JPH04150573A - Image processing device - Google Patents

Abstract

PURPOSE:To surely read a thin line, etc., plotted on an original by providing a switching means which changes weight or interrupts the addition itself of an error when the error value of the weight with a higher value allocated to the error of each peripheral picture element shows a negative value. CONSTITUTION:The switching means 26 fetches each error value of the peripheral picture elements (a-d), and is provided with a judging part 2 which judges whether or not the error value of the picture element (d) to which the weight with higher value is allocated shows the negative value. When the error value of the picture element (d) shows a positive value, a switching part 26b is operated, land connection to a first error allocation arithmetic circuit 27 is performed, while, when the error value of the picture element (d) shows the negative value, the switching part 26b is operated, and the connection to a second error allocation arithmetic circuit 28 is performed. In other words, the switching is performed so as to change the weight or to interrupt the addition itself of the weight when the error value with higher weight allocated to the error of each of the peripheral picture elements (a-d) shows the negative value. In such a way, it is possible to surely read the thin line, etc., plotted on the original.

Description

[Detailed Description of the Invention] First-year-old baby's skin ■ minutes! The present invention relates to an improved image processing device that can reliably read thin lines drawn on a document, and is used in, for example, a facsimile machine.

' and its amount The image reading device is, for example, an LED (LightEmitt
A light source consisting of a CCD (Ching Diode) and a CCD (Ch.
(large coupled device). The CCD includes at least a group of pixel photodiodes and a shift register. The operation of this type of image reading device will be explained as follows. That is,
The light source irradiates the document with light, and the reflected light (information light) reflected by the document enters the photodiode group.

The photodiode group accumulates optical signal charges corresponding to reflected light, and this optical signal charges are transferred to the shift register at a predetermined timing. And this shift register is
Concentration level information (brightness information) based on the optical signal is sequentially output based on a predetermined clock. The output density level information for each pixel is compared with a preset reference value and determined as either white or black. In other words, if the density level has a value higher (brighter) than the reference value, it is judged as white, and if the density level has a value lower (darker) than the reference value, it is judged as black. processing is performed.

20 In the binarization process, an error diffusion method has conventionally been used to improve halftone reproduction. This error diffusion method will be explained below.

FIG. 3 shows a pixel of interest e and surrounding pixels a, b, and c.

d and the next pixel of interest f. Here,
The pixel of interest e is the pixel to be read, and the surrounding pixels a and b
, c, and d are the pixels referenced at that time, and the next pixel of interest f is the pixel to be read next to the pixel of interest e.

The error diffusion method adds the weighted average of errors distributed from surrounding pixels a, b, c, and d to the actual density level of the pixel of interest e, and compares this corrected density level with a reference value using a comparator. In addition to determining whether pixel e is black or white, this pixel of interest e
, and accumulates it in an error buffer for use in determining the density of the next pixel of interest f.

However, in image processing based on such an error diffusion method, the following problems occur. Note that the concentration level is set to 0.
(assuming that there is a range from (white) to 63 (black) and 6 bits), the reference value is set to 31, and the weighting of errors is as shown in FIG. Here, if the corrected density level of the pixel of interest e is 35, the error of this pixel of interest e is 35.
-63=-28, which is accumulated in the error buffer. Then, in the processing of the next pixel of interest f, pixel e
The weighted average of the errors including the value of -28 multiplied by 7 is added to the actual density level of the pixel f of interest, and the corrected density level is calculated.

In this case, the weighted average of the errors tends to be negative because the pixel e to which a heavy weight is assigned is negative, and the actual density level of the pixel f of interest is 35 (this value would normally be black). If the corrected density level is less than the actual density level and falls below the reference value 31, the corrected density level may be determined to be white. For this reason, if the image to be read is composed of lines, for example, and these lines are extremely thin, discontinuities will occur in the reproduced thin lines.

In view of the above circumstances, it is an object of the present invention to provide an image reading device that can reliably read thin lines drawn on a document.

In order to solve the above-mentioned problem, an image processing device according to the present invention for reading a pixel calculates a corrected density level of a pixel to be read by taking into account errors of a plurality of pixels located around the pixel to be read. In an image processing apparatus that determines whether a pixel to be read is black or white based on this corrected density level,
If the error value of the heavy weight assigned to the error of each surrounding pixel is negative, a switching means is provided to change the weight or stop adding the error itself. It is a feature.

According to the above configuration, the inversion of the pixels to be read, in particular,
This prevents the image from being reversed to white when it should be black, and allows even fine lines to be reproduced well.

An embodiment of the present invention will be described below with reference to FIGS. 1 and 2.

In FIG. 1, 11 is a light source made of, for example, an LED;
2 is a CCD (reading means), and 15 is a document.

The light source 11 is provided so that its longitudinal direction coincides with the width direction of the document, and is configured to irradiate light onto the document 15 in a linear manner in the width direction.

The CCD 12 includes an optical signal storage section 12a and a shift register 12b. The optical signal storage section 12a is
It is provided so that its longitudinal direction coincides with the width direction of the document, and is provided with a group of photodiodes in this longitudinal direction.

Further, each photodiode is configured to take in reflected light (information light) reflected by the original 15 and accumulate optical signal charges corresponding to an image.

The shift register 12b transfers the image data signal based on the optical signal transferred from the optical signal storage section 12a to the line memory 18.
(shown in Figure 2). The line memory 18 has the ability to hold 6-bit data for 2048 pieces x 2 lines.

Note that the reason why the number is 2048 is that the above-mentioned CCD has an effective number of pixels of 2048.

FIG. 2 is a block diagram showing the main parts of the image processing device. In the figure, 18 is a line memory, 19 is an addressing control section, 20 is an addition section, 21 is a comparator, 22 is a subtraction section, 23 is an output buffer, 24 is an error buffer, 25 is an error succession control section, and 26 is a switch. 27 is a first error distribution calculation circuit, and 28 is a second error distribution calculation circuit.

