JPS6359272A - Picture processor - Google Patents

Abstract

PURPOSE:To obtain a binary image with high quality, by extracting only an area where high resolution is requested such as a character, etc., separately from a background part, or a photographing area, etc., and setting a threshold value corresponding to a picture signal at every small area. CONSTITUTION:An image is divided into blocks consisting of (m Xm)-number of picture elements, and the maximum value Smax, the minimum value Smin, and a mean value S0 of a picture signal level at every block are found. A differential value D=Smax-Smin is compared with a value T set in advance, and when it is D>T, the picture signal level of each picture element of the said block is binarized setting the S0 as a threshold value. In other words, the picture signal level Sij (1<=i<=m, 1<=j<=m) of a certain picture element is set as a black when it approaches to a black level nearer than the S0, and is set as a white when the former approaches to a white level nearer than the latter. Meanwhile, when it is D<=T, the signal level is compared with a value ST, and is binarized. Namely, the picture signal level is set as the black when it approaches to the black level nearer than the ST, and as the white when it approaches to the white level nearer than the ST.

Description

DETAILED DESCRIPTION OF THE INVENTION (Technical field to which the invention pertains) The present invention provides an image processing device that has a simple configuration and can read images with low contrast well.

(Prior Art) Documents read by facsimiles include those with high density contrast, those with low density contrast, and those with colored backgrounds.

Various types exist.

Conventional devices detect the white level of the background of the document and
This prevents the background from turning black even if the original is colored.

As a means to enable good reproduction of documents with low contrast, the binarization threshold is set closer to white.

However, for this reason, even for high-contrast documents, the binarization threshold is set to be whiter than necessary, which causes deterioration in resolution.

An effective method for improving the resolution is to perform binarization using the average value of the image signal levels of pixels in the vicinity of the first pixel of interest. However, this method cannot handle the case where all neighboring pixels have the same image signal level, and has the disadvantage that abnormal dots may occur due to the influence of noise contained in the image signal.

(Object of the Invention) An object of the present invention is to solve these drawbacks and provide an image processing device that can automatically set a threshold value according to the image signal content and obtain a high-quality binarized image. It is in.

(Structure of the Invention) (Characteristics of the Invention and Differences from the Prior Art) The present invention determines an independent threshold value for each small area, and in a part where the image signal level changes significantly, the signal level of the neighboring pixels is determined. The most important feature is that a threshold value is set to an intermediate value of , and a threshold value that is set independently of the pixel signal level value of a neighboring pixel is used in a portion where the change in signal level is small, regardless of the value of the pixel signal level of a neighboring pixel.

(Example) [Example 1] Divide the image into blocks of mXm pixels.

Maximum value S□, minimum value S of image signal level for each block
ml i, as well as the average value S. seek.

The value of the difference value D=Smaf Sml+ is compared with a predetermined value T, and if it is DOT, the pixel signal level of each pixel of the block is binarized using So as a threshold value.

That is, if the image signal level 5z (1≦i≦m, 1≦j≦l11) of a certain pixel is closer to the black level than So, it is considered black, and if it is closer to the white level, it is considered white.

On the other hand, if D≦T, is the predetermined value S? Compare and binarize. That is, if the one-picture signal level is S, closer to the black level, it is black, and if it is closer to the white level, it is white.

FIGS. 1 and 2 show examples of image signals and processed images for explaining the functions of the apparatus of the present invention.

FIG. 1 shows an example of an image signal within one block in the case of m=4. The small square indicates the image signal level (black level: 0
．． White level: 31).゛In this case, the average value is S, = 12. The maximum value and minimum value are S□x”23.
S−+−=8−D=5.

If T=10, then D>T, so
As the threshold value, So, that is, 12 is used, and each pixel signal is binarized as shown in FIG.

Note that when the image signal level is equal to the threshold value, it is assumed to be white.

3 and 4 show other examples of image signals and processed images for explaining the functions of the apparatus of the present invention.

FIG. 3 shows another image signal example, 50=30°S□r=
31. S-1-=28. In the case of T=10, where D=3, since D<T, the threshold value S9 is used.

The value of S is usually chosen to be close to the middle between the white level (31) and the black level (0), so if S, = 18, the full picture signal is larger than S, so the fourth As shown in the figure, all pixels become white.

Such an example occurs in a background portion of a manuscript where no characters or graphics are written.

