JPS61196675A - Image signal processing method - Google Patents

Abstract

PURPOSE:To make sharpness of a processed picture optionally adjustable, by changing the filter coefficient and taking a means to change over a correcting guy, in performing an MTF correction by the Laplacian filter processing of a picture signal. CONSTITUTION:By the combination of 2 shift registers SR in the portion of each 1 line and 4 FF circuits, the picture element data of A-E in a picture element area of 3 x 3 from a picture signal BS are successively extracted, the full addition of corresponding data B-E is performed by the adders 11-13, multiplied result is multiplied by a coefficient K2 by a multiplier 14, and the extracted picture element A is multiplied by a coefficient K1 by a multiplier 15, and the output data of the multiplier is substracted from the output data of a multiplier 14 in a substractor 16. In such a way, a processed data A making it as object is obtained.

Description

[Detailed description of the invention]

The present invention relates to a method for processing an image signal obtained by reading an image of a document pixel by pixel using a scanner, and more particularly to an image signal processing method for performing MTF correction on the quantized digital image signal. In general, as pre-processing when binarizing the digital image signal of a document image, MTF deterioration in the optical system of the scanner, vibration during sub-scanning feed, and image signal transfer unevenness in the image sensor are So-called MT, which uses spatial filter processing to correct image blur caused by
I am trying to have F correction performed. Conventionally, when performing such MTF correction, a uniform MTF correction that satisfies a fixed correction value is performed, centering on deterioration correction in the optical system and image sensor of the scanner as described above.
It is not possible to arbitrarily change the sharpness of the processed image depending on the image condition of the original. The present invention has been made in consideration of the above points, and provides an image signal processing method in which the intensity of MTF correction can be switched in order to appropriately change the sharpness of a processed image. be. In order to achieve the object, the present invention employs a means to stepwise switch the gain of the MTF supplementary 15 by changing the filter coefficient when performing MTF correction by Labrasian filter processing. . Hereinafter, one embodiment of the present invention will be described in detail with reference to the accompanying drawings. Fig. 5 shows an example of the configuration of a cedar plate for reading a document image, in which a document 2 placed on a contactor 1, a glass 1, and a guide 3 is moved parallel to the document 2 in the sub-scanning direction at a speed of ~r. The first traveling body 4 moving along the

