JPH0779425B2 - Image processing 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 processing apparatus for processing a general image containing halftone density.

[0002]

2. Description of the Related Art When an image including halftone density is displayed on a device such as a thermal transfer printer and an output device such as a liquid crystal display, a density display by a dither method is conventionally known.
However, in the dither method, when many gradations are generally expressed,
The display resolution is reduced, and irregular noise is noticeable.

On the other hand, to the human eye, even if a general image such as a photograph is decomposed into a line component and a halftone component and displayed in a display like an illustration, it does not feel so unnatural. .

Therefore, the local variation rate of the image density is obtained, the line component such as the outline of a character or an object is detected by the magnitude of this value, and the display resolution is lowered by binarizing it with a fixed threshold value. An image that is displayed at a high resolution without displaying and that other intermediate gradation components are represented by a dither method, that is, a portion having a large local variation rate is subjected to simple binarization, and the others represent the density of the image by a dither method. Regarding processing equipment,
Proposed.

[0005]

However, when an image signal is detected and input by a line sensor such as a CCD,
Since the element characteristic of each pixel of the detection element varies, the reproducibility of the obtained image varies. In particular, when the local variation rate of the image is obtained in order to switch the signal processing depending on the characteristics of the input image, the variation in the output characteristics from the individual pixels making up the line sensor may cause the variation in the obtained local variation rate to be conspicuous. Becomes

Therefore, when the signal processing is switched depending on the local variation rate of the image, the switching decision may be erroneous, and if it is erroneously processed, the image quality may be deteriorated.

It is an object of the present invention to eliminate the above-mentioned drawbacks and to provide an image processing apparatus capable of expressing an image with high resolution and halftone.

[0008]

Image processing apparatus according to the present invention
Is a one-dimensional line sensor for inputting image signals.
And the image signals for multiple lines from this line sensor
Two-dimensional image signal storage means that can be stored, and this image signal
Unit data of the image signal stored in the storage means and this unit
The product of the coefficient corresponding to the data is added and the input
Means for obtaining the local variation rate of the image signal, and
The input image signal according to the local variation rate
Means for applying different signal processing to the local
Each coefficient corresponding to the unit data for obtaining the fluctuation rate is
Included in the coefficient string corresponding to the unit data string in the same line direction
Than the fluctuation of the coefficient value
Fluctuation of coefficient values included in the coefficient string corresponding to the unit data string
Is set to be larger .

[0009]

In the means for obtaining the local variation rate of the image used in the present invention, the product of the unit data constituting the image data and the predetermined coefficient corresponding to the unit data is calculated for the input two-dimensional image data. It has a product-sum operation unit that performs addition processing. The two-dimensional image data used for the calculation is prepared by scanning the line sensor or moving the input image for a plurality of lines of output signals from the line sensor.

Since the characteristics of the detection elements constituting the line sensor vary, the output signal from the line sensor also varies from detection element to detection element. On the other hand, with respect to the same detection element that constitutes the line sensor, since the element characteristics are constant, there is no problem that variations occur among the detection elements.

Therefore, the change of the coefficient value multiplied by the two-dimensional image data is set small in the same line direction and large in the direction extending over a plurality of lines, so that the influence of the fluctuation of the element characteristic for each detecting element is affected. To reduce
In addition, in the direction in which the element characteristics do not change, it is possible to set the coefficient for effectively obtaining the local change rate.

[0012]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 shows an image of a document 1 with a linear light source 2 such as a fluorescent lamp.
Illuminated by the CC and the self-occlusion lens array 3
An image is formed on the one-dimensional sensor 4 such as a D sensor. The image signal output by the one-dimensional sensor is converted into a digital signal by the A / D converter 5.

Next, the density characteristics are adjusted through density conversion ROMs 6, 7 and 8 for matching the density characteristics of the display printer 22. At this time, depending on the density characteristics of the original document 1, that is, a dark original document, a thin original document, or an original document without contrast, the output of the density conversion ROMs 6, 7, 8 is selected by the switch 9 to select an appropriate one. Specifically, FIG.
As shown in (a), it is for ordinary manuscripts, (b)
Is a characteristic of the density conversion ROM for a thin original and (c) is a density conversion ROM for a dark original.