FIGS. 5 and 6 show further examples of image signals and processed images for explaining the functions of the apparatus of the present invention.

FIG. 5 shows an example of an image signal generated at the center of a black area, and since D<T, the binary signal becomes as shown in FIG. 6.

In the case of FIGS. 3 and 5 when all pixels are binarized using a constant threshold value 18 regardless of the present apparatus, FIG.

The result is the same as in Fig. 6, but in the case of the image signal example in Fig. 1, the 7th
The image is binarized as shown in the figure.

FIG. 7 is an example of an image processed by a conventional device.

Comparing this with FIG. 2, pixels 1, 2.3 and pixel 4 are different. This difference is the reason why a binarized image with good resolution can be obtained in this method.

In the above description, the threshold value S is used for blocks where D>T. was determined as the average of the image signal level, but it can be set variously between S□ and Sm1n.

[Embodiment 2] FIGS. 8 and 9 show an example of an image signal and an example of a processed image when processing is performed pixel by pixel.

In FIG. 8, a case will be described in which the central pixel of nine pixels is binarized. The pixel signal level of the relevant pixel is X, and the pixel signal levels of the surrounding eight pixels are at b* Cr.
Let d+ e+ L g+ h.

a-Xg b-x, c-x, d-x, e-x,
The values of f−x, g−X+h−x are determined, and the largest value among these absolute values is set as Dl.

Also, X and a, t't Ct d+ e* L gt
The average value of the No. 9 hardening level of h is assumed to be 81. The value of D is compared with a predetermined value T1, and if Dl>T, the image signal X is binarized using a threshold value S1.

That is, if X≧S, the color is white, and if X×S□, the color is black.

On the other hand, if D<T1, then the predetermined value S? It is binarized by

For example, the image signal example shown in FIG. 9 will be explained.

Considering the case where pixel P is binarized, in this case, D□=
12,51=16.7. If T1=8, then D>rx
Therefore, the pixel P is binarized using the threshold value S8, and 20) Since S 1 = 16.7, the pixel P becomes white.

On the other hand, in the case of pixel Q, =7. S□=25.8. Since D1≦T1, binarization is performed using S. S? =18, the pixel Q becomes white.

In FIG. 8, referring to the eight pixels surrounding the pixel of interest, Sl, D
I have calculated l, but it is possible to calculate only 4 pixels on the top, bottom, left and right, or to include 200 pixels as shown in Fig. 10, Sl, D□
Various methods are possible, such as finding the value of .

FIG. 10 shows another example of the configuration when processing is performed pixel by pixel.

In the first embodiment, a case has been described in which processing is performed block by block, and in Example 2, processing is performed pixel by pixel, but the common points are as follows.

(1) When the change in the image signal level near the pixel of interest is large, it is assumed that the change in the image signal level contains information, and the pixel of interest is binarized using a threshold value determined from the pixel signal level of the neighboring pixel. do.

(2) When the change in the image signal level near the pixel of interest is small, the change in the image signal level is assumed to be due to noise, etc., and is binarized using a threshold that is set independently of the image signal level of the nearby pixel. .

In the above explanation, white level, black level, threshold S? , the case where the parameter T (or T) for threshold selection is set to a constant value has been explained. However, these do not necessarily have to be constant values, and all or some of them can be changed depending on the content of the image signal.

This will be explained below.

If the contrast of the original image is low, even if information is included in changes in shading, D (or ratio) will be small, so D < T (or n t < T 1), resulting in poor resolution. Prone.

As a countermeasure for this, for example, if the contrast is small, T
It is sufficient if the parameters are automatically set so that the value of (or T□) also becomes small. To that end, as stated in Japanese Patent Application No. 156065/1980,
The white level and black level may be detected from the original image, and based on these, the threshold value S9, parameter T (or T1), etc. may be calculated.

[Embodiment 3] FIG. 11 is a diagram showing the configuration of an embodiment of the present invention.
Indicates the case where blocks of X m are processed as a unit, and 5
is a line memory, 6 is a comparator, 7 is a black and white level detection circuit, 8 is a block average (So) calculation circuit, 9 is an intra-block difference value CD) calculation circuit, 10 is a comparator, 1
1.12 is a memory.