[1] The light source 5 is illuminated sequentially from the bottom in the main scanning direction, and the reflected light from the document surface is directed to the first mirror [;, which is also attached to the first traveling body 4], and the guide 3 at a speed of /2. The second traveling body 7 moving in the sub-scanning direction along A solid-state image sensor such as a CCD is arranged in one line in the main scanning direction, and is configured to be sent to an image sensor 10 with a length of 1t and sequentially read as an original image.In this case, the linear velocity V of the sub-scanning feed is 180 mm. mm/See, and the reading density is set to 400 pixels/inch in both the main scanning and sub-scanning directions.Therefore, the scanning time per pixel is approximately 70 nSec (
14.3MHz). The image sensor 10 has 5,000 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. , apparent processing time per pixel] 40 ns'
ec. FIG. 6 shows an example of the configuration of the control system and data processing system in the image reading system. Here, the interface 61 with a printer etc. varies depending on the system configuration, but the basic one is listed here. The commands C8 are divided into commands related to the start, startup, or abnormal state of the image reading system, and commands related to the selection of the data processing method. Most of the commands are input by an operator via an operation panel or the like. The command C8 is once given to the controller 62, and after checking the system status in the controller 1-roller 62, it is sent to each necessary location. Commands related to data processing include specifying document size, scaling factor, background noise removal, MTF correction on/off specification, and document mode/photo mode. This includes selecting and specifying the density position. In the configuration shown in FIG. 6, when the controller 62 receives a command from the scanner star 1-, it drives the servo motor M to cause the optical system to star 1- and starts reading the original image. At this time, the position and speed of the servo motor M are feedback-controlled using an encoder, so that the accuracy is maintained within ±2% with respect to a linear velocity of 180 mm/Sec. Further, 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. Also, at that time, the clock generator 63 supplies control clocks CL, K to the image sensor 10, video signal processing circuit 64, and image processing device 65! j=4- get. The image sensor 10 reads image information of a document line by line in the main scanning direction. that time. Scanner reading density is 400 pixels/inch, line speed is 18
With 0mm/Sec, the reading time per line is 35
It becomes 37zSec. Odd pixel column odd and even pixel column e' read out in time series from the image sensor 10 by clock CLK
Each data of ven is sent to the video signal processing circuit 64, where in the circuit configuration shown in FIG.
Data amplification is performed by P. At this time, since the maximum magnitude (white level) of the output signal of the image sensor 10 is 120 to 200 mV, it is amplified to nearly 1 V. Next, in the image signal output from the image sensor 10, that is, the image signal that has been video amplified and taken out through the current buffer CB2, as shown in the CB2 output waveform in FIG. Since there is noise and undulation of the image signal in the SuriSera 1 (in the figure, t1 indicates the reset noise period and t2 indicates the image signal period), it is provided via the MOS FETI and the Karen 1 to Buffer CB3. By sequentially sample-holding using MO8FE1"2 in the next stage, the reset noise and the undulation of the image signal are removed. In other words, the image signal output from the mask current 1-buffer CB2 is transferred to the previous stage MOS. is applied to the sample and hold switch of the FETI, where the MOS shown in FIG.
Switching is performed at the timing of the FETI drive signal r)1, the image signal is sampled during the image signal period t2, and the reset noise is removed by holding the reset 1-noise period t1. In FIG. 8, the image signal from which noise has been removed in reset 1 is shown as the CB3 output waveform. Next, by controlling the MOS FET2 in the later stage, switching is performed at the timing of the drive signal D1 of the MOS FET2 shown in Fig. 8, that is, at a timing such that the MO8FET2 in the later stage samples while the MOS FET1 in the former stage is holding. By doing this, the undulation components in both signals are removed. In Fig. 8, the image signal from which the waviness component has been removed is transferred from Karen 1 to Buffer CB.
It is shown as the output waveform of No. 4. In addition, the timing of each switching of MOS FETI and MOS FET2 is based on the previous stage MOS
The timing to turn off FETI is within the image signal period t2,
In addition, it is necessary to ensure that the on time does not overlap with that of the MOS FET 2 in the subsequent stage. Furthermore, depending on the performance of the sample-and-hold circuit actually used, sampling signals having completely opposite phases may be used. Also, if we use such a two-stage sample and hold means,
It is no longer necessary to make the sampling period particularly faster than the period of the image signal, and the time constant of the second-stage sample mode can be limited to smooth the transition between the image signals. FIG. 9 shows a specific configuration example for video amplification and sample hold processing of the image signal output from the image sensor 10. Here, the image signal output from the image sensor 10 is passed through an emitter follower circuit 91. After that, a sample hold circuit F for high speed is provided in two stages.
12 and 93 perform processing for removing reset noise and waviness components sequentially according to the sampling signal 5Pl-, and the processed image signal is sent to the video amplifier 94.
The signal is then amplified to a required level and fed to a zero clamp circuit 95 for creating a DC reference. In the zero clamp circuit 95, while the image sensor 10 is generating a dummy sensor output, the clamp switch SW is turned on by the zero clamp signal ZCL, and the zero clamp circuit 95 is clamped to 0. This zero clamp circuit 95 can be provided before the sample bold circuit 92, but in that case, the clamp switch SW must be turned on during the image signal period, and the subsequent circuits must be DC coupled. be. In the figure, CB is a Karen 1-buffer. Furthermore, in the video signal processing circuit 64, based on the image signal which has been shaped into a clean boxcar waveform by removing noise and waviness components and having been video amplified by sample and hold processing as described above,
It corrects shading caused by uneven illuminance of the light source in the scanner and removes the background texture of the original, and finally converts the image signal into a 6-bit digital signal by AD conversion and sends it to the image processing device 65. Note that background removal is turned on and off as appropriate by command C8. In this way, the reset noise and undulation components can be removed through sample and hold processing, and both signals can be shaped into a clean boxcar waveform.
