JPS61137469A - Image processing system - Google Patents

Abstract

PURPOSE:To make optimum MTF corrections suitable to the response of an input system by using two-dimensional digital filters which have various characteristics and selecting a two-dimensional filter to be processed optionally according to filter specification. CONSTITUTION:When an MTF correcting circuit 11 corrects the image information BS of an original read by a scanner by two-dimensional digital filter processing, two-dimensional digital filters having various characteristics are used and an optional two-dimensional filter is selected according to a command FCS for filter specification suitable to the response of the input system. Further, the response of the input system for image information is different, scanner by scanner, and even one scanner differs in response between the main scanning direction and subscanning directions. For the purpose, this system uses plural two-dimensional filters (including Laplacean filters) suitable to responses of various input systems and uses the best two-dimensional filter selectively for image reading.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to image processing that digitally processes image information read by sampling an image of a document pixel by pixel using a scanner, and particularly relates to image processing that digitally processes image information read by a scanner.
The present invention relates to an image processing method that performs TF correction.

Hair Tohin 4 In general, the image quality of recorded and reproduced images is improved by correcting the response of the image input system due to non-uniformity of characteristics in the optical system when reading the image information of the original with a scanner, uneven feeding of the scanning system, etc. In order to achieve this, it is necessary to perform so-called MTF correction, in which digital image information read by a scanner and multivalued quantized is subjected to two-dimensional digital filter processing.

Conventionally, a simple method using a Laplacian filter has been often used for this type of MTF correction, but the Laplacian filter is simply used to emphasize edges in pattern images such as characters and line drawings, and is not intended to be It is not a value that takes into account the response of the input system.

It does not adapt well to the response of the input system, which differs from scanner to scanner, and as a result, the black and white edges of thick lines and solid black areas are overemphasized.

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

The present invention provides an image processing method that allows optimal MTF correction to be performed that matches the response of an input system.

In order to achieve the objective, the present invention prepares a plurality of two-dimensional digital filters with various characteristics when performing MTF correction on image information read by a scanner, and processes them according to filter specifications. This makes it possible to arbitrarily select a two-dimensional digital filter.

Hereinafter, one embodiment of the present invention will be described in detail with reference to the accompanying drawings.

FIG. 4 shows an example of the configuration of a scanner for reading an original image. A first traveling body moves an original 2 placed on a contact glass l parallel to the original 2 in the sub-scanning direction at a speed V along a guide 3. 4 is sequentially illuminated from below in the main scanning direction by a light source 5 attached to the document surface, and the reflected light from the document surface is reflected onto the first mirror 6, which is also attached to the first traveling body 4.
Pixel by pixel via a second mirror 8 for optical path length correction attached to a second traveling body 7 that moves in the sub-scanning direction along a guide 3 at a speed of V/2 and an imaging lens 9 provided on the fixed side. A fixed image sensor such as a CCD is
The original image is sent to a line image sensor 10 arranged in a line-distributed manner and is sequentially read. In addition, here, the linear speed V of sub-scanning feed is 180 + * m/ See
The reading density is 400 pixels/in both the main scanning and sub-scanning directions.
I try to make it in inches. Therefore, the scanning time per pixel is approximately 70 nSec (14.3 MHz). In addition, the line image sensor 10 has 5000 pixels, and its output is divided into two channels, odd and even pixel columns, and the output data from each of the odd and even pixel columns is processed in parallel. The processing time per pixel is apparently 140 nSec.

FIG. 5 shows an example of the configuration of a control system and a data processing system in a single image reading system. Here, the interface 51 with the printer etc. varies depending on the system configuration.

I'm listing the basics here. Commands C8 are divided into those related to control such as starting, stopping, or abnormal conditions of the single-image reading system, and those related to control such as selection of data processing method. Most of the commands are input by an operator via an operation panel or the like. The command C5 is once given to the controller 52, which checks the system status and then sends it to each necessary location.

Commands related to data processing include specifying the document size, scaling ratio, and turning on/off background noise removal.

These include document mode/photo mode, density position selection designation, MTF correction filter designation, etc.

In the configuration shown in FIG. 5, when the controller 52 receives a scanner start command, it drives the servo motor M to start the optical system and start reading the original image. At that time, the servo motor M is subjected to position and speed feedback control using an encoder, and the linear speed is 18
The accuracy is maintained within ±2% with respect to 0wes/See. Also, at this time, the return position and linear speed of the optical system are controlled in accordance with commands for the document size and magnification ratio. Further, at this time, the clock generator 53 provides a control clock CLK to the line image sensor 10, the video signal processing circuit 54, and the image processing device 55. The line image sensor 10 reads image information of a document line by line in the main scanning direction. At that time, the reading density of the scanner is 400 pixels/inch, and the linear speed is 180 +mw+/S.
ee, the reading time per line is 353μSee
becomes. Clock CLK from line image sensor lO
The data of the odd-numbered pixel columns and the even-numbered pixel columns read out in time series are sent to the video signal processing circuit 54.