In order to operate these, the white level Sw and black level S of the image are determined by the monochrome level detection circuit 7 while storing the image signals that are inputted line-by-line in the line memory 5. In this case, using the values of Sl and S8 obtained from the image signals up to the n-th line, the intra-block difference value detection circuit 9 that processes the image signal of the (n+1)-th line detects the maximum image signal level of each block. Find the difference between the value S□ and the minimum value s+wta.

The comparator 10 compares the output of the intra-block difference value detection circuit 9 with the output T of the black and white level detection circuit 7, and D)T
If so, the output is set to High level, and if D≦T, the output is set to Low level.

Since the comparator outputs independent values for each block, these are stored in the -time memory 11.

During this time, the block average calculation circuit 8 calculates the average value S of the image signal level for each block. are calculated and stored in the memory 12.

When the mth line image signal has been input.

The image signal level of each pixel, the average value S0 of each block in the first block row, and the comparison results between D and T are stored in the memory 5.12.11, respectively.

Immediately before storing the pixel signal on the m+1th line in the line memory 5, the already stored pixel signal on the first line is read out and input to the comparator 6.

The comparator 6 compares the image signal level a output from the line memory 5 with the output from the selector 13, and if a≧b, outputs a light level (corresponding to white).

If a < b, the output is set to Low level (corresponding to black).

In this case, the output of the memory 11 represents the threshold selection signal of the block to which the image signal belongs, and the selector 13 selects So if the output of the memory 11 is High level, and S? if the output of the memory 11 is Low level. is selected and input to comparator 6. As a result, the comparator 6 outputs an image signal whose threshold value is automatically selected from S, , S, depending on the intra-block difference value. Moreover, the black and white level is S,,S.

is automatically detected, so even in the case of an image with low contrast, an optimally binarized image signal can be obtained.

In this way, while sequentially processing the image signal of the first line,
The line memory 5 stores the image signal of the (m+1)th line. Hereinafter, while processing the image signal of the 2nd, 3rd, etc. lines, the m+2. The image signals of the m+3th line are sequentially stored in the line memory 5, and at the same time, the image signals of the m+1th line and subsequent lines are stored in So.

The values of S, , T, and D are determined and the above-mentioned process is repeated.

[Embodiment 4] FIG. 12 shows an embodiment in which the processing of the present invention is performed simultaneously while performing identification processing of an image containing mixed text and photographs.
5 is a line memory, 16.17 is a maximum value and minimum value detection circuit for each block, 18 is a calculation circuit for calculating a difference value for each block, 19 is a calculation circuit for a median value for each block, 2
0 is a black and white level detection circuit, 21 is an arithmetic circuit, 22 is a comparator, and 24 is a comparator for binarization processing.

To operate this, as in the third embodiment, while inputting image signals line sequentially and storing them in the line memory 15, the black and white level detection circuit 20 detects the white level SW, the black level S,
seek.

Find the value of.

The maximum value detection circuit 16 compares the sequentially input image signals and detects the maximum value S□ for each block.

Similarly, the minimum value detection circuit 17 detects the minimum value Smi for each block.
. Detect.

If the block size is mXm, then the mth line
At the time when the k (k=1.2...)th image signal is input, the maximum value S□2 and the minimum value Sm of the th block
l+x is found. Immediately from this Sva*Xs S+mlo, the arithmetic circuit 18 calculates D=S+++a! S+
ala is determined and inputted to the comparator 22 together with the signal Syu output from the arithmetic circuit 21. Comparator 22
If the value of D is large, the area is determined to be a character area requiring high resolution, and if the value of D is small, it is determined to be an area that requires shading expression such as a photograph. Therefore D
If ≧Si, the output S2 of the comparator 22 is immediately stored in the memory 23 for each block.

Since this is used as a threshold for binarization, it is stored in the memory 27 for each block.

Similar to the comparator 6 in FIG.
is input to the comparator 24, compared with the output of the selector 26, and output as a binary value.

In this case, the selector 26 receives the threshold signal S output from the ROM 25 by the signal S output from the memory 23.
4 and the threshold signal S0 outputted from the memory 27 is selected and inputted to the comparator 24 as a signal. That is, if S3 is at a high level, the same threshold value S0 is selected to be given to each pixel in the block so that high resolution can be obtained.

On the other hand, if S is at Lot+ level, the dither threshold value S4 stored in the ROM 25 is selected so as to enable gradation expression. Since the dither threshold value S4 is set so that the threshold value changes periodically depending on the pixel position, the density of black pixels changes depending on the value of the image signal level a, making it possible to express light and shade. In this way, the image signals that are line-sequentially input to terminal A are optimally binarized and then output from terminal B with a delay of 4 lines.

