JPH03109871A - Picture processor - Google Patents

Abstract

PURPOSE:To attain effective binarization from a picture end part as the processing result entirely by employing an arithmetic means obtaining an average density for a prescribed area through the use of a binarized multi-value data near a noted picture element. CONSTITUTION:This processor is provided with an input means receiving a data of a noted picture element, a 1st arithmetic means 1 using a binarized data to obtain an average density for a prescribed area binarized already before the noted picture element, and a 2nd arithmetic means 10 using the binarized multi-value data near the noted picture element to obtain the average density for a prescribed area. Then a binarizing means 3 binarizing the data of the noted picture element based on a 2nd arithmetic means output, correction means 4, 5, 20, 21 correction an error generated at binarization of the noted picture element and data storage means 19, 20 storing the binarizing means output or the binarizing means output and the correction means output for a prescribed period are provided. Thus, even when the binarized binary data is not already in existence, the multi-value data of the same position is used to obtain an average density with weight and effective binarization is attained from the edge of a picture.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an image processing device that performs binarization processing on image data.

[Prior Art] Conventionally, error diffusion methods, average density approximation methods, and the like have been proposed as pseudo intermediate processing methods for image processing devices such as facsimile machines and digital copiers.

The former error diffusion method is described in the literature R, FLOYD & L, 5T.
EINBERG, ”AN ADAPTIVE ALG
ORITHM FOR3PATIAL GRAY 5
CALE”, SID 75 DIGEST PP36~
As disclosed in No. 37, the multi-valued image data of the pixel of interest is binarized (converted to the darkest level or the shortest level), and the error between this binarized level and the multi-valued image data before binarization is calculated. is given a predetermined weight and added to pixel data in the vicinity of the pixel of interest.

Moreover, the latter average concentration approximation method is disclosed in JP-A-57-1043.
As described in No. 96, the pixel of interest is converted into black or white using the already binarized binary data near the pixel of interest.
When converted into values, weighting with each neighboring pixel is performed by obtaining an average value, and using the average of these two average values as a threshold value, the image data of the pixel of interest is binarized.

[Problems to be Solved by the Invention] The error diffusion method described above is a method for correcting errors between human image data and output image data, so it is possible to preserve the density of the input image and the output image processing device. It is possible to provide images with excellent resolution and gradation.

However, when correcting errors between input image data and output image data, many two-dimensional calculations must be performed, and the large amount of processing requires a very complex hardware configuration. .

On the other hand, since the average density approximation method performs calculations using binary data after binarization, it is possible to simplify the hardware configuration and achieve faster processing due to the extremely small amount of processing. is possible.

However, since binarization is performed by simply approximating the pixel of interest to the average value of the area including the pixel of interest, the number of gradations is limited and the low frequency characteristic of images with gradual density changes is limited. There were drawbacks such as texture generation and deterioration of image quality.

What is common to these methods is that the original image is
When divided into two or more regions, the data between each divided image after processing becomes discontinuous, resulting in deterioration of image quality.

That is, the error diffusion method has the drawback that the error for correction is not propagated near the binarization start point of the divided image, and the binarization process is performed using a so-called constant value.

On the other hand, in the average concentration approximation method, when calculating the average concentration,
This is due to the fact that already binarized data is not carried over.
There is a drawback that the halftone expression becomes discontinuous at the joint between each divided image after the binarization process, resulting in a decrease in image quality.

[Means for Solving the Problems] The present invention has been made for the purpose of solving the above-mentioned problems, and it is possible to significantly reduce the amount of processing for quantization, and also to achieve good halftones. The present invention provides an image processing device that can reproduce images.

This embodiment has the following configuration as a means for achieving the above-mentioned objective.

That is, an input means for inputting the data of the pixel of interest, a first calculation means for calculating the average density value of a predetermined area that has already been binarized before the pixel of interest using the binarized data, and a second calculation means for calculating an average density value of a predetermined area using binarized multivalued data near the pixel;
A binarization means for binarizing the data of the pixel of interest based on the output of the arithmetic means, a correction means for correcting an error that occurs when the pixel of interest is binarized, and an output of the binarization means or an output of the binarization means. and data storage means for storing the output of the correction means for a predetermined period of time.