Next, the high frequency component emphasizing circuit 10 obtains a local variation rate of this signal. Specifically, two-dimensional differentiation is performed by a convolution calculation circuit as described later. In this circuit, a line memory unit 11 for preparing a part of two-dimensional data based on a signal input as one-dimensional information, and a product-sum calculation unit 12 for performing two-dimensional differentiation from the two-dimensional data. Has become. Here, the high frequency emphasis circuit 10 will be described.

In a line sensor 4 such as a CCD, the pixels of the individual detecting elements generally have variations in sensitivity, so if the high frequency region is emphasized too much in the line direction, the variations become noticeable. However, the output signal does not change much in the document feeding direction or the sensor 4 moving direction. Therefore, a relatively large high-frequency emphasis is possible in this direction.

That is, when the following convolution calculation is performed, a difference in the magnitude of the variation of the parameter for obtaining the product is provided in the direction of the line sensor and the direction perpendicular to the line sensor, so that the variation of the detection element is made. It is possible to reduce the influence of and to perform effective high-frequency emphasis processing.

Therefore, it is practically preferable to change the degree of high-frequency emphasis in the line direction and the direction perpendicular to the line direction. The matrix of FIG. 3 shows a parameter group for high-frequency emphasis, and it is possible to control the degree of high-frequency emphasis in directions orthogonal to each other. The parameter a1 to c5 shown in this figure is sequentially producted to two-dimensional image data, and the result is added, which is called a convolution operation. High-frequency emphasis can be performed by selecting the parameter.

When executing the convolution operation, in order to execute the product, considering the matrix shown in FIG. 3, for example, data for five lines are required at the same time. The line memory unit 11 prepares this data.
Is. This configuration is shown in FIG. 4, and the line memory 70 has 6
Lines are prepared, and while one line is writing, the other 5 lines are reading.

That is, one write line is selected by the switching multiplexer 71 and data is written to the line memory. Next, for the line memory 70 which is not written by the decoding multiplexer 72,
The outputs out1 to out5 are output by switching so that the order of data is not disturbed in the paper feeding direction. Next, each time one data is input to the input I0, 5 outputs are read out to the respective outputs out1 to out5 to prepare 5 × 5 data. This data is input to the product-sum calculation unit 12.

FIG. 5 shows the configuration of the product-sum calculation unit 12. The operation matrix has the first row and the fifth row as shown in FIG.
The line, the second line, and the fourth line have the same parameters.
Therefore, the outputs out1 to out of the line memory 70
5 is input to SI1 to SI5 in FIG. 5, and first SI1 and S
The adder 80 calculates the sum of I5 and the adder 81 calculates the sum of SI2 and SI4. Next, the output results of the adders 80 and 81 and SI3 are respectively calculated as product-sum operation circuits 82, 83 and 8
Input to 4. This product-sum operation circuit takes in the respective parameters 85, 86, 87 and executes the product and the sum. That is, it is executed five times each time one data is input to I0 shown in FIG.

Using the result, all the sums are obtained by using the adder 88 and the adder 89, respectively, and the convolution operation is executed. Then, the high-frequency emphasized signal is output to the output SO. In this way, high-frequency emphasis is executed at high speed.

Next, a method of switching the signal processing depending on the input characteristic of the image will be described.

Here, in order to facilitate the explanation, consider a one-dimensional signal. It is assumed that the signal shown in FIG. 6A is input as the input image. Then, an electric signal obtained by the optical transfer characteristics of the self-locking lens array 3 or the like becomes a signal as shown in FIG. 6B.

When this signal is input to the differentiating circuit 10, it becomes a differentiated signal as shown in FIG. 6 (c). This signal is input to the simple binarization or dither method switching code conversion unit 13 (which is composed of a ROM). In the switching code conversion unit 13, of the signals shown in FIG. 6C, the level higher than that of the broken line 31 or the broken line 32.
For an input signal having a smaller level, a first code signal for selecting the processed signal is generated by the simple binarization described later and the signal is sent to the multiplexer 14.
When there is an input signal having a level between the broken line 31 and the broken line 32, a second code signal for selecting a signal processed by the dither method described later is generated and the signal is sent to the multiplexer 14.