However, it does not necessarily have to be this value, and S□,
and 5sln.

For example, if you want to express it darkly, you can choose a slightly thicker one, and if you want to make it lighter, you can choose a slightly smaller one; calculation is not necessarily necessary, and you may change the value in this vicinity depending on the requirements of the output image.

FIG. 13 is a block diagram of an embodiment showing the configuration of the black and white level detection circuit of the present invention, which determines the parameters of a new scanning line based on the image signal of the previous scanning line.
8.29 is a comparator, 30.31 is a counter, 3
2 is an arithmetic circuit for calculating contrast, 33 and 34 are latch circuits that output a white level and a black level, respectively, and 35 is a gate circuit.

Here, the beginning of one screen of the latch circuit is set to an appropriate initial value.

In the following description, the white level will be expressed as a large value, and the black level will be expressed as a small value.

The image signal is input to comparators 28 and 29.

The comparator 28, together with the counter 30, detects the density level expected to be white from among the image signals on one scanning line.

That is, the image signal and the output of the counter 30 are input to the comparator 28 in synchronization with each pixel signal input, and their magnitudes are compared.

If the image signal is larger, the output from the comparator 28 becomes High, so a count pulse is input to the counter 30, and the count value of the counter 3° increases.

If the image signal is smaller, the output of the comparator 28 is Low, so the counter value does not change.

When this operation is performed for each image signal, the count value of the counter 30 continues to count up as long as it is smaller than the image signal level.

Therefore, by resetting the count for each scanning line, the level expected for white can be output as the output signal of the counter 30 for each scanning line.

Similarly, the comparator 29 and the counter 31 detect the level expected to be black for each scanning line, and output it as the output Y of the counter 31.

For example, the blackest possible level of image signal level 0 (
31) is defined as the whitest possible level, the counter 31 is preset to level (31) at the beginning of each scanning line, and thereafter, the counter 31 is preset to level (31) in synchronization with each image signal, and every time the output of the comparator 29 becomes Low, the counter 31 is preset to level (31). 31 count value to 1
decreases by

The comparator 29 compares the image signal with the output of the counter 31, and is set so that if the image signal is larger, the output is )Ihigh, and if it is smaller, the output is LOν.

As a result, as long as the output of the counter 31 is greater than the image signal, the count is decremented, so that at the end of each line, the counter 31 outputs a level expected to be black as a signal Y.

When the image signal input for one scanning line is completed, the counter 30.3
The outputs X and Y of 1 are respectively for the scanning line.

White represents the expected level and black represents the expected level. Signal X becomes a new white level candidate, and signal Y becomes a new black level candidate.

In determining these signals a and b as new parameters, the following must be considered.

That is, in a document or the like, there are cases where there are no black parts at all between lines.

In this case, the output Y of the counter 31 is equal to or close to X, but this is the lowest level value (closer to black) among the white parts of the document and is not the black level.

Therefore, the arithmetic circuit 32 calculates the difference between the signals X and Y. The value of this difference represents the contrast between black and white, and when this contrast is greater than a predetermined value, the output of the arithmetic circuit 32 becomes ``light''.

At this time, the output of the gate circuit 35 becomes High, and the values of the signals X and Y are latched into the latch circuits 33 and 34 as new white and black levels, respectively.

If the difference between signals
Therefore, the output 2 of the arithmetic circuit 32 becomes Low, and no latch signal is output to the output of the gate circuit 35.
As a result, the signals X, Y are not reproduced as new values,
The previous value will be output as is to the latch circuits 33 and 34.

In this way, parameters for processing the next scan line can be determined sequentially based on the image signal content of the current scan line.

Of course, if you have one line of memory, you can temporarily store the image signal in memory while determining the white and black levels from the image signal of the current scanning line, and use the white and black levels determined from the current scanning line image signal. It is also possible to process the current scan line.

In addition, in the explanation of FIG. 13, the output of the gate circuit 35 is High.
Although the case is shown in which the latch circuit 33 is operated to replace the white level with a new value X only in the case of gh, this condition is not necessarily necessary for the white level, and even if it is always replaced with a new value a. good.