[Operation] In the above configuration, even when there is no already binarized binary data to refer to when binarizing the edge of the image, the multi-valued data at the same position is used to convert the binary data isometrically. It is possible to obtain a weighted average density based on the image density, and it is possible to convert the image into an effective 2 (i) from the edge of the image.

Also, when performing serial scanning, the binary data at the end of each scan is saved until the next scan is executed, and is referenced when executing the subsequent scan to calculate the weighted average density. As a result, data does not become discontinuous at the connection portion of each scanning amount, and an image with excellent gradation and resolution can be obtained.

[Example] Hereinafter, an example according to the present invention will be described in detail with reference to the drawings.

[First Embodiment] First, the principle of an embodiment according to the present invention will be explained with reference to FIG.

FIG. 1(A) is a diagram showing the multivalue density of each pixel of a human image.

In FIG. 1(A), f(tj) indicates multilevel density data of the input image at the pixel position of interest to be binarized, and is a normalized value of O to 1. In addition, the pixel positions above the broken line have already been binarized, and after the pixel of interest is binarized, the same binarization process is performed sequentially as f (i, j+1), f (i, j+2), etc. It will be done.

FIG. 1(B) is a diagram representing binarized image data, and B
(i, j) indicates the density of the pixel of interest after binarization (assumed to be a value of “0” or 1°). The portion surrounded by the broken line is pixel data that has already been binarized when processing the pixel of interest, and these are used when the pixel of interest is binarized.

FIG. 1(C) is a diagram showing a weighted mask. R is an example of a weighted mask for determining the average density, and is represented by a 3×3 matrix.

Here, the weight n (o, o)R(0
, -1)=0.

In this example, the weighted average density of the binary image in the vicinity of the pixel of interest is m(i, j), and the pixel of interest f(i, j) is expressed by the following formula %. ) and the already assigned binarization correction values E(i, , Ll) are binarized according to the following series of equations.

B when f (i, j) + E (i, j) > m (i, j)
(i, j)=1err(i, j)=f(i, j)+
E(i, j)-m(i+j)f(i, j)+E(
i, j) When 5m(i, j), B(i, j)=Oer
r(i,j)・f(i,j)+E(i,j)−m(i,
j) However, f(z,,+)+E(i,j) = m(z
, j) = 1, then B (f, j) = 1 E (i, 1 = E1 (i, j) + E2 (i, j) El (i
, j+1) =E2(i+l,j) =err(i
, j)/2Figure 2 (A), (E)? This is a diagram showing the series of 0 equations described above.

In formula ■, H(i, j) is 1 of the pixel of interest (i, j)
The multilevel density f (i,
The error that occurs when binarizing j-1) into a binary density B (i, j-1), that is, the multi-value density f (i, j-i
) and its neighborhood average density m(t, j-1), and the multilevel density f of the pixel (i-1, j) one line before the pixel of interest (i, j), that is, the pixel (i-1, j). Error that occurs when (i-1,j) is binarized into binary density B, that is, multi-value density f 04. j) and its neighborhood average concentration m (i-1,
j) is the sum of the difference value and the value of .

Therefore, this binarization error E(i, j) is expressed as the pixel of interest f (
By binarizing the corrected values in addition to i, j),
The image density after binarization can be completely preserved as an average density over the entire input image.

By performing processing that takes such binarization errors into consideration, the halftone reproduction ability is significantly improved compared to the above-mentioned average density approximation method.

In addition, in equation (■), El (i, j+1) is the error distributed to the pixel (i, j+1) one pixel after the pixel of interest (i', j), and E2(i+1.j) is the error distributed to the pixel (i, j+1) one pixel after the pixel of interest (i', , j) is distributed to the pixel (i+1.j) one line after.

As described above, although the binarization method in this embodiment requires an extremely small amount of processing compared to the conventional binarization method using the error diffusion method, it has the same or better image reproduction ability. Although what is obtained is only to correct the error in two adjacent pixels, by obtaining the average density using multiple data after binarization, it is possible to correct the error in multiple pixels isometrically. This is because the same effect as distribution and correction can be obtained.