As shown in FIG. 7, the contents of the ROM forming the switching code conversion unit 13 may be determined. According to this content, the switching multiplexer 14 to be described later selects a signal by the simple binarization when 0 and a signal by the dither method when 1.

However, since the high frequency component of the signal generated by the optical system such as the self-locking lens array 3 is deteriorated, the high frequency component is deteriorated. A clearer image can be obtained by correcting by adding.

Therefore, using the output signal of the central line of the line memory 11 and the output signal from the product-sum calculation unit 12,
Processing is performed by simple binarization or the dither method.

In this case, if all the high frequency components are added, the image quality is deteriorated due to overcorrection. Therefore, the correction coefficient K (0 <
K <1) is multiplied by the signal by the multiplier 15, and the adder 1
Enter in 6.

If the multiplier 15 is simplified so that the correction coefficient K is limited to one power of 2, even if the multiplier 15 is omitted and only the upper bits are input to the adder 16. The purpose can be achieved.

Now, the output signal of the product-sum calculation unit 12 is
Due to the processing time of the product-sum calculation unit, the output signal of the line memory unit 11 is delayed. Therefore, the signal of the line memory unit 11 is delayed via the delay circuit 17, and is synchronized with the output of the product-sum calculation unit 12 to be combined with the correction signal.
It is input to the adder 16. Since the output of the product-sum calculation unit 12 is added to the output signal of the line memory unit 11 in the adder 16, a signal with better high frequency characteristics than in FIG. 3B is output. However, it is better not to overcorrect.

This signal is then sent to the two comparators 1.
8 and 19 are input. The signal input to the comparator 18 is compared with a predetermined value 20 and simple binarization is performed. The purpose can be achieved even if the comparator 18 is omitted and the upper 1 bit of the output signal of the adder 16 is used as the binarized output signal. However, subtle adjustment cannot be expected.

On the other hand, the signal input from the adder 16 to the comparator 19 is sequentially compared with the data of the reference matrix memory 21 called a dither pattern and is output from the comparator 19 as a dithered signal.

The signal processed through these two processing paths is input to a signal switcher, for example, multiplexer 14, and either signal is selected. The input signal is switched by the output signal from the switching code conversion unit 13 described above, and either of these two types of signals is selectively output.

The output signal of the multiplexer 14 is displayed on the image output device 22 such as a thermal transfer printer. For example, when the image as shown in FIG. 6A is input, the output image of the image input device 22 becomes the image as shown in FIG. 6D. However, in this display, the binary densities are averaged and expressed as the density state when viewed.

In order to understand the effect of this embodiment, FIG. 7 shows an output example by the conventional image processing apparatus which does not perform the selective signal processing as described above. FIG. 7A shows a line sensor output signal. This is an output waveform in which the high-frequency characteristics in which the edges of the waveform are taken off are degraded due to the frequency characteristics of the optical system. When this is displayed by simple binarization, it is shown in FIG.
(B). This simple binarization process has a drawback in that an image signal smaller than a certain threshold value cannot be reproduced. Further, when the processing is performed only by the dither method and is displayed, it becomes as shown in FIG. In the processing by the dither method, all image signals are reproduced regardless of the amplitude of the image signal, but there is a drawback that the resolution of the image is lowered.

On the other hand, in the present invention, as shown in FIG. 6 (d), the portion including the high frequency component is binary and the portion not including the high frequency component is displayed by dither, so that the resolution is high. And halftones can also be expressed.

By thus preparing a plurality of signal processing systems and switching the processing according to the characteristics of the image, the present invention detects the line component and performs simple binarization on this signal to obtain the line component. Can be expressed without blurring, characters and photographs can be combined, and a sharp image can be obtained.

