermined area centered on the specific pixel, a difference calculation means for sequentially calculating a difference from the average value of 2 for each target pixel, and within the predetermined area, the specific pixel Above at the scanning start position side Selecting means for selecting the second average value of the first average value at a portion closer to the scanning end position than the pixel of interest; change threshold calculating means for calculating a change threshold based on the output of the selecting means; An image signal binarizing apparatus comprising: a threshold value calculating unit that calculates a threshold value based on a first average value and a second average value, and correcting the threshold value by the change threshold value. 2. The selecting means, when the difference of the target pixel has a maximum value within a predetermined area centered on the target pixel, a portion of the predetermined area closer to the scanning start position than the target pixel. 2. The image signal binarizing apparatus according to claim 1, wherein the first average value is selected as the second average value at a portion closer to the scanning end position than the pixel of interest. 3. The selecting means pays attention to the variation of the absolute value of the difference calculated by the difference calculating means, and when the absolute value of the difference having a magnitude equal to or larger than a certain constant value continues continuously, the continuous Detecting the maximum of the absolute values of the difference, and within a predetermined area centered on the pixel corresponding to the difference having the maximum absolute value, in the portion closer to the scanning start position than the corresponding pixel, 2. The image signal binarization apparatus according to claim 1, wherein the second average value is selected in a portion of the first average value closer to the scanning end position than the pixel of interest. 4. The weighted average in which the first average value calculation means and the second average value calculation means assign a weighting coefficient that becomes exponentially smaller according to the distance from the pixel of interest to each of the reference pixels. The image signal binarization apparatus according to claim 1, 2, or 3, wherein a value operation is executed.