JPH0435166A - Picture processor - Google Patents

Abstract

PURPOSE:To obtain a picture with an excellent pseudo medium gradation by using a least means error method so as to binarize an input picture data subject to multi-gradation quantization and correcting the result by a Laplacian operation. CONSTITUTION:A picture input section 101 reads picture information of an original, quantizes it and the result is immediately subject to Laplacian operation and inputted to a binarizing means 102 as an original picture data. The means 102 applies binarization by using the least means error method, which reads a binarization error caused by surrounding picture elements after binarization and obtains a weight mean. The result is fed to an original picture data of the noted picture element to obtain a correction data and it is compared with a prescribed threshold level to apply binarization. A difference caused therein is stored as a binarization error. The processing above is repeated to obtain the binary data based on the least means error method and an excellent pseudo medium gradation picture is obtained.

Description

[Detailed Description of the Invention] [Summary] An image processing device that binarizes multi-tone quantized input image data using the minimum average error method to obtain a pseudo halftone image, the binarization means binarizing the pixel of interest and its surroundings. The goal is to correct the density data of the pixel of interest using Laplacian calculation using the original density data of the pixel and emphasize the sharpness of the image.Binarization involves converting the input image data that has been multi-gradation quantized in the previous stage into binarization. an image processing apparatus comprising: a control means for receiving binary output data of the binarizing means and controlling printing; and a printing means for printing the binary output data of the binarizing means; comprises a minimum average error method calculation unit and a Laplacian calculation unit, which binarizes the quantized input image data by the minimum average error method, performs a Laplacian calculation on the input image data of the pixel of interest and surrounding pixels, and calculates the value of the pixel of interest. The configuration is configured to correct density data.

[Industrial Application Field] The present invention relates to an image processing device that binarizes input image data using the minimum average error method to obtain a pseudo-halftone image, such as a pseudo-halftone image processing device used in a facsimile machine or an image scanner device. This invention relates to improvements in methods for obtaining toned images.

[Conventional technology]

A dither method is used for boat fishing as a pseudo halftone display in an image processing device. The dither method is a method of binarizing the density value of each pixel using a threshold value that changes for each pixel, and the pixels of the original image correspond to the pixels of the expression on a one-to-one basis. The systematic dither method, which is a type of dither method, includes Bayer type, halftone type, spiral type, etc. depending on the arrangement of threshold values. Furthermore, these formats are divided into dot concentrated type (spiral type) and dot dispersed type (Ba
yer type, halftone type).

Specifically, the systematic dither method considers the screen as a set of nxn dot submatrix, and determines the threshold value in the submatrix based only on the image information in the submatrix.

As mentioned above, the systematic dither method uses binary recording, so it is easy to record and has good recording stability, and because it is binary, the hardware configuration is simple, so it is relatively inexpensive and easy to use. It also has the advantage that it is possible to realize

On the other hand, as another type of dither method, %anfre
d, R, 5chroeder (Bell Labs, USA), rl
mages from C: computers J, IB
EB Spectrum, Vol. 6. pp66-
78° in 1969, and later J, F, Jarv.
is, C,N, Judice.

and W, H, N1nke (U.S. Health Research Institute) etc., rA 5urvey of Techniq
ues for the Display ofCon
Tiny Tone Pictures on
B11evel DisplaysJ.

Computer Graphics and Ima
ge Processing, Vol5, pp13
-40.There is a minimum average error method published in a paper in 1976. The minimum average error method is a method in which the error between the read density and display density of a reference pixel is weighted by the distance to the pixel of interest, and the averaged error is added to the pixel of interest and binarized using a constant threshold. That is, as will be described later, by correcting the density data of the pixel of interest using a weighted average value of binarization errors generated in surrounding pixels, the densities of the original image and the output image are made equal.

In this minimum average error method, the number of expressed gradations is not uniquely determined by the matrix size as in the systematic dither method described above, and the so-called moiré, which is a periodic striped pattern, does not occur. This is an excellent method in terms of performance and resolution.

[Problem to be solved by the invention]

By the way, the systematic dither method has the following disadvantages. In other words, ■ If the original is a halftone image such as a print, moiré that does not exist in the original will occur in the output image; ■ Due to binarization, the processing results of characters, line drawings, etc. will be cut off and the gradation will be poor; Therefore, the reproduction quality is extremely poor, and ■There is a contradictory property between increasing the matrix size to obtain multiple gradations and decreasing the matrix size to obtain high resolution. For example, it is not possible to achieve both.

On the other hand, the minimum average error method has the following disadvantages. That is, (1) Compared to simple binarization processing, the reproducibility of characters and line drawings is still inferior. That is, minute noise in the original image is also treated as an error, and granular noise is noticeable in black and white areas. (2) In areas where m degrees are low, there is a spatial delay in the appearance of dots. Furthermore, when processing a part with --like density, a unique striped pattern in which neighboring dots are connected occurs, giving an unpleasant feeling.

The purpose of the present invention is to solve these problems, and to binarize input image data that has been multilevel a-quantized using the minimum average error method, and further correct it by Laplacian operation to correct characters, line drawings, photographs, and halftones. Both parts are intended to provide an image processing device that can obtain sharp and good pseudo-halftone images.

[Means to solve the problem]

FIG. 1 is a diagram showing the principle configuration of the present invention.

The present invention comprises: a binarization means 11 for binarizing input image data that has been multi-gradation quantized in the previous stage; a control means 12 for receiving binary output data from the binarization means and controlling printing; In an image processing apparatus having a printing means 13 for printing the binary output data of the binarizing means, the binarizing means 11 includes a minimum average error calculation unit and a Laplacian calculation unit, and The binarization means 11 is characterized in that the image data is binarized by the minimum average error method, and the input image data of the pixel of interest and surrounding pixels is subjected to Laplacian calculation to correct the density data of the pixel of interest, and the binarization means 11 further comprises: A correction value C12 is obtained by calculating the average error between the binarization error E and the weighting coefficient, and the correction value, the quantized input image data D my n, and the output R□ of the Laplacian calculation section. , and the added value D m,.

゛ is binarized using a predetermined threshold value T, and the addition value is calculated. ′ and the binarized value 0°. is added and fed back to the binarization error E.

[Effect]

Binarization means for binarizing multi-gradation quantized input image data by the minimum average error method;
A printing unit that performs printing based on value output data is provided, and when correcting the original image data by the average error generated in the processed surrounding pixels during processing by the minimum average error method in the binarization unit, the pixel of interest and This adds correction using Laplacian calculation values based on surrounding pixels.

In this case, the binarization means and the printing means may coexist in one device, or the binarization means and the printing means may be separated and connected by a communication line. .

〔Example〕

FIG. 2 is a block diagram of an embodiment of the present invention. Second
In the figure, the image input unit 11 reads image information of a document (not shown), and the minimum density (white) is 0 and the maximum density (black).
is quantized to 255 and inputted to the binarization means 12 as original image data. The binarization means 12 performs binarization using the minimum average error method, and the method will be described later.

The obtained binary data is microprocessor (CPU) memo! The information is sent to the printing section 15 of the own device under the control of the main control section 13 composed of J (ROM, RAM), etc., and based on instructions from the operating section 14 by the operator. If the printing section is separated into another device, it is sent to the communication control section 16, and then sent to the printing section 1 in the receiving side device 17 via a communication line.
Sent to 8th. In this case, it is appropriate that the printing units 15 and 18 use thermal heads as described above.

First, the binarization process using the minimum average error method will be briefly explained. Now, suppose that the n-th dot in the main scanning direction and the m-th line pixel in the sub-scanning direction in the image space are to be focused on and binarized, and the original image data representing the density of each pixel is from the minimum density to the maximum density. It is assumed that the input is quantized to a value indicating .

