JPH079672B2 - Image signal processor - Google Patents

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image signal processing device having a function of reproducing multi-valued image information including a gradation image.

2. Description of the Related Art In recent years, along with the mechanization of office work and the rapid spread of image communication, there is an increasing demand for reproduction of high quality in gradation images and printed images, in addition to conventional black and white binary documents.

In particular, many proposals have been made for the pseudo gradation reproduction by using a binary image of a gradation image because it has good compatibility with a display device and a recording device.

The dither method is best known as one means for reproducing these pseudo gradations. This method attempts to reproduce gradation in a predetermined fixed area by the number of dots reproduced in that area. While comparing the threshold value prepared in the dither matrix with the input image information for each pixel. Binarization processing is performed. In this method, gradation characteristics and resolution are directly dependent on the size of the dither matrix, and are incompatible with each other. In addition, it is difficult to avoid the occurrence of moire patterns in reproduced images used for printed images.

Therefore, as a method that achieves both the above gradation characteristics and high resolution and has a large effect of suppressing the generation of moire patterns, the error diffusion method (reference: `` Adaptive algorithm for spatial gray scale '', Floyd and Steinberg (R.FLOYDO & L. STEINBERG, “An Adaptive Algorithmfor
Spatial Gray Scale ", SDI 75 DIGEST, pp36-37))
Is proposed.

FIG. 3 is a block diagram of an essential part of an apparatus for realizing the above error diffusion method.

When the coordinates of the pixel of interest in the original image are (x, y), 3
01 is an error storage means for storing an error, 302 is an unprocessed pixel area around the pixel of interest indicated by the error distribution count matrix,
303 is the storage position of the accumulated error Sxy at the coordinates (x, y), 304
Is an input terminal of level Ixy at coordinates (x, y), 305 is an input correction means for correcting the input level and outputting it as I'xy (= Ixy + Sxy), 306 is a binary signal Pxy of level 0 or R
307 is a signal terminal for applying a constant threshold value R / 2, 308 is a signal terminal for comparing the input signal I′xy with a constant threshold value R / 2, and Pxy = R when I′xy> R / 2, Otherwise, Pxy = 0
Numeral 309 is a binarization means for outputting the difference, and 309 is a difference calculation means for obtaining a binarization error for the pixel of interest and outputting it as Exy = (= I'xy-Pxy).

Now, the integration error Sxy for the pixel of interest is the following (1),
It is expressed by equation (2).

Sxy = ΣKij · Ex−j + 2, y−j + 1 (1) (where i and j represent coordinates in the error distribution coefficient matrix) This error distribution coefficient Kij is the distribution of the error Exy to the peripheral pixels of the target pixel. The weighting is done in the above document. (However, * indicates the position of the pixel of interest).

In the configuration of FIG. 3, in the above calculation, the binarization error Exy for the pixel of interest is multiplied by the distribution coefficient corresponding to each pixel A to D in the unprocessed peripheral pixel region 302, and the error storage means 30 is obtained.
This is realized by an error distribution calculation means 310 that adds the value within 1 and stores it again in the corresponding position. However, the error storage means 3
The integration error at the pixel position B of 01 has been cleared to O in advance.

Problems to be Solved by the Invention The error diffusion method described above has excellent performance in terms of gradation characteristics and resolution as compared with the dither method, and the appearance of moire patterns is extremely small even when reproducing a printed image. In principle, this is a method in which the gradation of black pixel (or white pixel) density according to all the input levels Ixy can be reproduced in a pseudo manner. On the other hand, by utilizing these characteristics, it is possible to easily realize multi-valued output for a device capable of recording or displaying about several gradations, and to realize an image signal processing device that can obtain smoother gradation characteristics.