Therefore, data amplification is first performed. At that time, the magnitude of the output signal of the line image sensor [0 is the maximum (white level)
In order to obtain 120-200 mV, amplify it to nearly 1 V. Next, in order to remove the DC offset in the output signal of the line image sensor 10 and the DC offset of the video amplifier, a MOS F is installed in the dummy sensor output.
Use ET to zero hold. Furthermore, a sample and hold is performed to remove the undulations of the line image sensor output signal. The video signal processing circuit 54 then corrects shading caused by uneven illuminance of the light source in the scanner, removes the background of the document, and finally converts the data read pixel by pixel by the scanner into a 6-bit digital signal. and sends it to the image processing device 55. Note that background removal is turned on and off as appropriate by commands.

In the image processing device 55 according to the present invention, as shown in FIG. 1, the image information BS (in this case, odd number (separated into two channels: image information BSI for pixel columns and image information BS2 for even-numbered pixel columns)
A first processing system includes an MTF correction circuit 11 and a binarization circuit 12 that performs optimal binarization processing for the pattern image, and also performs optimal binarization processing for the grayscale image of the input image information BS. Halftone processing circuit 13 that performs processing
a second processing system consisting of a data switching control circuit 14 that outputs a switching signal S in accordance with the area designation of the grayscale image portion in the original image; a data switching circuit 15 that switches between the output data of the conversion circuit 12 and the output data of the halftone processing circuit 13 in the second processing system; and an output buffer circuit 16 that stores the appropriately switched processing data.
(Here, it consists of a RAM 1 for storing processing data of odd-numbered pixel columns and a RAM 2 for storing processing data of even-numbered pixel columns). Here, especially when a pattern image area and a grayscale image area coexist in the same document, the optimal binarization process can be performed separately for each image area according to the image information read by the scanner. That's what I do.

As shown in FIG. 2, the data switching control circuit 14 uses position data X (PI), Y ( Ql), X (Pl), Y
(Q2) and the counter outputs of the main scanning counter 141 that counts the pixel clock D-CLK in the main scanning direction and the sub-scanning counter 142 that counts the line synchronization signal L-5YNC in the sub-scanning direction, respectively. Comparator 1 to compare
43 to 146, and a logic gate circuit that outputs a switching signal S as appropriate depending on the combination of the output states of these comparators. Further, the data switching circuit 15 turns on the output side of the binarization circuit 12 and cuts off the output side of the halftone processing circuit 13 when the switching signal S is Ll, Vt, and when the switching signal S is set to II O7H. It is comprised of a gate circuit that turns off the output side of the binarization circuit 12 and turns on the output side of the halftone processing circuit 13 when . Note that the comparator 143 compares the position data X (Pl) on its input A side with the output data of the main scanning counter 141 on its input B side, and when the output is A3B, the output becomes "1"'. compares the position data Y (Ql) and the output data of the sub-scanning counter 142, and the output becomes '1'' when it is A3B, and the comparator 145 compares the position data When A≧B, the output is II 171, and the comparator 146 compares the position data Y (Q2) with the output data of the sub-scanning counter 142, and when A≧8, the output is ## L II. It is designed to be.

In the data switching control circuit 14 configured in this way, for example, as shown in FIG.
When setting the upper hatched area to the grayscale image mode, the operator inputs the scanning start point (pt, Ql) and the scanning end point (Pl) on the two-dimensional coordinates using an input means such as a keyboard.

Q2) is input. Accordingly, the system controller converts the positions of the input coordinate points into main scanning count values X (PL), X (Pl) and sub-scanning count values Y (Ql), Y (Q2), respectively. is applied to the switching control circuit 14. Therefore, when the scanner is scanning the grayscale image part in the original image, that is, when X (PL)≦C1≦X (P2), Y (Ql)
When ≦C2≦Y (Q2), the switching signal S output from the data switching control circuit 14 becomes ITOIt, and the halftone processing circuit 13 side is selected by the data switching circuit 15. Further, when the scanner is scanning a grayscale image part in the original image, the switching signal S becomes '1'', and the data switching circuit 15 selects the binarization circuit 12 side. Note that the data switching control circuit Instead of providing the switch 14, it is also possible to directly control the switching signal S using a CPU or the like.