First, pre-stored binarization errors occurring in neighboring pixels that have already been binarized are read out, and a weighted average is determined based on a predetermined weighting coefficient. This value is added to the original image data of the pixel of interest to obtain correction data, and a predetermined threshold value (intermediate density) is calculated.
By comparing it with , binarization is performed (the value is uniquely equivalent to the maximum density or the minimum density) to obtain binary data.

Here, the difference generated during binarization is stored as a binarization error. By repeating the above steps over the entire image space, binary data binarized by the minimum average error method is obtained, and by displaying and printing, a pseudo halftone image is obtained.

FIG. 3 is a block diagram of the minimum average error binarization unit 12 of the present invention. The image input unit 11 reads the original image, quantizes it so that the minimum density (white) is O1 and the maximum density (black) is 255, and inputs it as quantized original image data to the minimum average error binarization unit 12. be done.

In the mean error minimum method binarization unit 1 (12), the target pixel to be currently binarized is set to the m-th line and n-th dot in the image space, and the quantized original image data of this n-th dot is are D□, h. Note that m and n are positive integers, and 1≦n≦N.

First, in the image input unit 11 shown in FIG. 1, an image is read by a line image sensor in which N (N is an integer) dots are arranged in the main scanning direction, and the original image data is quantized according to the image density (minimum density; 0, highest density: an integer of 255) is input to the average error minimum method binarization unit 1 (12). This original image data is input to the original image data line buffer 2 (11
is stored in the corresponding location.

The original image data line buffer 2 (11) is composed of a RAM, etc., and stores original image data for the pixel of interest to be processed and a total of three lines before and after it (m-1st, m1th, m+1th line).

The original image data DITh+n of the pixel of interest and its surrounding pixels are immediately input to the Laplacian calculation unit 204, and the following Laplacian calculations R17, . . . are performed. That is, R,,n
-K ・C(Dnn □D, -,,.) +(Dn.

D-,--1) + (D-,-D-,-+)+ (
Dnh-Dnnn. ) will be carried out. Here, K is an arbitrary integer.

On the other hand, the first occurrence occurs in binarized pixels around the pixel of interest (here, 4 pixels are used to simplify the explanation, but it is not limited to this, and a wider range may be referred to). The binarization error E is read from the error data line memory 2 (13).

On the other hand, a predetermined weighting coefficient K corresponding to each pixel position of the error data line memo 'J2 (13) is read out from the weighting coefficient matrix 2 (12).

Then, using these as weighting coefficients, the binarization error E is input to the average error calculation unit 205 to obtain a weighted average of the binarization errors.

If this weighted average is used as the correction value Cmn, the arithmetic expression is expressed as follows.

Molecule-K m l+ n-1' E m-1+ n-1
+Km-11n ・E 111-11n +Km
-l + h + l・Em-1,. Ya + + K l
l+1-1・El,.

Denominator - Km-I, n-1+ Km-1, h+ K, -+
, h+l+Kn.

Cffl+ n - numerator/denominator Next, the original image data D□1 of the pixel of interest in the original image data line buffer 2 (11) and the Laplacian calculation unit
04 calculation results R1, 7 and the average error calculation unit 205
The calculation results Cmy n are input to an adder 206, and these data are added to obtain addition results Da and h'.

And the obtained addition result. +h′ is the binarization unit 208
and is sent to the adder 207, and first, the binarization unit 208 compares it with a predetermined threshold value, for example, threshold value T=128.

(1) When correction data Dm, h'>threshold T=128 → binary data O1, n-255 (black pixel) (2) correction data D. 2. ,” When ≦Threshold T=128→
Binary data O□n-0 (white pixel) It is determined whether the pixel is a black pixel or a white pixel, and binary data (11..) is obtained.This binary data ○, n.

is sent to the binary decoder output unit 209 and output as “1” (black pixel) or “0” (white pixel), and simultaneously input to the adder 207.