By the way, if the processing method is to be realized by a practical integer arithmetic processing circuit, the multilevel error E (n) xy and the sum ΣKij · E (n) xy of the error distribution values to the peripheral pixels do not always match. do not do. This means that all the values of the multi-value quantization error E (n) xy are not distributed to the peripheral pixels, and the gradation corresponding to all the levels of the input level Ixy cannot be reproduced. When the density is low and the density is high, this phenomenon is remarkable and there is a problem that a reproduced image in which the gradation reproduction area is narrowed is obtained.

SUMMARY OF THE INVENTION It is an object of the present invention to provide an image signal processing device which has improved tone reproduction characteristics in implementing the above n-valued error diffusion method and has an extremely small moire pattern.

Means for Solving the Problems The present invention is directed to an error storage unit that stores a multi-valued error of a target pixel in association with pixel positions around the target pixel, an input level of the target pixel, and an attention level in the error storage unit. Input correction means for adding an integration error corresponding to a pixel and outputting a correction level, and inputting upper n bits of the correction level and outputting the same as multi-valued data, and a multi-valued level corresponding to the multi-valued data A multi-valued output selection means for selectively outputting, an error calculation means for obtaining a multi-valued error that is a difference between the correction level and the multi-valued level, and an unprocessed pixel around the target pixel for the multi-valued error. The distribution count generating means for generating a distribution count to be distributed to, the multi-valued error from the error calculating means and the distribution count from the distribution count generating means to calculate an error distribution value corresponding to an unprocessed pixel around the pixel of interest. Calculate The error distribution means and the sum of the error distribution values corresponding to the unprocessed pixels around the pixel of interest from the error distribution means are obtained, and the residual error which is the difference from the multilevel error from the error calculation means is calculated and output. A residual error calculating means for adding one of the error distribution values and the residual error from the residual error calculating means,
An error updating means for adding the remaining error distribution value and the integrated error of the corresponding pixel position in the error storage means and storing again.

According to the present invention, with the above configuration, the upper n bits of the input correction level are used as the multi-valued output data, and the multi-valued level corresponding to the multi-valued output data is selected and output to obtain the multi-valued error. And the sum of the error distribution values corresponding to the peripheral pixels of the pixel of interest are corrected at the time of the next pixel processing by a residual error that is a difference between the multi-valued error and the sum of the error distribution values of the peripheral pixels. The tone reproduction characteristics in the low density and high density areas are improved so that moire patterns do not occur.

Embodiment 1 FIG. 1 is a block diagram of essential parts of an image signal processing apparatus in an embodiment of the present invention.

In the figure, error value storage means 101, peripheral pixel area 102, target pixel error storage position 103, input terminal 104, input correction means 10
5, the binarizing means 108 is the error storing means 301, the unprocessed pixel area 302 around the target pixel, the storage position 303, the input terminal 304, the input correcting means 305, and the binarizing means 3 in FIG. 3, respectively.
Similar to 08.

The multi-valued output selection means 107 inputs the upper n bits including the sign of the input correction level I ′ (x, y) (N-bit signal including sign bit) corrected by the input correction means 105,
The multi-valued output data P (n) (x, y) is output to the output terminal 106, and the multi-valued level corresponding to the multi-valued output data is selected and output to the error calculating means 108.

The error calculation means 108 calculates the difference between the input correction level I ′ (x, y) and the selected output level, and the error distribution means 109.
Output to. Therefore, the multi-value quantization error E (n) (x, y) when selected at a certain multi-value level is E (n) (x, y) = I '
It is calculated as (x, y) -P (n) (x, y).

Next, the multi-value output selection means 107, the error distribution means 109, the error update means 111, and the distribution coefficient generation means 110 will be described with reference to FIG.

In the embodiment shown in FIG. 2 (a), multi-value output selection means
A case where the output of 107 is a quaternary output will be described.

Now, assuming that the input image signal I (x, y) is 8-bit data, the input correction level I ′ (x, y) corrected by the input correction means 105 has an error calculation value of ±, so the sign is It is 10-bit data including. Input correction level I '(x,
y) (0: 9) upper 4 bits I '(x, y) (6: 9) are input to the quaternarization determination circuit 221, and the quaternarized data shown in Table 1 is output unit of FIG. Output to A and B, and output level P0 to P respectively
Corresponds to 3.