In the above explanation, counters were used to determine the white level and black level for each scanning line, but it should be noted here that the outputs of the counters 30 and 31 do not necessarily strictly correspond to the whitest level and blackest level. It is not expressed.

That is, even if there is only one particularly black pixel or one particularly white pixel, the counter value changes only by 1, so it is hardly affected by abnormal signals due to dust or noise.

In the above embodiment, the case where the threshold value is explained for each block is shown, but a similar circuit can be easily realized for the case where the threshold value is selected for each pixel as explained in FIGS. 8 and 9. .

(Effects of the Invention) As explained above, the device of the present invention extracts only areas that require high resolution, such as characters, by distinguishing them from background areas, photographic areas, etc., and converts image signals into image signals for each small area. Since the threshold value is set accordingly, a high quality binarized image can be obtained.

Furthermore, by automatically identifying the black level and white level in the document and determining parameters based on this, even images with low contrast can be binarized with good results.

Therefore, if this device is applied to an image input device such as a facsimile, a binarized image with good resolution can be obtained.

[Brief explanation of drawings]

FIGS. 1 and 2 show examples of image signals and processed images for explaining the functions of the apparatus of the present invention. FIGS. 3 and 4 show other examples of image signals and processed images for explaining the functions of the apparatus of the present invention. Figures 5 and 6 are further examples of image signals and processed images for explaining the functions of the device of the present invention, Figure 7 is an example of images processed by the conventional device, and Figures 8 and 9 are processed pixel by pixel. An example of an image signal and an example of a processed image when
Figure 1 shows an embodiment of the device of the present invention. Figure 12 shows an example of a case that has a function to automatically detect the black and white level and a function to identify areas such as text and photographs, and also allows for 1-tone expression. Figure 13 has a function to automatically detect the black and white level. This is an example. P, Q...Pixel of interest, 5...Line memory, 6,10...Comparator, 7...Black and white level detection circuit, 8...Block average calculation circuit, 9...Intra-block difference value Detection circuit, 11.12,2
3,27...Memory. 13.26...Selector, 15...Line memory,
16... Maximum value detection circuit. 17... Minimum value detection circuit, 18... Difference value arithmetic circuit, 19... Median value arithmetic circuit, 20... Black and white level detection circuit, 21.32... Arithmetic circuit, 25... ROM, 22.
24, 28.29... Comparator, 30.31.
...Counter, 33.34...Latch circuit, 35...
・Gate circuit. Figure 1 Figure 2 Figure 3 ¥, Figure 5 Figure 6 y? ;7 Figure 8 Figure 9 Figure 11

Claims (6)

[Claims] (1) Means for determining the maximum and minimum values of the image signal level for each block consisting of a plurality of pixels in the image, and binarizing the image signal of the block from the maximum and minimum values of the image signal level. means for calculating a first threshold value for each block; selecting either the first threshold value or a second threshold value given independently of the maximum value and minimum value of the image signal level for each block; and means for binarizing the image signal of at least one block using either the selected first threshold value or the second threshold value. Processing equipment. (2) A means for determining the magnitude of change in the pixel signal level from neighboring pixels for each pixel, and a means for binarizing the pixel signal of the pixel from the pixel signal level of the pixel and neighboring pixels for each pixel means for calculating a first threshold; means for selecting for each pixel either the first threshold or a second threshold that is given independently of the pixel signal level of the pixel and neighboring pixels; An image processing apparatus comprising: means for binarizing a pixel signal of at least one pixel using either the selected first threshold value or the second threshold value. (3) Automatically detects the black level and white level from the original image,
The automatically detected difference value S_1 between the black level and white level and the difference value D between the maximum and minimum values of the image signal level for each block are determined, and the difference value S_1 and the difference value D are compared, and based on this result, , the first threshold value, or the second threshold value is selected.
). (4) Automatically detect the black level and white level from the original image, calculate the difference value S_1 between the automatically detected black level and white level, and calculate the magnitude of the change in image signal level between the difference value S_1 and the neighboring pixel. The image processing apparatus according to claim 2, wherein the image processing apparatus compares C and selects either the first threshold value or the second threshold value based on the result. (5) The image processing apparatus according to claim (3), further comprising means for determining and sequentially updating the black level and white level for each predetermined number of scanning lines. (6) The image processing apparatus according to claim (4), further comprising means for determining and sequentially updating the black level and white level for each predetermined scanning line.