In the image processing system according to the present invention, when the image information BS of the document read by the scanner is subjected to the two-dimensional digital filter processing in the MTF correction circuit 11, the image processing method according to the present invention has various characteristics in advance. A plurality of two-dimensional digital filters are prepared, and an arbitrary two-dimensional digital filter is selected according to a filter designation command FC5 that matches the response of the input system.

MTF correction essentially uses a two-dimensional digital filter to correct the response of the input system caused by variations in the characteristics of each CCD in the line image sensor 10 in the scanner, uneven aberrations of the imaging lens 9, and vibrations of the traveling body 4. This is corrected through processing. Figure 6 shows input image information B.
MTF is applied to S using a Laplacian filter F with a 3×3 configuration.
This shows the case of correction, and the target pixel D to be processed is
The processing output O by the Laplacian filter F of No. 5 is given by ◯=3D5-(D2+D4+D6+D8)/2, taking into account the surrounding pixels. In the figure, X indicates the main scanning direction, and Y indicates the sub-scanning direction.

The response of the image information input system varies depending on the scanner.
Furthermore, even in one scanner, the response in the main scanning direction and the sub-scanning direction often differ. Therefore, in the present invention, a plurality of two-dimensional digital filters (
(including a Laplacian filter), so that an optimal two-dimensional digital filter can be selectively used when reading an image.

Specifically, as shown in FIG. 7, data from each processing result when using a plurality of two-dimensional digital filters is written in the ROM in advance, and the command FC5 for specifying the filter and the data for the target pixel D5 are written. A predetermined processing result by an appropriately selected two-dimensional digital filter is read out by using each data of the data and the surrounding pixel combinations D2+D8 and D4+D6 as addresses, respectively.

At this time, if the two-dimensional digital filter has the same filter constants corresponding to surrounding pixels as shown in FIG. 6, for example, data from the surrounding pixel combination D2+D4+D6+D8 can be used as the address. Also, MT using such ROM addressing
The F correction means can be applied even when only one type of two-dimensional digital filter is used (in this case, the filter designation command FC3 is not required).

In the configuration shown in Fig. 1, the binarization circuit 12 is provided with seven threshold values, and the optimal threshold for the state of the original image to be read is determined by the density position command given from the controller. The value can be selected arbitrarily. Further, the halftone processing circuit 13 uses a dither method, a density pattern method, or a partial matrix method located between these methods.

FIG. 8(,) shows an 8×8 threshold matrix used in the partial matrix method.

Then, cut out a 4×4 submatrix and make it correspond to the input data in pixel units and that submatrix.
6 In the same figure (b), it is possible to reproduce four gradations.
A 4×4 submatrix is shown. In other words, if we use that threshold matrix, the gradation unit is 8×8,
The unit of resolution is 4×4. Also, the size of the threshold matrix and the size of the submatrix.

It goes without saying that the shape can be applied in various ways.

FIG. 9 shows the principle of the partial matrix method using a threshold matrix. In the figure, BS indicates six-value digitized input data, and P indicates an output pattern. In FIG. 8, when the number of pixels of the input data is compared with the number of pixels of the output pattern, the number of pixels of the output pattern is 16 times larger. If the size of the input pixel and the output pixel are the same, the output will be 16 times the area of the input.

For example, the size of both the input pixel and the output pixel is 4.
In the case of 00 dots/inch, if one person is using the same output, it is necessary to group the input data into units of the size of the partial matrix so that the input and output sizes are the same. In this case, a method of taking an average value may be considered. Taking an average value is smoothing, and it is advantageous for grayscale images because it can remove high frequency components.If you add pixel data for the size of the partial matrices, the result of the addition will be The gradation of data is greater than the gradation of one pixel. For example, if the data for one pixel is 6 bits (64 gradations) and addition is performed using a 4 x 4 submatrix, the result is 10 bits (10
24024 gradations. If we take the average value, we will divide the addition result by 16 and get the original 6 bits.

The 4 bits that are discarded at this time also have meaning. However, since the human eye is said to have a range of 64 to 256 gradations,
Here, the upper 8 bits (256 gradations) are used. In other words, the size of the threshold matrix is set to 1
The size is 6×16, the unit of gradation is 256 gradations, and the unit of resolution is 4×4.

Therefore, 64 gray levels per input pixel to 256 output gray levels are thereby obtained.

10 shows a specific configuration example of the halftone processing circuit 13 for realizing this. here.