The quaternarization determination circuit 201 has a positive sign bit I′9 (x, y) and a hexadecimal display data in the range of “00” to “1FF”, that is, “000” ≦ I ′ (x, y) ≦ “. In case of "OFF", I'6
The data of (x, y) and I'7 (x, y) are directly used as 4-value output data. However, when the sign bit I′9 (x, y) is positive and I ′ (x, y)> “OFF”, the four-value output data is forcibly set to “11”, and the sign bit I′9 If (x, y) is negative, that is, if I ′ (x, y) <“000”, then 4
The value output data is forcibly set to "00". The above is summarized in Table 1.

Now, the decoder 222 inputs the four-valued data output from the four-valued determination circuit 201 and outputs the four-valued output level corresponding to the four-valued output data of the four-valued determination circuit 221 to the error calculation means 108. . If the 4-level output level of the decoder 222 is equally divided, P0 = “00”, P1 = “7F”, P2 = “80”, P
Set to 3 = "FF" (all hex display). The error calculating means 108 calculates the difference between the input correction level I ′ (x, y) (0: 9) and the output level P (n) (x, y) selected and output by the decoder 222, and quaternarized. The error E (n) (x, y) is output to the error distribution unit 109.

The adaptation to the ternary recording system can be easily realized from the quaternary output data shown in Table 1 by pulse width modulation. Second
FIG. 2D is an example of a simple pulse width modulation, and FIG. 2E is a timing chart thereof.

In the figure, 401 is a four-valued signal input terminal for inputting a four-valued signal output to the output terminal of 106, and 402 is a pulse generator for generating and outputting a 1 / 2-delayed pulse in synchronization with a synchronization signal. , 403 is a selector for selecting and outputting a ternary pulse width signal by a quaternary signal, and 404 is a latch for temporarily storing the quaternary input signal. Next, the operation will be described. A maximum pulse width is set for one cycle of the synchronizing signal, then a signal having a pulse width of 1/2 of the maximum pulse width and 0 is prepared in the selector 403, and is output from the latch 404 by the synchronizing signal. It is intended to output a ternary pulse width modulation signal by switching the signal according to the binarized signal. Therefore, a pulse width of 1/2 can be obtained by delaying the sync signal by 1/2. When the four-valued signals A and B output by the synchronizing signal are "0" and "0", respectively, the pulse width 0 of the selector 403 is selected, and the four-valued signals A and B are "0" and "0", respectively. 1 "or" 1 ",
When it is "0", select the signal that is 1/2 delayed from the sync signal of the selector 403, and when the four-valued signals A and B are "1" and "1", select the pulse width 1 of the selector 403. Output. Thus, the ternary pulse width modulation signal can be output.

The distribution coefficient generating means 110 prepares a distribution coefficient set for unprocessed pixels around the target pixel in advance,
Value E (n) (x,
The y) distribution coefficients K _{A to} K _D are output to the error distribution unit 109.

The error distribution unit 109 synchronizes with a synchronization signal synchronized with the periphery of the pixel processing, and multi-valued error E (n) (x, for the distribution coefficient K _{A to} K _D and the pixel of interest from the error calculation unit 108.
y) and error distribution values G _{A to} G _D corresponding to the pixel positions A, B, C, and D in the peripheral pixel area 102 of the error storage means 101 are calculated by the equation (3).

Further, the error distribution values G _{A to} G _D are output to the error updating unit 111 and the residual error calculating unit 112.

The remainder error calculation means 112 calculates the sum of the error distribution values G _{A to} G _D and the multi-valued error E (n) from the error distribution values G _{A to} G _D and the multi-valued error E (n) (x, y). ) (X, y) The residual error J _B , which is the difference from (x, y), is calculated by the equation (4).

J _B = E (n) (x, y)-(G _A + G _B + G _C + G _D ) …………
(4) Further, the residual error J _B is output to the error updating means 111 described later.

