JPS62163456A - Digital image reader - Google Patents

Abstract

PURPOSE:To execute an image processing such as an MTF correction and a half tone processing, etc., by a simple constitution, by providing an optical system for reading simultaneously an image data of at least a two line portion in the sub-scanning direction of a sampling pitch. CONSTITUTION:Two kinds of optical paths 8a, 8b are provided, and set to different points (a), (b) on an original surface of the upper face of a contact glass 2, respectively. A distance between two points is set to as to be equal to an image sampling interval. Two image sensors 7a, 7b read simultaneously an image data of an original of a two line portion in the main scanning direction by said two kinds of optical paths 8a, 8b, and send out its output to an image processing circuit. The image processing circuit executes a character image processing and a photographic image processing by using the data of the two line portion, which has been inputted, and also, decides whether a picture element of a processing object is in a character area or a photographic area, and selects and outputs one of the character or photographic image data which has been obtained by said processing, based on a result of its decision.

Description

[Detailed description of the invention] Five-year division The present invention relates to a digital image reading device.

Digital image reading devices such as conventional facsimiles and digital copying machines use raster scanning to read image data one line at a time in the main scanning direction, and then read the data of the pixel of interest to be processed and each adjacent pixel vertically and horizontally. Based on pixel data, MTF correction is performed through filter processing using a 3x3 matrix digital filter, etc., and halftone processing is performed using a 12N value matrix.
Since it is necessary to take out multiple lines of image data at the same time, a large number of line memories and the like are required, and the hardware is also extremely complex.

1 Kei The present invention has been made in consideration of the above points.

An object of the present invention is to provide a digital image reading device in which an image processing circuit can be easily configured when performing image processing such as MTF correction and halftone processing based on image data of a read document. .

DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, one embodiment of the present invention will be described in detail with reference to the accompanying drawings.

FIG. 1 shows the external appearance of a digital image reading device according to the present invention, in which a document is placed on the contact glass 2 on the main body 1 side and pressed down with the i-document pressure plate 3, and the operation unit 4 is used to operate the reading start button and the density. It is possible to set various functions such as selection.

FIG. 2 shows the configuration of an optical system for reading image data of the original.

As shown in the figure, in the present invention, there are two optical paths 8a and 8b.
are provided at different points a and b on the document surface on the upper surface of the contact glass 2, respectively.

The distance between two points is equal to the image sampling interval.

5a and 5b are light sources, 6a and 6b are self-hock lenses, and 7
a and 7b are contact type image sensors.

A contact image sensor is a sensor that reads the density information of a document using a 1-magnification optical system.The size of the sensor itself increases depending on the width of the document, but the optical system is simple as shown in Figure 2, and the device is easy to use. It is small in size.

In the present invention, however, in the configuration shown in FIG. 2, image data of two lines of the original in the main scanning direction can be read at the same time.

The optical system includes light sources 5a and 5b, self-hock lenses 6a and 6b, and a contact image sensor 7a.

The movable body 9, which includes the scanner 7b, is moved in the direction of the arrow in the figure to read the entire surface of the document.

Assuming that the main scanning direction is X and the moving direction of the movable body 9 is the sub-scanning Y direction, the document surface is read as shown in FIG.

First, in the first sampling, 1
Read the image information of line Q1 and second line Q2,
In the next sampling, the second line Q2 and the third line Q3 are read, as shown in FIG. 4(b). That is, the image information is read by moving one line at a time in the sub-scanning direction, and two lines of information are simultaneously read at each sampling time.

The arrows in the main scanning X direction in FIG.
It shows the direction in which the charges accumulated in b are sequentially transferred and read out.

Next, a method for processing the read image data will be described.

In general, digital image processing is broadly divided into two types based on its characteristics.

One is that in characters and line drawings, there is a clear distinction between white "0" and black "1", and the smaller the unit for determining whether it is "0" or '11 II, that is, the sampling pixel, the smaller the resolution. As the number of characters increases, an image that is more faithful to the original can be reproduced (hereinafter referred to as a 0-character image).

The other is things that have intermediate tones, such as photographic images, that is, the gray value of the pixels rather than the fineness of the pixels.

The greater the number of level decompositions, or in other words, the greater the gradation, the more faithful an image can be reproduced (hereinafter referred to as a photographic image).

MTF correction processing is normally performed on character images in order to correct the reduction in resolution due to the MTF of the reading system.

The most common MTF correction method is to use a 3×3 spatial filter. Figure 4 shows commonly used M
This is an example of a digital filter for TF correction. Figure (a)
multiplies the data of the pixel to be corrected by 3, and divides it by 4
The pixel data adjacent in the direction are multiplied by 1 1/2, all are added, and the result is used as correction data. The same figure (b
), the correction is made by multiplying the pixels adjacent in 8 directions (up, down, left, right, and diagonally) by 11 and multiplying itself by 9. There are various ways to select the 6 coefficients, and weights may be given depending on the direction. , the sum of the coefficients is usually 1.

The effect of MTF correction is as shown in FIG. Increases the entrast of low-entrast signals or increases the sharpness of edges. In the figure, (a) shows the characteristics before correction, (
b) shows the characteristics after correction. Also, S.L.
is the threshold level.