Adder 207 receives input data D from binarization section 208 . .

The difference between □′ and the output data (11..) is calculated and the binarization error generated by binarizing the pixel of interest is Em2..=D□+n Omy.

As, the binarization error data line no < tufa 2 (1
Store it in the corresponding position of 3.

Repeat the above steps in the main scanning direction and sub-scanning direction 2
By transferring the binary data from the value data output section 209 to the printing section, a pseudo-halftone image subjected to the binarization process is obtained.

Note that in this embodiment, the Laplacian calculation is performed using the pixel of interest and adjacent pixels on the front, rear, left, and right sides, but the invention is not limited to this, and the calculation may be performed using a wider range of surrounding pixels.

Further, although the original image data Dmr h of the pixel of interest is inputted from the original image data buffer as input to the adder 206, the addition may be performed within the Laplacian operation section.

Furthermore, the reference range of the binarization error in the average error calculation unit 205, the minimum density O1, the maximum density 255, and the binarization threshold 1
27 is not limited to this.

FIGS. 4(a), (b), and (C) are explanatory diagrams of various Laplacian filters. In each figure, the center is the pixel of interest, and the surrounding pixels are surrounding pixels. Moreover, each numerical value is a coefficient. The Laplacian operation is calculated as the sum of the differences in density between the pixel of interest and surrounding pixels in a discrete space, and various proposals have been made as a coefficient matrix (Laplacian filter) as shown in FIG.

Correction using Laplacian calculation improves image contour (sharpness)
It has the effect of emphasizing the gradation, and is generally used in the simple binarization method.In the conventional dither method, it has negative effects such as noise occurring at the boundary where the gradation changes. Therefore it was not used.

However, in the minimum average error method of the present invention, there are no substantial gradations or boundaries, so good image quality with emphasized outlines can be obtained in text and line drawing areas, and good image quality with no noise occurring in photographic areas. image quality. Therefore, according to the present invention, by adding the Laplacian operation to the minimum average error method, a good pseudo-halftone image can be obtained for any image.

FIG. 5 is a block diagram of the thermal head. If a thermal head is used as the printing section, the effects of the present invention will be significant. The thermal head generates Joule heat by passing a current through a heating resistor 3 (12) corresponding to each pixel formed on a ceramic substrate 3 (11), and transfers this heat to thermal recording paper, thermal transfer film, etc. , forming a black dot on the recording paper.

That is, various signals ST and LA are output from the thermal head control circuit 3 (13) connected to the system bus.

The dots are turned on (black dots) and off (white dots) by SD, 5CLK, and a power supply unit 304 that supplies current to the heating resistor 3 (12).

Thermal head control circuit 3 (13 is the main control section 1 in Fig. 2)
(Provided within 13. Therefore, binary data output section 20
The binary data from 6 is input to the thermal head control circuit 3 (13) via the system bus SB.

Specifically, first, serial data SO1 for one line in the main scanning direction, that is, pixel data (High: printing, Low
w: non-printing) is taken into the shift register 305 in synchronization with the rising edge of the shift clock 5CLK. Next, when the latch signal LA is at the low level, the pixel data is held in the latch circuit 306 and simultaneously input to one of the two-input NAND circuits 307 corresponding to each heating resistor.

When the above is completed, the strobe signal ST is set to High level. Current flows through the heating resistor 3 (12), and the generated Joule heat is transmitted to the paper, forming a black dot at the corresponding pixel position. After the processing for one line is completed, the recording paper is conveyed by a predetermined sub-scanning pitch by a step motor or the like (not shown), and the same processing is repeated to obtain an image.

FIG. 6 is an explanatory diagram of printed dot sizes when a thermal head is applied to the printing section of the present invention. As is clear from the figure, the size of the formed black dots is influenced and changes depending on whether or not neighboring pixels are black dots.