The error updating means 111, in synchronism with the periodic signal, surrounds the error storage means 101 and the residual error J _B from the residual error calculation means 112 of the error distribution prices G _{A to} G _D from the error distribution means 109. The integration error S _A ′, S _C ′, S _D ′ in the previous pixel processing process stored in the storage device corresponding to the pixel positions A, B, C and D in the pixel area 102 is read out and a new integration is performed. The errors S _{A to} S _D are calculated by the equation (5).

Further, the error updating means 111 performs an updating process of writing the new integrated errors S _{A to} S _D in the storage devices corresponding to the pixel positions A to _D of the error storage means 101.

However, in the equation (5), the residual error J _B is calculated as the peripheral pixel area 102
Pixel position A, B, C, D
May be added to any position of
Will be described as an addition.

Further, the residual error J _B may be temporarily stored in an internal register, read at the time of the next pixel processing, and added to any one of the pixel positions A ′, B ′, C ′ and D ′ in the peripheral pixel region 102. .

Specific examples of the error distribution means 109, the distribution coefficient generation means 110, the error update means 111, and the residual error calculation means 112 are shown in FIG. 2 (b). In the figure, the residual error J _B is temporarily stored in an internal register and reflected in the peripheral pixel position B.

The distribution coefficient generation means 205 is provided with a storage device 206 for storing the distribution coefficients K _A , K _B , K _C , and K _D in advance, and stores them before starting the pixel processing. Alternatively, the storage device 206 stores the distribution coefficient K
_A ~K _D may be used in advance the written ROM (read only memory).

The error distribution means 207 obtains by multiplying the error distribution values G _{A to} G _D from the multi-valued error E (n) (x, y) and the distribution coefficients K _{A to} K _D, and the error updating means 210 and the remainder. It outputs to the error calculation means 208.

The residual error calculation means 208 is the sum 209 of the error distribution values G _{A to} G _D from the error distribution means 207 and the multi-valued error E (n).
The residual error J _B , which is the difference from (x, y), is calculated and temporarily stored in the internal register 203 (R _J ). Internal register 203 (R _J )
The residual error J _B ′ at the time of the previous pixel processing read from is output to the error updating means 210.

The error updating means 210 synchronizes with the synchronization signal 214 input from the synchronization signal input terminal 204 in synchronization with the pixel processing, and the error distribution value G _A and the pixel position A read from the error storage means 201.
Internal register 211 for adding the integrated circuit S ′ _A corresponding to
Temporarily store in (R _A ). The integrated error for the pixel position B is added to the error distribution value G _B and the residual error J _B ′ temporarily stored in the previous pixel processing from the residual error calculation means 208, and the integrated error corresponding to the pixel position B (S _B ) Is temporarily stored in the internal register 212 (R _B ). The error distribution value G _C is added to the data of the internal register 212 (R _B ) which is temporarily stored in the previous pixel processing, and is temporarily stored in the internal register 213 (R _C ) as the integrated error (S _C ) of the pixel position C. . Error allocation value G _D
Is an internal register 21 that is temporarily stored in the previous pixel processing.
3 (R _C ) data is added and the integration error (S
_D ) is stored in the storage device corresponding to the pixel position D of the error storage means 201.

With such an error updating means 210, the access to the storage device in the error storage means 201 is only a read access corresponding to the pixel position A and a read access corresponding to the pixel position D, which is easily realized.

Further, FIG. 2 (c) shows a distribution coefficient for distributing the binarization error Exy to unprocessed pixels around the pixel of interest. A specific example of the residual error calculating means 208 in the case of
In the figure, the multi-valued error E ′ (n) (x, y), which is the lower 3 bits of the multi-valued error E (n) (x, y), is
The two-bit residual error J _B is input to the decoder 216 having the relationship shown in the table. The residual error J _B is temporarily stored in the internal register 217 (R _J ) and the read residual error J _B ′ is output during the next pixel processing.