However, this method has two drawbacks. One is that it also increases the amplitude of noise and moiré.

Another problem is that in order to obtain a correction value for one line, data for the lines before and after it must be retrieved at the same time, so conventional methods require line memory for at least two lines. Above, there was a delay of at least one line between inputting a certain line and obtaining the output of that line.

The present invention employs an automatic threshold value selection method that compensates for these drawbacks.

The method uses pixels to the right (or left) of the pixel to be subjected to MTF correction and pixels below (or above), that is, adjacent pixels in two directions. The absolute value of the difference between the pixel to be corrected and the right pixel and lower pixel is taken. That is, the differential values of the pixel in the main scanning direction and the sub-scanning direction are taken, and the larger one is taken out. If the differential value is smaller than a certain value, it is determined that the change is due to noise or moiré, or a gradation change in an intermediate tone, and the threshold value for the character image is left at its initial value (fixed value), and the differential value When the value is larger than a certain value, the threshold value is changed to an intermediate value between the level of the pixel to be subjected to MTF correction and the level of the adjacent pixel that has undergone the larger change. This makes it possible to sharpen the image, which is the main purpose of MTF correction. Text images are MT
After F correction, it is binarized using a fixed threshold value.

For photographic images, the dither method, the density pattern method, and the submatrix method, which is an intermediate method between the two, have been widely used.

These methods utilize the integral effect of the human eye, which perceives shading depending on the ratio of black dots to white dots within a certain area, and are realized by systematically changing the threshold value.

FIG. 6 is an example showing how the threshold values change systematically, which is called a threshold value matrix, and this matrix is used repeatedly in the vertical and horizontal directions.

The dither method is a method of determining the output of one pixel by associating one pixel of human data with one element of a matrix.

The density pattern method is a method of associating one input data with all elements of a matrix and determining output pixels having the same size as the matrix. The submatrix method divides the matrix into several small matrices and associates one input with each of the small matrices to determine the output.

When the size of the output image is assumed to be constant, the dither method uses the smallest data input unit and has excellent resolution, but the noise in the signal tends to appear in the output image, resulting in a noisy image and a high resolution. The tonality cannot be reproduced satisfactorily. On the other hand, in the density pattern method, the data input unit is the coarsest, but noise is removed, natural density changes are reproduced, and the gradation is excellent, but the edges are noticeably blurred and the resolution is poor. The submatrix method is between these two methods, and depending on how the size of the submatrix is selected, it is possible to obtain an image with a good balance between resolution and gradation.

However, no matter which method is used, the resolution is lower than that of an output image processed for a character image. In other words, text images and photographic images must be separated and processed separately.

In the present invention, division into character image areas and photographic image areas is performed at the same time as each process is performed, and an automatic area division method is adopted that provides optimal output for all parts of any input image.

As the area division method, similar to the automatic threshold selection described above, differential values in the main scanning direction and the sub-scanning direction are taken, and the larger one is extracted. In the case of region determination, the sum of both may be used instead of the larger one.

In the case of a character area, the black level and white level change frequently, so large values of this differential value appear frequently. However, in the case of photographic images, this differential value is generally small, and even if a large value occurs, the frequency is small.

This area division method utilizes this, and accumulates the differential value by multiplying it by a certain coefficient with sampling synchronization, and at the same time subtracts the value accumulated up to the previous time by a certain coefficient. Therefore, if large values occur frequently, the accumulated value will gradually increase, and conversely, if the differential value is small or the frequency of large values is small, the accumulated value will gradually decrease. That is, by looking at this accumulated value, it can be determined whether the area containing the pixels at that time is a text area or a photo area.

FIG. 7 shows an image processing circuit according to the present invention. As mentioned above, two lines Q j read simultaneously,
Input image data of Q (j+1). The data input here is a digital signal of several bits/pixel that is read out by the image sensors 7a and 7b shown in FIG. It is. Generally, it is 4 to 8 bits, and here it is 6 bits (64 gradations).

In FIG. 7, X is first processed in the differential value detection circuit 9. Take a pixel from the Qj line and take the same <12j from the X line. xl, which is the right neighboring pixel of , from the ρ(j+1) line. Take out the X1 pixel next to below. And the differential value IXOxll, lXo−X mentioned earlier
21, and compare them to find the larger value and the larger value of the adjacent pixel (X,.

Select r X2). These are then input to the next threshold selection circuit 1o.

The threshold value selection circuit 1o first checks whether the differential value is larger than a certain fixed value TRI. If it is smaller than TRI, a predetermined threshold value TR2 is selected as the threshold value for the character image. If it is larger than TRI, the first selected X□
orX2, the differential value is the coefficient α1 (-1
≠0) and then add the product and use this as the threshold value. The delay circuit 12 in front of the character/line character processing circuit 13 in FIG. 7 is for adjusting the time taken to select the threshold value.

Then, the pixel data is compared with the selected threshold value to determine the output.

On the other hand, processing for photographic images is also performed in parallel.