First, the image information BS of odd-numbered pixel rows sent from the scanner
Addition of four pixels in the main scanning direction is performed by the first adder 17, latch 18, and second adder 19 at the timing shown in FIG. , the addition result A for the first sub-scanning line
Write DDI to RAM 20. Next, at the second sub-scanning line, the result of addition is ADD2 and R
Addition result A of the first sub-scanning line written to AM20
DDI and is added by the third adder 21 and the result is written into the RAM 22. In the third sub-scanning line, the adder 21 adds the addition result ADD3 of the third line and the addition result previously written to the RAM 2'2, and writes the result to the RAM 23. Similarly, what is added to the fourth sub-scanning line becomes an addition result corresponding to 4×4.

This is stored in the RAM 24a, and while the addition result of the subsequent four lines of sub-scanning is stored in the RAM 24b,
The data stored in the AM 24a is read out, and compared in the comparator 25a with the threshold values corresponding to the main scanning position and the sub-scanning position read out from the PROM 26a in which a 16×16 threshold matrix is written in advance. , if data≧threshold, re 1 rr is output, and if data is smaller than threshold, RO)g is output. The binarized output is arranged in two systems: output data Di for odd pixel columns and output data D2 for even pixel columns. Furthermore, PR
The OM26 stores seven types of threshold matrix data as described above, and the threshold matrix is appropriately selected according to the density position command DO5. In the figure, 27 indicates an input/output switching circuit. Also, the timing in the sub-scanning direction is set to 12th.
Shown in the figure.

In the figure, W is a write cycle and R is a read cycle. In the figure, since the usage timings of RAM 20 and RAM 23 are reversed, it is also possible to share them with one RAM.

Also as a command given to PROM26.

There is a concentration position. There are seven density positions, and input-output conversion as shown in FIG. 13 is considered.

In Fig. 13, normal halftone processing in which the number of output gradations (black area ratio) is linear with respect to input data X is shown as y=f4(
x), fl (x) - f3 (X) side is dark idiom, f
For the 5(x) to f7(X) sides, a light tone function is considered, and the elements of the threshold matrix are determined so as to realize the conversion. fL(x) ~ f7(x)
After determining the seven threshold matrices corresponding to , they are stored in the FROM 26.

Writing to the FROM 26 is performed using parallel processing for odd and even numbers, so the threshold data is also divided into odd and even numbers in the main scanning direction and written to the two FROM 26a and FROM 26.
Each is stored in b. Note that the address of FROM 26 is as shown in FIG.

The output data Ct of the lower 3 bits of the main scanning counter is assigned to addresses AO-A2, and the output data C2 of the lower 4 bits of the sub-scanning counter are assigned to addresses A3-86, so that the threshold value can be read. The density position command DO3 is given to the addresses A7 to A9.

As described above, in the image processing method according to the present invention, when performing MTF correction of image information read by a scanner, a plurality of two-dimensional digital filters with various characteristics are prepared, and a plurality of two-dimensional digital filters are prepared according to the filter specification. This allows the user to arbitrarily select a two-dimensional digital filter to be processed, and has the excellent advantage of being able to perform MTF correction optimally suited to the response of the input system using simple means. are doing.

[BRIEF DESCRIPTION OF THE DRAWINGS] FIG. 1 is a block diagram showing an example of the configuration of an image processing device for specifically implementing the image processing method according to the present invention, and FIG.
The figure is a block diagram showing a specific configuration example of the data switching control circuit and the data switching circuit in the same embodiment, FIG. 3 is a diagram showing an example of area specifying means for a grayscale image part in a document, and FIG. 4 is a diagram showing a scanner. 5 is a block diagram showing an example of the configuration of the control system and data processing system of the image reading system. FIG. 6 is a diagram showing a Laplacian filter according to input image information. FIG. 7 8 is a diagram showing addresses of a ROM in which data of MTF correction results by a plurality of two-dimensional digital filters is stored, and FIG. 8 is a diagram showing an example of the configuration of a threshold matrix used in the partial matrix method.
Fig. 9 is a diagram showing the principle of the partial matrix method, Fig. 10 is a block diagram showing a specific configuration example of the halftone processing circuit in the same embodiment, and Fig. 11 shows the addition timing in the halftone processing circuit. A time chart, FIG. 12 is a time chart showing the timing in the sub-scanning direction.
FIG. 13 is a diagram showing characteristics of the number of output gradations with respect to input data in the halftone processing circuit, and FIG. 14 is a diagram showing FROM addresses in the halftone processing circuit.

Claims (1)

[Claims] When performing MTF correction by two-dimensional digital filter processing of image information read by a scanner, multiple two-dimensional digital filters with various characteristics are prepared, and the two-dimensional digital filter to be processed is adjusted according to the filter specification. An image processing method that uses means to make selections.