FIG. 3 is a block diagram of the image processing apparatus of this embodiment.

In FIG. 3, the human power sensor unit A is composed of a photoelectric conversion element such as a CCD and a driving device that scans the element, and reads and scans the original.The image data of the original read by the input sensor unit A is Sequentially sent to A/D converter B. The A/D converter B converts the data of each pixel into 8-bit digital data and quantizes it into data having 256 levels of gradation. Next, in the correction circuit C, shading correction and the like for correcting the CCD sensor sensitivity unevenness of the input sensor section A and the illuminance unevenness due to the illumination light source are performed by digital calculation processing.

The data that has been corrected by the correction circuit C is sent to the binarization circuit. In the binarization circuit, the 8 input from the correction circuit C
Bit multi-value image data is quantized into 1-bit binary data using the method of this embodiment described above.

Printer E is a printer constructed using a laser beam or inkjet method, and reproduces a read image on storage paper by controlling print dots on/off based on a binary data sent from a binarization circuit.

Here, if the input sensor section A is configured to perform so-called serial scanning, in which the input sensor section A reads the original in a strip shape of width ρ and length k and then reproduces the binary value, as shown in FIG.
The connection between each scan quantity (for example, scan (1) and scan (II)
) may cause a problem that the processing becomes discontinuous.

That is, the right end in the j direction of scanning (I) is f as shown in FIG.
(i, j), B (j−
2, i+1) and B (j-1, i+1) are the next scan (■
This is because the data should be binarized in step 1) and has not yet been binarized.

Therefore, in this embodiment, the multi-value data of the two pixels to be processed in the next scan (II) are read in an overlapping manner, and the read multi-value data is replaced and applied as binarized data.

On the other hand, if the pixel of interest f(i, j) is located at the left end in the j direction of scanning (1 pixel), then B(j-2, 1-
1), B(i-1, j-1), and B(i, j-1), but contrary to the above, the binary data is ) is binarized.

Therefore, in this case, the data storage means of this embodiment stores the binary data corresponding to the three pixels when they were binarized in the previous scan (I), and replaces and applies them.

Here, in FIG. 4, the above-mentioned processing is required at the upper end in the i direction for each scan and at the left end in the j direction of scan (I), but in this embodiment, the area below the top few lines is processed as an effective image. When recording as a range, the top number line is ignored.

FIG. 5 shows a detailed block configuration of the binarization section based on the above-mentioned algorithm in the binarization circuit in this embodiment.

Hereinafter, the binarization processing (binarization processing) based on the above-mentioned algorithm of this embodiment will be explained in detail with reference to FIG.

In FIG. 5, 1 is a D type flip-flop (F
/F) A delay unit 11 consisting of a FIFO buffer that delays 12 to 18 and the binarized data by 2 lines allows the binarized data to maintain the positional relationship of the 7 pixels surrounded by the broken line in Fig. 1. The above-mentioned average concentration value m
This is an average value calculation unit that outputs the following data, and is constituted by a ROM that performs data conversion based on a preset weight R. 2 operates only at the end point of binarization, and based on the multi-value data calculated by the multi-value data calculation unit 10, the weighted average density (added value) for 2 pixels is calculated from the ROM of the average value calculation unit 1. This is an adder that adds to a part of the weighted average density m based on the other binary data output.At this time, the binary data corresponding to the output from the multi-value data calculation section
The 1215 output is uniquely reset to 0° by the output of a timing generation circuit (not shown).

3 is a binarization and error calculation unit composed of a comparator and a subtracter, and the binarization result of the image is based on the multivalued data (f+E) corrected with a predetermined error (described later) and the weighted average density m. Output.

Note that this binarization result output can also be output to the delay section 11.

Error E2 (=
E, ) is output to the adder 9. The adder 9 first corrects the multivalued data f input from one side using this error E2. The corrected output from the adder 9 is input to the E2 line memory 7, where it is delayed by one line.

On the other hand, the error E output from the binarization and error calculation section 3,
(=E2) is also output to the adder 5.

One of the adders 5 receives E, which is the correction output from the adder 9.
The delayed output from the two-line memory 7, that is, the data corrected only by E2 of the i-th line of interest, is input, and the adder 5 adds the error E1 to the upper two correction data to perform further correction processing.