In Figure 7, the submatrix method described above is used.
In the 2×2 average value circuit 14, the Qj line and (1(j+
1) A 2×2 small pixel area is averaged from simultaneous input of lines.

That is, x,, x,, x2. An average value is calculated from x3, and D is used as a new input unit, and the 8×8 threshold matrix is cut into 2×2 small matrices, as shown in FIG. 9, to correspond to D.

In FIG. 7, in the halftone processing circuit 15, the threshold value matrix table (ROM) contains the ρj line and the X to be processed.
. XY address corresponding to the pixel (X=O~7.Y=O~
7), output the threshold value, and compare it with D to determine the output.

On the other hand, since it is also used when outputting the n (j+1) line, it is written in the RAM at the same time as the comparison, and is sequentially read out from the RAM for the next line.

Here, a delay circuit 16 is inserted in order to match the timing with the appearance of the character image, but this may be done after the comparison output as shown in FIG. 7, or after the appearance of the average value or before the average value processing.

The halftone processing does not need to be a submatrix method, and may be a dither method or a density pattern method. The selection can be made according to the purpose of use of the device. However, the density pattern method has the drawback of requiring a large number of memories in order to synchronize with the appearance of characters. On the other hand, if the dither method is used, there is no need to use any RAM.

Up to this point, we have reached the point where the appearance of text images and the appearance of photographic images are retrieved in synchronization. Next, the area determination circuit 11 will be explained. The larger differential value of the main scanning and sub-scanning, which was previously extracted by the differential value detection circuit 9, is input, the value is multiplied by a coefficient α2 (o<α2<1), and the result is input into an adder for accumulation.

On the other hand, the value output from the adder is the value accumulated up to the previous time, but it is added by the coefficient α3 (0 < α3 × 1
) to reduce the accumulated value and put it into the adder again.

In other words, the area determination circuit 11 performs accumulation and reduction at the same time, and if this equilibrium is broken and the accumulation progresses, the area is judged to be a character image area, and conversely, if the reduction progresses, it is judged to be a photographic image area. This is realized by comparing the accumulated value with a certain fixed value TR3.

Since this comparison value is output corresponding to each pixel, the selector 17 selects the optimal one between output for character images and output for photographic images.

FIG. 8 shows the operation timing of each part in the image processing circuit shown in FIG. 7. In the figure, dl.

d2 indicates the delay time.

As shown in the figure, in this circuit, the Ω1 line and Q(j+1
) line, and the output of the Q1 line can be obtained in real time from the data of the two lines, although there is a slight delay in the main scanning direction. Also, the QCj+1) line is Q1
It is mainly used for line image quality improvement (MTF correction) and area determination, so the Q1 line and Q (j+
1) Setting the distance from the line does not require much precision.

In addition to the method shown in FIG. 2, several other methods can be considered for simultaneously inputting the Q1 line and the Q (j41) line to the image processing circuit.

Although Fig. 2 uses a contact type sensor, it can also be realized using a reduction optical system.

Even with the conventional method of reading one line at a time, the circuit shown in FIG. 7 can be realized by having a line memory for one line before inputting it to the image processing circuit. It can also be realized by using an image sensor in which sensor chips are arranged in two lines in parallel. Alternatively, it is also possible to apply a color image sensor.

For color sensors, there are methods such as dividing one sensor chip into three parts, separating the light into R, G, and B using color filters, etc., and detecting each with each divided chip.
This can be achieved by using a sensor divided in the sub-scanning direction and without using a filter.

In the digital image reading device according to the present invention,
Since it is possible to simultaneously read at least two lines of image data from a document, appropriate image processing can be performed using a simplified circuit configuration that does not require any line memory as in the past. In addition, it has the excellent advantage that image data for doubling the sampling pitch can be easily created.

[Brief explanation of drawings]

FIG. 1 is a perspective view showing the external appearance of a digital image reading device according to the present invention, and FIG. 2 is a simplified diagram showing an example of the configuration of its optical system.
3(a) and 3(b) are diagrams showing the reading and scanning state of the document surface according to the present invention, and FIGS. 4(a) and 4(b) are diagrams showing an example of the configuration of a digital filter for MTF correction. 5 is a diagram showing characteristics of read image data and its MTF-corrected data, FIG. 6 is a diagram showing a threshold value matrix, and FIG. 7 is a block diagram showing an example of the configuration of an image processing circuit according to the present invention. FIG. 8 is a time chart of the operation of each part in the image processing circuit, and FIG. 9 is a diagram showing the correspondence between processing data and threshold matrix during halftone processing. 7a, 7b... Image sensor 9... Differential value detection circuit 10... Threshold value selection circuit 11... Area determination circuit 12.16... Delay circuit 13... Character line processing circuit 14...・2X2 average value circuit 15... Halftone processing circuit 17... Selector

Claims (1)

[Claims] In a digital image reading device that reads image data of a document pixel by pixel at a predetermined sampling pitch for each line in the main scanning direction, and finally produces a binary output of each pixel, the sampling pitch is at least two in the sub-scanning direction.
A digital image reading device characterized by having an optical system that simultaneously reads image data for lines.