As described above, by setting the distribution coefficient as shown in the equation (6), complicated calculation becomes unnecessary and a practical circuit configuration is obtained.

EFFECTS OF THE INVENTION As described above, in the image signal processing device according to the present invention, the distribution rate of the multi-valued error with respect to the peripheral pixels of the pixel of interest is not fixed, and a plurality of distribution coefficients in the distribution coefficient set are sequentially changed along with the pixel processing. By using this, it is possible to significantly suppress the false image (texture) seen in the conventional error diffusion method, and by using the upper n bits of the input correction level I ′ (x, y). By using multi-value output data, selecting a multi-value output level according to the multi-value output data, and performing multi-value error calculation processing, multi-value output can be easily realized, and It is possible to easily obtain an output image that matches the gradation characteristics. Further, in the present invention, when the multi-valued error is distributed to the peripheral pixels of the pixel of interest, the residual error is calculated from the sum of the error distribution values to which the multi-valued error is distributed, and the residual error is corrected at the time of the processing of the next pixel. It is possible to provide a practical processing circuit in which the gradation reproduction characteristics in the low-density and high-density areas of the input level, which was a problem when the method is configured with a practical integer arithmetic processing circuit, are significantly improved. It will be possible.

[Brief description of drawings]

FIG. 1 is a block diagram of an essential part of an image signal processing apparatus according to an embodiment of the present invention, FIG. 2 (a) is a block diagram of multi-valued output selecting means of the apparatus, and FIG. 2 (b) is the apparatus. Wiring diagram of error distribution means, distribution coefficient generation means, and error updating means,
FIG. 2C is a block diagram of the residual error calculating means of the device, FIG. 2D is a block diagram of a main part for performing pulse modulation in the device, and FIG. 2E is a diagram of FIG. 2D. FIG. 3 is a waveform diagram of essential parts, and FIG. 3 is a block diagram of essential parts of an image signal processing apparatus for implementing the conventional error diffusion method. 101, 102 ... Error storage means, 105 ... Input correction means, 107
...... Multi-value output selection means, 108 …… Error calculation means, 109 ・ ・ ・
... error distribution means, 110 ... distribution coefficient generation means, 111 ... error updating means, 112 ... remainder error calculation means, 221 ... four-value determination circuit, 222 ... decoder, 206 ... storage device, 217 ...
…decoder.

Claims (2)

[Claims] 1. An error storage means for storing a multi-valued error of a target pixel in association with pixel positions around the target pixel, and an input level of the target pixel and an integrated error corresponding to the target pixel in the error storage means. Input correction means for adding and outputting a correction level, and multi-value conversion for inputting the upper n bits of the correction level and outputting it as multi-valued data, and selectively outputting the multi-valued level corresponding to the multi-valued data Output selecting means, error calculating means for obtaining a multilevel error that is the difference between the correction level and the multilevel level, and a distribution count for distributing the multilevel error to unprocessed pixels around the pixel of interest. Distribution count generating means, an error distribution means for calculating an error distribution value corresponding to an unprocessed pixel around the pixel of interest from the multi-valued error from the error calculating means and the distribution count from the distribution count generating means, The error A residual error calculating means for calculating a sum of error distribution values corresponding to unprocessed pixels around the pixel of interest from the dividing means, calculating and outputting a residual error which is a difference from the multilevel error from the error calculating means, An error in which one of the error distribution values and the residual error from the residual error calculation means are added, and the remaining error distribution value is added together with the integrated error of the corresponding pixel position in the error storage means and is stored again. An image signal processing device including an updating unit. 2. A residual error calculating means obtains a sum of error distribution values corresponding to unprocessed pixels around a target pixel from the error distributing means, and calculates a residual error which is a difference from a multi-valued error from the error calculating means. The image signal processing apparatus according to claim 1, wherein the image signal processing apparatus calculates, temporarily stores, and reads out and outputs at the time of processing the next pixel.