Further, since the determination of the switching of the output signal between the simple binarization and the dither method is performed by measuring the local variation rate of the image density, the determination process becomes easy. That is, the local variation rate can be easily obtained by spatial differentiation processing.

Further, since there is no concept of blocks as seen in the block-by-block image area separation method and the like, there is no problem such as generation of noise between blocks. (Embodiment 2) Next, the case of a color image processing apparatus will be described. In FIG. 9, an analog electric signal (for example, three colors of white, yellow, and cyan) obtained by separating an input signal from the color original input device 90 is converted into a digital signal by the A / D converter 91. On the other hand, it is supplied to a circuit 93 for correcting the ink density characteristic of the color printer 92 and the input signal characteristic of the color original input device 90. Next, a matrix circuit 94 for color conversion is passed through a signal obtained by reading the original so as to match the spectral characteristics of the ink. This circuit 94 is composed of, for example, a product-sum calculation circuit as shown in FIG. 10, and performs the following calculation.

M = a11w + a12y '+ a13c' Y = a21w + a22y '+ a23c' C = a31w + a22y '+ a33c' where w, y ', and c'correspond to the signal from the input device 90, respectively to white, yellow, and cyan. Correspond. Further, M, Y and C are color conversion matrix circuits 94.
Are converted into signals according to the color development characteristics of the color printer 92 and correspond to magenta, yellow and cyan, respectively. a11 to a33 are matrix coefficients for conversion.

The operation of this circuit will be described with reference to FIG. The gate circuit 101 produces the w signal at terminal 102. Next, the product-sum operation circuit 103 extracts the coefficient of a11 from the parameter memory 104 for this signal, calculates the product, and stores the product in the internal accumulator. Next, the gate circuit 10
By generating the y'signal at the terminal 102 by 1 and performing the same operation as described above, the coefficient a12 is multiplied,
Add by the internal accumulator. Further, the gate circuit 101 selects c ′, and the coefficient a
Multiply by 13 and add with the internal accumulator. If this result is output, M can be obtained. Y and C are calculated in the same manner. In this way, it is possible to convert the signal into a signal that matches the color development characteristics of the color printer 92.

Next, as shown in FIG. 9, the converted signals M, Y, and C are output by selecting the simple binarization process or the dithering process as described in the previous embodiment. Input to the color printer 92.

This processing circuit is a circuit 95 surrounded by a broken line.
The configuration is similar to that of the previous embodiment. This circuit 95
Are independently provided for M, Y, and C, respectively.
In this circuit, the parameters can be selected independently for each color, so that delicate color adjustment is possible.

For example, by arranging the contents of the reference memory for dithering so that the positions of the starting points of the respective colors are different, it is possible to reduce the color turbidity.
It is also possible to make the degree of high-frequency emphasis different depending on the type of color, which makes it possible to clearly display a specific color.

By changing the parameters for performing the dithering process, it is possible to change the density gradient for each color, and the parameters are set according to the characteristics of the ink used in the printer to obtain a more natural color. It is possible to realize the reproduction of. When the contrast of the original is poor or when the white background has noise, the density conversion RO of FIG. 1 used in the previous embodiment.
It is preferable to use the circuit 96 having the functions of M6, 7, and 8.

Next, another embodiment of the color image processing apparatus will be described. In FIG. 11, as in FIG. 9, w, y ′, and c ′ separated for each color of white, yellow, and cyan are input by the color original input device 90, and converted into a digital amount by the A / D converter 91. To be done.

Next, the logarithmic conversion circuit 110 reduces the number of representation bits. Specifically, a conversion ROM having the characteristics shown in FIG. 12 is prepared, and the ROM data is addressed to perform logarithmic conversion. However, for example, in the case of converting from 6 bits to 4 bits, it is possible to obtain better characteristics by using the dead region as shown by the following equation than by simply performing logarithmic conversion.

When x> a y = 15 · lnx / (ln64-lna) −15 · lna / (ln64-lna) (2) When x <a or x = a y = 0 where x is the input signal And y is the output signal. Further, a is the upper limit value of the dead zone, and about 2 to 3 is appropriate.