Since the image signal has a long stable period during conversion, it is possible to take a longer acquisition time. Also, A.G.
It becomes possible to provide an accurate image signal to the peak hold circuit for applying C. Therefore, it is possible to obtain an accurate image signal with high accuracy in reading the original image. In addition, in the image processing device 65, as shown in FIG.
a first processing system consisting of an MTF correction circuit 101 and a binarization circuit 102, which perform optimal binarization processing for pattern images such as characters; Similarly, a second processing system consisting of a halftone processing circuit 103 that performs optimal binarization processing for grayscale images such as photographs of the input image signal B S is provided in parallel, and each processing system Controller 1-Roller 6 in the diagram
Output buffer circuit 104 temporarily stores and sequentially outputs data selectively processed in accordance with command CS issued from 2 (Here, R and A M for processing data of odd pixel columns and R and A M for product of processing data of odd pixel columns and RA for processing data
? , 4). The MTF correction circuit 101 corrects uneven transfer of image signals in the image sensor IO and MT in the optical system of the scanner (J).
This method uses a Laplacian filter to correct the blurring of micrometry that occurs in the high frequency range due to F deterioration and Ihj movement during sub-scanning feed. For example, the 3×
Figure 3 (
Each pixel data A- in the 3×3 pixel area shown in b)
By performing convolution integration with D, processing data A' of the pixel of interest located at the center of the 3×3 pixel area to be processed is obtained. A'=1・A-2 (B0C+D+E)...(
1) Here, 1 and 2 are coefficients in the filter shown in FIG. 3(a), and K1=3. 2=]/2. Also, if the DC gain of the Laplacian filter is made to be a constant n, the relationship K1=4XK2+n (K2≧0)−= (2) holds, and by changing 2 to the coefficient Kl,
As shown in FIG. 4, it becomes possible to change the spatial frequency characteristics. The present invention focuses on such points, and when performing MTF correction using a Laplacian filter, the filter coefficient KL
, and 2 to change the spatial frequency characteristics, the gain of the M TF correction, especially the high-frequency gain, can be switched stepwise to adjust the sharpness of the processed image. As is clear from the characteristics shown in Fig. 4, when the gain is small], if the aforementioned optical system and image sensor of the scanner perform deterioration correction, the MTF-corrected digital image signal will be filtered as a whole by a low-pass filter. By applying a high-frequency gain that exceeds the deterioration correction amount, it becomes possible to emphasize the sharpness of the processed image in the high-frequency region. FIG. 1 shows an example of the configuration of a ληTF correction circuit 101 for concretely implementing the present invention. Depending on the circuit configuration, the hardware is designed to be inline. That is, 2
3 from both signals BS by combining shift registers SR and four flip-flops FF for one line each.
The pixel data of A to E in the pixel area of The coefficient is multiplied by 2, the extracted pixel data A is multiplied by 1 by the multiplier 15, and the subtracter 16 converts the output data of the multiplier 14 to the multiplier 15.
By reducing the output data of , the target processed data A' is obtained. For example, each coefficient signal of 2 is given a value calculated according to equation (2) by an MPU that interprets the command. Further, FIG. 2 shows another configuration example of the MTF correction circuit 101, in which case the multipliers 14 and 15 in FIG.
, 1.8, and the data read out from the ROMs 17 and 18 are added by an adder ]9 to obtain processed data Δ'. Here, it is also possible to replace ROMs 17 and 18 and adder 19 with one ROM, but in that case, the address lines will be 7 + 3 + 6 = 16 trees, so the ROM capacity will be 655:36 bytes or more.
- In the configuration shown in Figure 2, ROM+7 is 102
4 by i, R(,) M 98 or 512 bytes, and compared to that, the R,OM capacity becomes too large to be practical. Table selection (+1 'r' s
is generated by a command, but ROM 17. I
8 may be replaced with RAM, respectively, so that the table can be converted by MPH. Tables also allow for nonlinear data that is not possible with multipliers. The MTF-corrected image signal is finally subjected to 91 pure binarization processing in the binarization circuit 102. The threshold level can be selected in several stages using the density position command 1-. FIG. 11 shows a specific configuration example of the binarization circuit 102, which includes a ROM+, 1] from which data of a predetermined threshold value is read out according to the density position signal P;
image (comparing No. a BS with its read threshold value)
It consists of a comparator 112 that converts into values. Here, the picture (
The fi number BS is data proportional to the light reflectance R of the document surface, and the relationship with the image density is R=, 10=, that is, a logarithmic relationship. Now, as data N of image signal BS, inverse 1]=J
When the reflectance is 1, it is 02, and when the reflectance is 0, it is 255, then N''
From 255-255 X R, N=255 255
X 10"", and its characteristics are as shown in FIG. 12. Since human vision is linear with density,
The selection of concentration positions must generate a threshold proportional to concentration. Therefore, ROM], 1.1 contains each density position Pa, , P2 . P3. A conversion table of data N of the image signal BS for ... is stored. Actually ROM
III is provided with a plurality of groups of conversion tables, and the conversion tables are used depending on the type of manuscript and the editing operation mode. A table with nonlinear data is also stored in the ROM11. It is also possible to have the MPU perform the same role as this ROM]11. Further, the halftone processing circuit 103 performs binarization processing of the image signal BS in consideration of gradation using a threshold matrix and partial averaging using a partial matrix method that combines the advantages of the dither method and the density pattern method. is realized. As described above, in the image signal processing method according to the present invention, when performing MTF correction by Laplacian filter processing of a digital image signal read by a scanner and quantized, MTF correction is performed by changing the filter coefficients. It uses a method to switch the gain.
The sharpness of the processed image can be arbitrarily adjusted by changing the strength of the MTF correction, which not only corrects deterioration caused by the scanner's optical system and image sensor, but also optimally corrects image blurring depending on the image condition of the original. It has the great advantage of being easy to use.