In this way, the adder 5 is jointly corrected by E, , E2.
The output data is delayed by one pixel by the D-type F/F 4, and then input to the binarization and error calculation section 3, where the above operation is repeated for each pixel.

Note that the multivalued data calculation section 10 can be configured with two D type F/Fs and an adder. In other words, multi-value data for one pixel belonging to the next scanning area is latched and held for each line, and the internal F/
This can be easily implemented by having a configuration in which F is shifted and held.

Further, the binarization result output from the binarization and error calculation unit 3 is also output to the binary linking memory 19. The binary data memory 19 stores only the binarized data at the end of each line out of the binarized output from the binarized and error calculation section 3 in synchronization with the one line periodic signal, and stores it in synchronization with the one line periodic signal. After being delayed, it is output to multiplexer 30. Then, it is selected by the multiplexer 30 and input to the F/F 18 and the delay section 11.

This processing operates only at the line start end in each scan after scan (II), and the average value calculation unit 1 treats the data as binary data located one pixel before the line binarization start point.

Therefore, in this embodiment, the average density m can be calculated and determined as a continuous value at the transition between each scanning angle.

As explained above, according to this embodiment, the binary data at the transition between each scan is saved until the next scan is executed, and is referred to when the subsequent scan is executed to calculate the weighted average density based on the binary data. , data is not discontinuous at the connection portion of each scanning amount, and an image with excellent gradation and resolution can be obtained.

[Second implementation example] In the first example described over a month, only the binary data was carried over and the connection processing was performed at the connection part of each scanning amount, but the present invention is not limited to the above example. , various modifications can be applied as appropriate.

For example, the error E that occurs when the binarization circuit binarizes the end pixel of each line during the previous scan (that is, the first pixel of each line in the next scan area should be corrected) as shown in Fig. By configuring to perform carryover correction processing, more accurate binarization can be achieved.

A second embodiment of the present invention in which the binarization circuit is configured as shown in FIG. 6 will be described in detail below.

9 In FIG. 6, the same components as in FIG. 5 are given the same numbers, and detailed explanations are omitted.

In the second embodiment shown in FIG.
The error E outputted from the adder 5 is not directly input to the adder 5, but is manually inputted to the adder 5 via the multiplexer 21.

At the same time, the error E1 is stored in the error connecting memory 20.
The output from the error connection memory 20 is output to one input of the multiplexer 21.

This error connection memory 20 binarizes each line in the j direction when executing the next scan, that is, the scan (I) in FIG.
This is a memory for storing E and errors to be carried over and added to the first pixel of 0 until the next scan, that is, when scanning (II) is executed. selected by the multiplexer 21 and input to the adder 5 only when the first pixel is binarized;
Only the first pixel is corrected.

By controlling in this way, the binary data at the end of each scan is saved until the next scan is executed, and is referenced when executing the subsequent scan to calculate the weighted average density for each scan. It is possible to correct even the first pixel of the scanning amount, and data does not become discontinuous at the connection portion of each scanning amount, making it possible to obtain an image with excellent gradation and resolution.

[Other Embodiments] As explained in the first embodiment, the binary data is held with a delay of one scan and the average value m is calculated when the first pixel of each line in the next scan is binarized. The transition memory 19 and the error transition memory 20 for holding E1 error data to be corrected for the same pixel during binarization of the same pixel with a one-scan delay described in the second embodiment can be written and read at the same timing. controlled. Also, both data are within 8 bits in total.

For this reason, all of the above configurations can be configured within the same chip.

Furthermore, in the first and second embodiments, FIG.
As shown in A), a 3 x 3 matrix weight mask was used, but - in order to smoothly binarize the halftone part within the film, it is necessary to set the weight of the pixel adjacent to the pixel of interest to be small (set). is desirable.

Therefore, when using the 3×5 matrix weight mask shown in FIG. 7(CB), R (i−
1, j), R(i, j-1) is 5/21 = 0.24, whereas that in Figure 7(B) is 7/48・0.1
5, it is possible to more smoothly reproduce the middle tones in binary.