In other words, the black dot of the pixel of interest becomes wider as the neighboring pixels are black. This phenomenon is caused by the fact that the thermal head handles thermal energy, and this heat is dissipated and accumulated in many directions.

As shown in FIG. 6, even for the same black dot, the dot size becomes larger in areas where there are more black dots nearby, and conversely, in areas where black dots are not adjacent, the dot size becomes smaller and there is no connection.

That is, if the black dots are placed close to each other, the heat dissipation of the heating resistor will not follow suit, and the next dot will be struck before the black dots have been sufficiently cooled. On the other hand, in a laser printer or the like, since such a heating resistor is not used, black dots are not affected by adjacent black dots and are black dots of the same size.

In other words, the granular noise and unique striped patterns in which neighboring dots are connected, which were cited as disadvantages of the minimum average error method, can be suppressed by obtaining print output using a thermal head, and by adding the Laplacian operation, A good pseudo-halftone image printout with excellent tonality and resolution can be obtained.

[Effects of the Invention] As described above, according to the present invention, in an image processing method for obtaining a pseudo-halftone image by binarizing using the minimum average error method, in addition to binarizing using the minimum average error method, a Laplacian operation is performed. This correction improves the reproducibility of text and line drawings with emphasized outlines, as well as correcting the delay in the appearance of dots in areas with low density, making it possible to improve the reproduction of text and line drawings, photographs, and halftone areas. A good pseudo-halftone image can be obtained.

[Brief explanation of drawings]

Fig. 1 is a block diagram of the principle of the present invention, Fig. 2 is a block diagram of an embodiment of the present invention, Fig. 3 is a block diagram of a binarization section of an embodiment of the present invention, and Fig. 6 is a thermal diagram. FIG. 3 is an explanatory diagram of print dot size by the head. (Explanation of codes) 1 (11... Image input section, 1 (12... Minimum average error binarization section, 1 (13...
- Main control unit, 104... Operation unit, 105, 108... Printing unit, 106... Communication control unit, 107... Receiving side device, 2 (11... Quantized original image data, 2 ( 12... Weighting matrix, 2 (13... Error data line memory, 204...
Laplacian calculation section, 205... Average error calculation section, 206, 207... Adder, 208... Binarization section, 209... Binary data output section, 3 (12... Heating resistor , 3 (13... thermal head control circuit, 304...
Power supply unit, 305...Shift register, 306...Latch circuit. Diagram for explaining printed dots by the thermal head

Claims (1)

[Claims] 1. Binarization means (11) that binarizes the input image data that has been multi-gradation quantized in the previous stage, and receives the binary output data of the binarization means and controls printing. a control means (12) for printing, and a printing means (12) for printing the binary output data of the binarization means.
3), in which the binarization means (
11) is equipped with a calculation means for the minimum average error method and a Laplacian calculation means, and binarizes the multi-tone quantized input image data by the minimum average error method, and converts the input image data of the pixel of interest and surrounding pixels to the Laplacian calculation means. An image processing device that performs calculations and corrects density data of a pixel of interest. 2. The binarization means (11) calculates the average error of the binarization error (E) and the weighting coefficient (K) to obtain a correction value (C_m
_, _n), and calculate the correction value, quantized input image data (D_m_, _n), and the output (D_m_, _n) of the Laplacian calculation unit.
R_m_, _n) and the added value (D_m_, _n'
) is binarized using a predetermined threshold (T), and the added value (D_m_, _n') and the binarized value (O_m_, _n
) is added and fed back to the binarization error (E). 3. The image processing apparatus according to claim 1, wherein the printing means is a thermal head. 4. The image processing apparatus according to claim 1, wherein the control means (12) is a microprocessor. 5. The image processing device according to claim 1, wherein the binarization means (11) and the printing means (13) are provided in one device. 6. The binarizing means (11) is provided in one device, and the printing device (13) is provided in the other separated device,
The image processing apparatus according to claim 1, wherein these are connected by a communication line (L).