The reason for reducing the number of expression bits here is as follows. First, it is necessary to match the gradation of the color printer 92 capable of output display. Further, the calculation with the color conversion matrix 94 found in the previous embodiment may not provide good color reproduction depending on the characteristics of the ink.

In such a case, in particular, all combinations of colors are calculated in advance to create a table, and the color is calculated by drawing this table. By doing so, even extremely complicated calculations can be processed in real time. Further, since it is sufficient that the table created here corresponds to the combination of colors that can be expressed by the color printer, the number of expression bits can be reduced for these reasons.

As described above, the color conversion is performed by looking up the color conversion table 111. Here, the table 111 is composed of a RAM, and the contents of this table are stored in the ROM.
By inputting from 112, free conversion is possible. For example, when the contrast of the document is low, or the background of the document is too dark, the contents of this table can be exchanged to make an intelligent printer or copy.

As described above, in the color expression of a four-color printer or the like in which the three colors are expressed in color and black is added to the lack of the density level, if the added black is blurred by the dither method, it becomes very unsightly. Therefore, if only the line component of the black signal is detected and displayed, such unpleasantness is eliminated.
Furthermore, when this black is displayed with high resolution, there is no unnaturalness even if the resolution with other colors decreases, and it is also possible to improve the fidelity of hues, so that it is natural to the human eye. It has the effect of being felt as an image. (Third Embodiment) Next, as shown in FIG. 11, a selection signal for selecting a signal processed by simple binarization processing or dither processing is not obtained independently for each color, but A / It is also possible to generate the selection signal using the D-converted output signal (the portion surrounded by the broken line in FIG. 11).

Now, when the convolution operation is performed in the portion surrounded by the broken line in FIG. 11, the following procedure is required.

That is, in the embodiment, the line memory section 10 is provided with a line memory of 5 lines or more.
In order to process an image signal included in one line at the center of the lines, it is necessary to capture at least two lines of image signals for each of the front and rear. This means that in order to perform image processing for one line, the image signal delayed by two lines is waited for and processed.

Therefore, in the case of performing a 5 × 5 operation, first, the data for two rows is sent to the convolution operation section,
Next, try to actually calculate. Then, the deviation between the selection signal processed in the part surrounded by the broken line in FIG. 11 and the color conversion output signal of the color conversion table 111 is two lines. Therefore, the delay circuit 113 delays the output of magenta, yellow, and cyan by two lines, respectively, and synchronizes with the selection signal for image processing.

Next, simple binarization processing and dithering processing are performed for each color. That is, for each of the output signals M, Y, and C from the color conversion table, the simple binarization circuit 1
The binarization process is performed at 14, and the dither process is performed by the comparator 115 while sequentially comparing with the dither pattern from the dither pattern reference memory 116.

Each of these signals is input to the multiplexer 117 for selecting simple binarization / dithering, and is switched by the output signal of the switching code conversion ROM 13. When all the outputs of the color multiplexers 117 are 0 or 1, white or black is printed.
This is detected by the circuit 118. At this time, the black and white signal gate circuit 119 is opened, and at the same time, the color signal gate circuits 120, 121, 122 are closed.

The signal thus obtained can be input to a color printer to obtain an output image including color halftones.

In this method, there is a possibility that the degree of freedom such as high-frequency emphasis independent of each color is limited, but for white and black, a sharp image with high-frequency emphasis is obtained. Also for a color image, high resolution display by binarizing the line component and halftone expression by dither are possible. Further, it is sufficient to provide at least one convolution circuit having a somewhat complicated configuration, so that the circuit scale can be reduced and the system can be downsized.
Further, when performing color conversion, it is only necessary to subtract the corresponding data from the table, and therefore the circuit can be further simplified.

FIG. 13 is a circuit obtained by further simplifying the circuit of the embodiment shown in FIG. This circuit is characterized in that the convolution circuit 10 and the decision ROM 13 are provided with a portion for generating a selection signal for selecting simple binarization processing or dithering processing. And the color conversion table 11
1 is white, yellow, cyan to magenta (M),
It is a circuit that converts yellow (Y), cyan (C), and black (B). For each of the magenta, yellow, cyan, and black color components, a processing signal by simple binarization or dithering is selected as in the previous embodiment. By doing so, it becomes possible to further simplify the circuit.