Furthermore, when using the 3×5 matrix weight mask shown in FIG. 7(C), R(2,-1)=16/255,
R(2-2)=8/255, R(1,-1)=24/2
55, R (1, -2) = 16/255, and the multi-valued data used for the connection processing of each scan amount is mainly calculated by a shift operation, which can further simplify the hardware.

Furthermore, when the weighted mask becomes 3×3 or more in this embodiment, the multi-value data calculation unit 10 requires a large amount of calculation;
For example, if f has an 8-bit data length, sufficient effects can be obtained even if the upper 2 or 3 bits are used.

Furthermore, in the above description, the calculation of the average density m was simply realized using the ROM table built into the average value calculation unit 1, but this calculation can also be performed using a plurality of adders. By configuring 4 in this way, it is possible to further increase the processing speed. Furthermore, it goes without saying that the hardware scale can be significantly reduced by incorporating it inside a gate array or the like.

In the first and second embodiments described above, the error is evenly distributed to two pixels, but the error distribution is not limited to two pixels, and furthermore, the distribution rate is not limited to the even distribution. Instead, any pixel distribution can be achieved. or,
It is not limited to the average value calculation mask area or its weight.

In each of the above embodiments, there is one type of human data (one color).
The case has been explained as an example, but the input data of the present invention is 1
It is not limited to the color, and the input data is red (R),
By using three colors, green (G) and blue (B), it can also be applied to a color image processing device.

As explained above, according to this embodiment, even when there is no already binarized binary data to refer to when binarizing the edges of an image, the multi-level data at the same position is used to perform equal measurement. It is possible to obtain a weighted average density based on binary data, making it possible to perform effective binarization from the edge of the image.

Furthermore, even when performing serial scanning, the binary data at the end of each scan is saved until the next scan is executed, and the weighted average density based on the binary data is calculated by referring to it when executing the subsequent scan. Data does not become discontinuous at the connection between each scanning amount, and an image with excellent gradation and resolution can be obtained.

[Effects of the Invention] As described above, according to the present invention, effective binarization can be performed as a result of processing from the edge of the image.

Further, even when serial scanning is performed, data does not become discontinuous at the connection portion of each scanning amount, and an image with excellent gradation and resolution can be obtained.

[Brief explanation of drawings]

FIG. 1(A) is a diagram showing a multivalued image for each pixel in one embodiment of the present invention, FIG. 1(B) is a diagram showing a binarized image for each pixel in this embodiment, Figure 1 (C) is a diagram showing the weighting of each pixel in this embodiment as a mask, Figures 2 (A) and (B) are diagrams for explaining the principle of binarization processing in this embodiment, Fig. 3 is a block diagram showing the configuration of the image processing device in this embodiment, Fig. 4 is a diagram showing binary reproduction by serial scanning in this embodiment, and Fig. 5 is a diagram showing binary reproduction in this embodiment. A block diagram showing the details of the circuit, FIG. 6 is a block diagram showing the details of the binarization circuit in the second embodiment of the present invention, and FIGS. 7(A) to (C) show examples of weight masks. It is a diagram. In the figure, 1... Average value calculation unit, 2, 5.9... Adder, 3... Binarization and error calculation unit, 4, 12-18...
・Flip-flop, 7...E2 line memory, 10
...Multi-value data calculation section, 11...Delay section, 19...
Binary link memory, 20...E1 error link memory,
21.30... Multiplexer, A... Input sensor section, B... A/D conversion section, C... Correction circuit, D...
Binarization circuit, E... printer.

Claims (1)

[Claims] an input means for inputting data of the pixel of interest; a first calculation means for calculating an average density value of a predetermined area that has already been binarized before the pixel of interest using the binarized data; and a first calculation means in the vicinity of the pixel of interest. a second calculation means for calculating an average density value of a predetermined region using the multivalued data to be binarized, and a binarization method for binarizing the data of the pixel of interest based on the outputs of the first and second calculation means; means, a correction means for correcting an error that occurs when the pixel of interest is binarized, and an output of the binarization means, or
An image processing apparatus comprising: a data storage means for storing the output of the value conversion means and the output of the correction means for a predetermined period of time.