[Brief explanation of the drawing]

1 and 2 are block diagrams respectively showing configuration examples of an MTF correction circuit for concretely implementing the present invention, and FIG. Figure (b) is a diagram showing the data structure in a certain pixel area that is subject to MTF correction, Figure 4 is a diagram showing the spatial frequency characteristics, Figure 5 is a simplified diagram showing an example of the configuration of the scanner, and Figure 6 is A block diagram showing an example of the configuration of a control system and a data processing system in an image reading system, FIG. 7 is a block diagram showing an example of a circuit configuration of a video signal processing circuit in the same configuration example, and FIG. Figure 9 is an electrical wiring diagram showing the actual circuit configuration of the video signal processing circuit, and Figure 10 is a block diagram showing an example of the configuration of the image processing device in the same configuration example.
FIG. 1 is a block diagram showing a specific example of the configuration of the binarization circuit, and FIG. 12 is a diagram showing data characteristics of an image signal with respect to image density. 10... Image sensor 11, 12, 13. 19.
- Adder 14.15... Multiplier 16... Subtractor 17.18... ROM 56... Interface 62... Controller 63... Clock generator 64... Video signal processing circuit 65. Image processing equipment 101... M'r F correction circuit 102... Binarization circuit 103... Halftone processing circuit 104... Output buffer circuit

Claims (1)

[Claims] When performing MTF correction on a digital image signal that is optically scanned by a scanner and read pixel by pixel by an image sensor on a quantized digital image signal, a method is adopted to switch the gain of the MTF correction. Characteristic image signal processing method.