Depending on how the ink is used, black processing may be performed. At this time, the black processing may be omitted in FIG.

In the above embodiment, even if the local fluctuation rate is large to some extent, as long as it does not exceed the threshold for simple binarization,
No white-to-black or black-to-white inversion occurs. However, for the human eye, it may be possible to improve the image quality by providing the portion where the above-mentioned color inversion occurs with sharpness. In order to perform such processing, the circuit as shown in FIG. 14 may be used by changing the output signal selection determination circuit of the simple binarization processing and dithering processing of the previous embodiment.

In this circuit, the convolution circuit 1
0 and the decision ROM 13 outputs the decision signal of the output signal of the simple binarization process and the dither process, but here, the following decision is made.

First, when the absolute value of the local fluctuation rate is sufficiently large (when it is larger than a certain first threshold value), that is, when the local fluctuation rate is sufficiently large or sufficiently small,
If the value is large, it is determined to be black, and if it is small, it is determined to be white. Next, when the absolute value of the local fluctuation rate is large to some extent (smaller than the first threshold value and larger than the second threshold value), that is, when the local fluctuation rate is large to some extent or small to some extent. Determines to select the output signal by the simple binarization process. Further, when the absolute value of the local fluctuation rate is smaller than a certain value (smaller than the above second threshold value), it is determined to select the output signal by the dithering process.

By doing so, if there is a large local variation, the inversion of the density will always occur and a sharp image can be obtained. However, if the input image has noise, it may be unsightly.

[0067]

As described above, according to the present invention, it is possible to obtain a local variation rate in which the influence of the variation in the element characteristics of the line sensor is reduced, and the image can be displayed with high resolution and halftone. Can be expressed.

[Brief description of drawings]

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

FIG. 2 is a diagram for explaining the contents of a density conversion ROM.

FIG. 3 is a diagram showing a matrix of convolution operation.

FIG. 4 is a diagram showing a line memory unit.

FIG. 5 is a diagram showing a product-sum calculation unit.

FIG. 6 is a diagram illustrating signal processing of the present invention.

FIG. 7 is a diagram illustrating an example of contents of a switching code conversion ROM.

FIG. 8 is a diagram illustrating conventional signal processing.

FIG. 9 is a diagram showing another embodiment of the present invention.

FIG. 10 is a diagram showing a circuit for performing a color conversion matrix operation.

FIG. 11 is a diagram showing another embodiment of the present invention.

FIG. 12 is a diagram for explaining bit reduction conversion.

FIG. 13 is a diagram showing another embodiment of the present invention.

FIG. 14 is a diagram showing a modification of the simple binary / dither switching circuit.

[Explanation of symbols]

 10 ... Convolution circuit 13 ... Judgment ROM 14 ... Multiplexer 22 Printer

Continuation of front page (51) Int.Cl. ⁶ Identification number Internal reference number FI Technical display location G06F 15/31 A (72) Inventor Shuzo Miura 1 Komukai Toshiba-cho, Kouki-ku, Kawasaki-shi, Kanagawa Toshiba Corp. (56) References JP-A-57-78275 (JP, A) JP-A-57-95759 (JP, A)

Claims (1)

[Claims] 1.One-dimensional line for inputting image signal
A sensor, Stores image signals for multiple lines from this line sensor
A two-dimensional image signal storage means capable of The unit data of the image signal stored in the image signal storage means
Data and the coefficient corresponding to this unit data
Means for obtaining a local variation rate of the input image signal
When, Depending on the calculated local fluctuation rate, the input image
And means for performing different signal processing on the image signal, Each corresponding to the unit data for obtaining the local variation rate
The coefficient is the coefficient corresponding to the unit data string in the same line direction
Relative to multiple coefficient lines in a column
Coefficient included in the coefficient string corresponding to the unit data string in the direction
It is set so that the fluctuation of the value becomes larger.
is there An image processing device characterized by the above.