The address specification control unit 19 specifies the address of the pixel e to be read (hereinafter referred to as the pixel of interest), and the line memory 18 stores the density level f of the pixel of interest e stored at the specified address. , 7 is output. Note that pixel f indicates the next pixel of interest.

In the addition section 20, a weighted average of errors Ea1.7, which will be described later, is added to the density level f1.7, and the resulting corrected density levels f'll and I'l are output to the comparator 21. Corrected density level f'. 1. is expressed by the following first equation.

f' a+lI =f, , , +Eas, n '''1st
The equation comparator 21 compares the corrected density level f'll+11 with the reference value T (T=P/2: P is the black level 63), and if f'll+ll≧T, gva+n
63 is assigned as ``gl'', and this is sent to the output buffer 23. On the other hand, if f', . The output buffer 23 holds gl and a, and has the ability to hold data for one line.

The subtraction unit 22 calculates an error value E, , 4, which is a value obtained by subtracting g, , 7 from f'll+fi, and outputs this to the error buffer 24 . The error value E0,7 is expressed by the second equation below.

E,,,l=f-111+,-g,,i...The second equation error buffer 24 holds the error values E1 and 7, and has the ability to hold two lines of data.

The error succession control section 25 specifies addresses necessary for extracting error values of surrounding pixels a, b, c, and d stored in the error buffer 24.

The switching means 26 includes a determining unit 26a that takes in each error value of surrounding pixels a, b, c, and d and determines whether or not the error value of d, which is a pixel to which a heavy object to be described later is assigned, is negative. However, if the error value of pixel d is positive, the switching unit 26b is operated to connect to the first error distribution calculation circuit 27, while
If the error value of pixel d is negative, operate the switching unit 26b,
Connection with the second error distribution calculation circuit 28 is made.

The first error distribution calculation circuit 27 assigns a weight of 1 to the error value of pixel a, a weight of 5 to the error value of pixel S, a weight of 3 to the error value of pixel C, and a weight of 3 to the error value of pixel C. A weight of 7 is distributed to the error value of , and the weight average Ea, . . . 1 is obtained. The above weight average Ea, 7 is expressed by the following third formula.

E as, n = (Ell-1, Fl-1+ 5 E
IN-1,R+3 E,-1,□117 E,,,-1
) / 16 ... Third equation On the other hand, the second error distribution calculation circuit 28 assigns a weight of 1 to the error value of pixel a, a weight of 5 to the error value of pixel S, and a weight of 5 to the error value of pixel C. A weight of 3 is allocated to the value and a weight of 5 is allocated to the error value of pixel d, and the weighted average E a m-n is calculated.

The above weight average Ea, , A is expressed by the following 4th formula.

E am, n = (Elm-1++1-1 + 5
Elm-1+11 + 3 E, -1, □++5E,
, □l)/14 ... According to the fourth formula above,
This prevents pixels to be read from being inverted, especially from being inverted to white when they should be black, and it is possible to reproduce fine lines well.

That is, the corrected density level f·1. of the pixel of interest e. Assuming that ゎ is 35, the error of this pixel of interest e is 35-63
= -28, which is stored in the error buffer 24 as E,...
=-28. Then, in processing the next pixel of interest f, since the error of pixel e is negative, the second error distribution calculation circuit 28 whose weight is 5 is selected, and the error including the value multiplied by this 5 is selected. The weighted average is the pixel of interest f
is added to the concentration level of Therefore, compared to the case where the first error distribution calculation circuit 27 (weight 7) is selected, the weighted average of the errors E am+n is less likely to become negative, and the density level of the pixel f of interest is 35 (which is the original value). ), the corrected density level f'
Since IN+l'l is prevented from falling below the reference value 31, black is determined as expected, and black->white reversal is prevented, even thin lines can be reproduced satisfactorily.

In this embodiment, only the pixel d to which a heavy weight is assigned is considered, but other pixels to which a heavy weight is assigned may also be considered. That is, pixels d, b
The weight may be changed when both of the pixels d and b are negative, or the weight may be changed when one of the pixels d and b is negative.

4゜Also, in this embodiment, an example is shown in which the weight is changed when the error value of a heavy object is negative, but in addition, means for switching to stop adding the error itself is provided. It may be done as follows.

According to the present invention as described above, it is possible to reliably read thin lines drawn on the original document.

[Brief explanation of drawings]

1 and 2 show an embodiment of the present invention, in which FIG. 1 is an explanatory diagram showing an image reading section, FIG. 2 is a block diagram showing main parts of an image processing device, and FIG. The figure is an explanatory diagram showing a pixel of interest and surrounding pixels, and FIG. 4 is an explanatory diagram showing error weighting in the error diffusion method. DESCRIPTION OF SYMBOLS 11... Light source, 12... COD, 1B... Line memory, 19... Address specification control part, 20... Addition part, 21... Comparator, 22... Subtraction part, 23・・・
・Output huffer, 24...Error buffer, 25...
Error succession control section, 26... switching means, 26a...
Judgment unit, 26b...Switching unit, 27...First error allocation calculation circuit, 28...Second error allocation calculation circuit. Patent applicant Murata Machinery Co., Ltd. Figure 3 n+1 n+1

Claims (1)

[Claims] (1) An image in which the corrected density level of the pixel to be read is calculated by taking into account the errors of multiple pixels located around the pixel to be read, and whether the pixel to be read is black or white is determined based on this corrected density level. The processing device includes switching means for changing the weight or stopping adding the error itself when the error value of the heavy weight assigned to the error of each surrounding pixel is negative. An image processing device characterized by: