JPH0220957A - Image reader - Google Patents

Abstract

PURPOSE:To execute operation processing for compensating a picture quality in a wide frequency area by constituting a CCD picture element of a CCD linear image sensor means so that the light receiving sensitivity is varied stepwise or continuously in the sub-scanning direction. CONSTITUTION:A CCD picture element of a CCD linear image sensor means is constituted so that the light receiving sensitivity is varied stepwise or continuously in the sub-scanning direction. For instance, the sensitivity of a part of 7mum in the center of a photosensitive part of the CCD picture element is set to '1', and on the other hand, to respective parts of 3.5mum of both ends, an optical filter is vapor-deposited uniformly so that the sensitivity becomes 0.5, and the sensitivity is varied stepwise. That is, when a shape of the photosensitive part of the CCD picture element is changed so that width of the photosensitive part becomes narrower as it goes toward the end along the sub-scanning direction, a sensitivity distribution can be given to the CCD picture element, therefore, a frequency area which can be utilized effectively can be widened.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to an image reading device that reads images using a linear CCD image sensor.

(Prior Art) In an image reading device that reads an image of a document, a CCD linear image sensor, which is a photoelectric conversion element, is used. The photosensitive area of the photosensitive part of one pixel of the current CCD linear image sensor is, for example, 5 μs ×
7- (see FIG. 10. In this figure, the shaded area indicates the photosensitive area of each pixel. The size of one pixel is the same 7- in both the vertical and horizontal directions.) In order to obtain images with high resolution, the photosensitive area must be made small.

However, when the photosensitive area becomes smaller, the photosensitive area correspondingly decreases, resulting in an insufficient amount of light.

On the other hand, in order to read a 7Ji image using a CCD linear image sensor, it is necessary to illuminate the document, and a light source such as a halogen lamp or a fluorescent lamp is used for the illumination. Light reflected from the original by the illumination is detected by a CCD linear image sensor and converted into an electrical signal. Then, a digitized signal is obtained by A/D conversion.

(Problems to be Solved by the Invention) As explained above, the CCD linear image sensor has the following problems:
Only reflected light with a width of about 7 μs is received. However, when using a halogen lamp, the illumination light for the original can only be narrowed to about 2 to 3 city widths at most due to the position of the filament of the halogen lamp and the manufacturing precision of the reflector, and the illumination light of the original can only be narrowed to about 2 to 3 city widths. Light lighting is difficult. On the other hand, even when using a fluorescent lamp, the illumination is based on light emitted from the fluorescent material on the tube wall, so condensed illumination is not possible.

Therefore, the lighting efficiency of both halogen lamps and fluorescent lamps was very poor, and it was necessary to provide high-power illumination with large wattage lamps.

Therefore, in order to utilize illumination light efficiently, the applicant of the present application made the photosensitive area of the CCD pixel longer in the sub-scanning direction than that of conventional ones, and at the same time, in order to prevent the reduction in resolution in the sub-scanning direction that would occur due to this, the applicant 1 in the sub-scanning direction from the storage means
The line image data is read out and Fourier transformed, and the above Fourier transformed value is multiplied by the frequency characteristic of light reception determined by the shape of the photosensitive part of the CCD pixel of the CCD linear image sensor or the reciprocal of its approximate function. An image reading apparatus has already been proposed that includes an image quality compensation calculation means that compensates for the deterioration in the image quality of image data by performing inverse Fourier transform to obtain image reading data.

The image quality compensation calculation means of this image reading device will be described below.

In general, as shown in Figure 2(a), in a certain system f(y)
Suppose we input the input and get the output g(y). As shown in Fig. 2(b), when a pulse with no width (in terms of an image, a point brightness) is input, h(y)
If an output appears, this h(y) can be considered to indicate the characteristics of the system.

The frequency characteristics of these f(y), g(y), and h(y) are expressed as ω), it can be expressed as follows.

F(ω)=, f″f(y)exp(-iωy)dy
−(la)G(ω)=S″g(y)exp(-iωy)
dy −(lb)H(ω)=J: h(y) exp
(-iωy) dy - (IC) Using this frequency characteristic, the output G(ω) is the input F(ω) and the system characteristic H(ω
) is expressed as arithmetic.

G(ω)−F(ω)・H(ω) ・・・(
2) Inverse Fourier transform, which transforms based on frequency characteristics, also holds true at the same time.

f(y)=' f-: F(ω)exp(icc
+y)dω-C3a>g(y)=J-: G(ω)e
xp(iωy)dω-(3b)h(y)=2', j-
: H(ω)exp(iωy)dcc+ ・-(3a)
A general explanation of the above f, g, and h is given, for example, in Hiroshi Kubota, "Applied Optics" (Iwanami Zensho, 1970), page 211.

FIG. 3 shows an example of the shape of a pixel configured in the CCD linear image sensor 22. The shaded area is the photosensitive area 24 of each pixel.
, 24. ...is shown. The length b of the photosensitive section 24 in the X direction is made larger than the pitch a in the X direction (for example, b=
14#1m, a=7 μs, b/a=2
).

Note that the width c (=5 μs) of the photosensitive portion 24 is shorter than the pitch a in order to separate pixels.

By making the shape of the photosensitive section 24 longer in the sub-scanning direction,
The amount of light received increases.

However, the contrast and resolution in the sub-scanning direction are reduced,
Image quality deteriorates. Therefore, the following arithmetic processing is performed.

First, the response characteristics of the CCD linear image sensor 22 will be explained. Photosensitive section 24 of the CCD pixel that constitutes this sensor
The characteristic of the light intensity (2) should originally be read as a point, but due to the influence of the mechanical movement of the optical system and the shape of the photosensitive section 24, it becomes a rectangular shape as shown in FIG. That is, if the coordinate axis is set at the center for analysis, then b is the length of the photosensitive section 24 in the X direction.

To be precise, one pixel in the CCD is mechanically scanned even while receiving light. However, the time that one pixel operates and scans is about 0.1 μs, and the mechanical scanning speed for sub-scanning is usually about 100 mm / second, so during this 0.1 μs, 0. Only about 0L4a++ is scanned. This length is sufficiently small compared to the element size 14-, so it is ignored here.

In the theory of the response of the general system mentioned above, the rectangular function 1
(y) corresponds to h(y). Therefore, by Fourier transformation of I (y), the function H(ω
) is obtained. That is, H(ω)=sinπω/πω...(
5) Here, ω=b−N.

Here, b is the actual length in m, and N is the spatial frequency, expressed in units of (hon/gull). FIG. 5 shows the frequency characteristics of the function H(ω). Naturally, the contrast in areas where the frequency is high, that is, in fine areas, is reduced.

The function H(ω) will be specifically explained by applying the actual spatial frequency N. Figure 6 shows a case where the length b of the photosensitive section in the sub-scanning direction is 14 μs (curve A) and a case where the length b in the sub-scanning direction is 7-μs (curve B).
) shows the frequency characteristics of H(N). When ω=b−N=1, H(ω) becomes 0. The value of N at this time is b=14
When um, it is 71 lines/aIm, and when b=7-, it is 143 lines/ffll11. Incidentally, a fine image reading device normally has a reading density of 400 dots per inch. Expressing this in terms of spatial frequency N, 8 lines/
I! It is I11. If the number of pixels of the CCD linear image sensor is 5000 and the pixel pitch is 7 pixels, the photosensitive size of the CCD linear image sensor is 35 m+n.

When A3 size 297 mm is reduced and projected onto this CCD IJ near image sensor, it is 297 8.5, and the spatial frequency of 8 lines/ffll11 is 68 lines/1
Ilffl, so the spatial frequency of the CCD pixel with b = 14 μs is 71 lines/ff1I+, which means that although this CCD pixel can theoretically achieve a reading density of 400 dots/inch, various factors If variations occur due to this, it may not be possible to achieve this reading density, and it goes without saying that it is desirable to have more margin in the reading density.

When the size of b (=7μs) is doubled, from the formula (5) obtained by Fourier transforming the formula (4) expressing the rectangular light intensity distribution characteristics, 5in(π・2b-N)/π・2b-N/(sin(π
・b−N)/π・b−N)=cos(π・b−N)=cos(0,022N), and the value of the frequency characteristic for the same N is multiplied by cos(0°022N). Therefore, by simply multiplying its reciprocal, 1/cos(0,022N), by the expression for the frequency characteristic when b=147a (represented by curve A in Figure 6), The formula for the frequency characteristic (represented by curve B in FIG. 6) is: By increasing the size of the photosensitive area in the sub-scanning direction, the light-receiving area increases and the sensitivity increases, but normally the resolution in the sub-scanning direction decreases, but by performing the above calculation processing, the resolution decreases. It can be prevented.

In the above explanation, a rectangular light intensity characteristic is used, but in general, a function H(ω) that expresses the frequency characteristic of the system is used to calculate the input value f( Find y). This restores the frequency characteristics that were degraded by enlarging the photosensitive area of the CCD. At this time, the following formula is used.

G(ω)X(1/H(ω)) =F(ω)・H(ω)・(1/H(ω))=F(ω)
...(6) That is, the output value g(y)
Find G(ω) by the Fourier transform of (formula (lb)),
Next, calculate F(ω) by multiplying by the reciprocal of H(ω) (Equation (6)), then perform inverse Fourier transform of F(ω) (Equation (3a)), and convert the original image input f( y) is calculated.

However, as is clear from equation (6), the above improvement in frequency characteristics is effective only when H(ω)≠0. That is, in the method of performing image quality compensation by performing the above-mentioned arithmetic processing, if the frequency characteristic becomes 0, its reciprocal cannot be obtained and it becomes impossible to perform the inverse Fourier transform processing.

Further, even when calculating the reciprocal when the frequency characteristic is close to 0, the calculation error becomes large, so even if the correct value is calculated, the original image input cannot be reproduced correctly. In other words,
The above method is limited to a frequency region in which the frequency characteristic does not become O.

An object of the present invention is to expand the frequency domain so that arithmetic processing for image quality compensation can be performed in a wider frequency domain.

(Means for Solving the Problems) An image reading device according to the present invention illuminates a document, projects an image of the document onto a CCD linear image sensor means, and a CCD linear image sensor.
In an image reading device that reads an image by mechanically scanning an original in a sub-scanning direction perpendicular to the main scanning direction in which the CCD pixels of the linear image sensor means are lined up, the CCD linear image sensor means is It consists of a CCD pixel having a long photosensitive area in the scanning direction, and includes an A/D conversion means for digitizing the electric signal detected by each CCD pixel of the CCD linear image sensor means into image data, and a CCD for converting the image data into image data. The CCD further comprises image quality compensation calculation means for performing calculation processing based on the frequency characteristic of light reception determined by the shape of the photosensitive part of the pixel or the reciprocal of its approximation function to compensate for the image quality deterioration of the image data,
The CCD pixels of the linear image sensor means are characterized in that their light-receiving sensitivity changes stepwise or continuously in the sub-scanning direction.

(Function) When the light receiving sensitivity of the CCD pixels of the CCD linear image sensor is changed in the sub-scanning direction, the frequency characteristics of the system are extended to the high frequency side.

(Example) Hereinafter, the present invention will be described in detail with reference to the accompanying drawings.

This example utilizes the phenomenon that when a sensitivity distribution is given to the photosensitive area of a CCD pixel such that the sensitivity decreases as it goes from the center to the edge along the sub-scanning direction, the characteristics at high frequencies improve. to expand the frequency range. here,
Examples of how to provide the sensitivity distribution include using an optical filter and depending on the shape of the photosensitive portion, but the present invention is not limited to these methods.

FIGS. 11 and 12 are graphs showing the light intensity characteristics of the photosensitive portion of the CCD pixel when a sensitivity distribution is provided by an optical filter.

In FIG. 11, the sensitivity of the central portion 71a+ of the photosensitive area of the CCD pixel shown in FIG. 3 is set to 1, while the sensitivity at both ends is 3.5
This is the characteristic of light intensity I when an optical filter is uniformly deposited on each JJI part so that the sensitivity becomes 0.5, and the sensitivity is changed stepwise.

Further, FIG. 12 shows the characteristics of the light intensity ■ when the sensitivity distribution is Gaussian distribution (normal distribution) so that both ends of the photosensitive part of the CCD pixel shown in FIG. 3 are 2σ. here,
σ is the standard deviation of the Gaussian distribution. In this case, an optical filter is deposited over the entire surface of the photosensitive area, and the film thickness is adjusted by linear control to give a Gaussian distribution.

FIG. 13 is a graph showing the frequency characteristic H(N) obtained by Fourier transforming the expression expressing the sensitivity distribution shown in FIGS. 11 and 12. As shown in the figure, the value of the spatial frequency (N) for which H(N) = 0 is 95 lines/m in the case of Figure 11.
m, it can be seen that in the case of Fig. 12, it is 115 lines/M. As mentioned above, in the conventional CCD pixel in which the sensitivity distribution shown in Fig. 3 is uniform, that is, the light intensity characteristic is rectangular, the spatial frequency is significantly improved compared to 71 lines/M. There is.

The above optical filter is made by laminating a high refractive material and a low refractive material on a glass substrate (BK7, white glass), and the high refractive material is titanium oxide (TiO=
), zinc oxide (Z no-), cerium oxide (Ce
nt), hafnium oxide (Hf'o t'), zinc sulfide (ZnS), etc.; low refractive materials include silicon oxide (SiO=), magnesium fluoride (MgF), cryolite (Na3A I F g) etc.

Next, examples of giving the sensitivity distribution depending on the shape of the photosensitive part of the CCD pixel are shown in FIGS. 15(a), (b) and 16(a).
), shown in (b).

FIG. 15(a) shows a CCD sensor 22 in which the size of one pixel is 107a×104, and the shape of the photosensitive portion is an ellipse with the major axis in the sub-scanning direction. The light intensity characteristics of this CCD pixel are shown in FIG. 15(b).

The CCD sensor 22 shown in FIG. 16(a) also has a pixel size of 10/JIXIO1a and a hexagonal photosensitive portion. FIG. 16(b) shows the light intensity characteristics.

In Fig. 14, C of Fig. 15(a) and Fig. 16(a) is shown.
For comparison with CD pixels, the same pixel size (10u
Ifl×10 prisoners) is shown below.

FIG. 17 is a graph showing the frequency characteristics of the CCD sensors shown in FIGS. 14, 15, and 16. Line E,
F and G are the frequency characteristics of the CCD sensor shown in FIGS. 14, 15, and 16, respectively. CC of Figure 14
It can be seen that the frequency range of the D sensor is approximately 143 lines/M, whereas the frequency range of both the CCD sensors in FIGS. 15 and 16 is 150 lines/M or more.

As described above, by changing the shape of the photosensitive area of the CCD pixel, specifically, by making the width of the photosensitive area narrower toward the edges along the sub-scanning direction, a sensitivity distribution can be given to the CCD pixel. I can do it. As a result, the frequency characteristics improve, and the frequency range that can be effectively used can be expanded.

CCD 'J consisting of CCD pixels with the above sensitivity distribution
The near image sensor 22 is used and the processing procedure shown in FIG. 7 is used to compensate for the deterioration in image quality of image data.

First, all images of a document are sequentially scanned by an optical system (step S1), and two-dimensional information of one page's worth of images is stored in an image memory (step S2).

Next, X is specified (step S3), read data g(y) on one line in the sub-scanning direction is read out from the image memory (step S4), and G(ω) is obtained by Fourier transformation (step S35).

Next, the reciprocal (1/H
(ω)) is multiplied by G(ω) to obtain F(ω) (step S6).

Next, image data f(
y) and store it in the image memory (Ste knob S7)
.

Next, if it is determined that it is not the final value of X (step S8)
, returns to step S3, sequentially changes X, and performs the same process for the lines in the sub-scanning direction.

When all X processing is completed, f(y) is output from the image memory as an image reading signal (step S9).

Thereafter, processing such as binarization is performed on this image reading signal, and the signal is output to an external printer or the like.

Even if the shape of the CCD pixel is not a rectangle but a parallelogram to avoid a moiré pattern, the same calculation process can be performed as long as the frequency of the photosensitive section determined by the shape of the CCD pixel is determined.

By performing processing such as multiplying by this reciprocal, the characteristics of the signal read by the CCD with the original small photosensitive area can be further improved.

It goes without saying that to improve the characteristics, it is effective to perform an operation that approximately increases the high frequency value, without multiplying it by its reciprocal (1/H(ω)).

FIG. 8 shows a schematic arrangement of a mirror scanning type optical system, which is an example of an image reading device. The light emitted from the halogen lamp 2 illuminates the original 10 placed on the original glass via the reflecting mirror 4.6. The image of the original is mirrored 12.14.1
6, passes through a lens 18, passes through a filter 20, and forms an image on a CCD linear image sensor 22.

The halogen lamp 2, the reflector 4.6 and the mirror 12 are
The mirrors 14 .
16 also moves at half the speed to keep the optical path length up to lens 9 constant. The CCD linear image sensor 22 is
By adding an optical filter to what is shown in Figure 3, Figures 11 and 1
The one with the sensitivity distribution shown in Figure 2, and the one with the 15th
It has a shape as shown in FIG. 16(a) and FIG. 16(a).

FIG. 9 shows a method of reading the entire image by redrawing the configuration of FIG. 8 as a model. In the sub-scanning direction, optical systems 2, 4,
6.12.14.16 refers to the direction in which the document 10 is scanned by movement. The CCD linear image sensor 22 outputs signals of each pixel in the order perpendicular to the sub-scanning direction (referred to as the main-scanning direction). 10 originals in both main scanning and sub-scanning.
Read the entire image. Hereinafter, the main scanning direction will be referred to as the X direction, and the sub-scanning direction will be referred to as the X direction.

Note that mechanical sub-scanning methods include moving the original while illuminating it in the main scanning direction as shown in Figure 9, or projecting the entire screen of the original using a lens and then using a CCD. There is a method to move the linear image sensor.

FIG. 1 is a block diagram of a circuit of an image reading device.

The clock generation circuit 40 provides a necessary sample and hold signal to the CCD linear image sensor 22, and is also connected to a CPU 42 that controls document reading and is used as a clock signal. CCD linear image sensor 22
converts an optical signal into an electrical signal. The A/D converter 44 converts the analog output of the CCD linear image sensor 22 into a digital signal. The shading circuit 46 corrects the unevenness of light amount in the main scanning direction and the CCD with respect to this digital signal.
This is for correcting variations between bits (pixels) of the linear image sensor 22, and its timing is given by a shading signal from the CPU 42. The output of the shading circuit 46 is input to an image memory 48. Image memory 48 is a RAM that stores all read data regarding document 10.

When the read data for one page is stored in the image memory 48, the Fourier transformer 50 first converts the data in the sub-scanning direction (X direction).
The read data of one line of
) to Fourier transform. Next, the multiplier 52 converts the reciprocal 1/H(ω) of the frequency characteristic of the system response to G(ω
) to obtain F(ω). This 1/H(ω) is CC
Photosensitive section 24 of CCD pixel of D linear image sensor 22
Although it is given by the length b in the sub-scanning direction and the scanning frequency N in the sub-scanning direction (see equation (5)), an appropriate approximation function may be used. Next, the inverse Fourier transformer 54 receives F(ω) for all frequencies, performs inverse Fourier transform using equation (3a), and sends it to the image memory 56. The above calculation process is repeated at each position y in the sub-scanning direction. In this way, one page's worth of arithmetic-processed image read data is stored in the image memory 56.

In the above description, two image memories 48 and 56 are used for convenience of explanation, but in reality, only one image memory can be used for calculation processing. Note that the CPU 42 is
Binary or dither attributes are determined for each predetermined area based on the information written in the image memory 56, and the attributes are stored in the attribute RAM 5.
Write in 8.

Next, binarization processing is performed in the same manner as before, and the result is output to the outside. That is, the above correction calculation read data is sequentially sent from the image memory 56 to the comparison circuit 60.

The pattern generation circuit 62 generates threshold values during dither selection, and the threshold values are generated in an (m×n) matrix. The selector 64 switches the threshold between binary and dither based on the information in the attribute RAM 58 during data transfer, and outputs the threshold to the comparison circuit 60. The comparison circuit 60 is
The corrected read data is compared with the threshold value sent from the selector 64, and the result is outputted in one bit to the output circuit 66.
Output to. The output circuit 66 outputs 1 bit when an effective image signal (synchronization signal) is sent from the CPU 42.

Outputs the image signal and synchronization signal to an external device (such as a printer).

Note that the CPU 42 inputs a motor signal, a lamp signal, a fixed position signal, a command signal, etc., and controls the operation of the image reading device.

In addition, in this embodiment, since the reading device that outputs image data to a printer or the like has been explained as an example, the image data that has been A/D converted is configured to be binarized and output. In some image reading apparatuses, each pixel may be output as a multivalued image, and the present invention can be applied to this case as well.

(Effects of the Invention) According to the present invention, in order to effectively utilize illumination light, the size of the photosensitive portion is increased in the sub-scanning direction, and the reading density in the sub-scanning direction is improved. In an image reading device that performs image compensation calculation processing for image compensation, the effective frequency range in which image compensation calculation processing can be performed accurately can be expanded, and sufficient reading density can be ensured.

[Brief explanation of the drawing]

Figure 1 is a block diagram of the circuit system of the image reading device, and Figure 2 (
a) and (b) are diagrams showing the response of the system, Figure 3 is a diagram of the photosensitive part of the CCD linear image sensor, and Figure 4 is C
A graph of the response of a CD pixel, Figures 5 and 6 are graphs of the frequency characteristics of the response of the system, Figure 7 is a diagram showing the flow of arithmetic processing, Figure 8 is a diagram of the optical system of the image reading device, 9th
The figure shows the scanning of the entire image, and Figure 10 is a conventional CCD.
A diagram of the photosensitive part of an example of a linear image sensor, Figures 11 and 12 are graphs of the response of CCD pixels given the sensitivity distributions, respectively, and Figure 13 is a graph of the response of the CCD pixels given the sensitivity distributions shown in Figures 11 and 12. Fig. 14 is a graph of the frequency characteristics of the response of the CCD pixel shown in Fig. 10, and Fig. 15 (a
) is a diagram of a CCD linear image sensor having an elliptical photosensitive part shape, and FIG. 15(b) is a diagram of the CC of FIG. 15(a).
Graph of pixel response of D linear image sensor, 16th
Figure 16(a) is a diagram of a CCD linear image sensor having a hexagonal photosensitive part shape, and Figure 16(b) is a diagram of Figure 16(a).
A graph of the pixel response of the CCD linear image sensor,
Figure 17 shows the C shown in Figures 14, 15 and 16.
It is a graph of the frequency characteristic of the response of the pixel of CD linear image sensor. 22...CCD linear image sensor, 42...C
PU, 48... Image memory, 50... Fourier transformer, 52... Multiplier, 54... Fourier inverse transformer, 56... Image memory. Patent applicant Minolta Camera Co., Ltd. Agent
Patent attorney Aobai Ao and 1 other person Evening 5v 1-11u+1 Fig. 6 lN1 1゜2゜4゜5゜ (book/mml Fig. 7

Claims (1)

[Claims] (1) The original is illuminated, the image of the original is projected onto the CCD linear image sensor means, and the original is mechanically scanned in the sub-scanning direction perpendicular to the main scanning direction in which the CCD pixels of the CCD linear image sensor are arranged. In an image reading device that reads an image using a CCD linear image sensor, the CCD linear image sensor means has a CCD linear image sensor having a photosensitive section that is longer in the sub-scanning direction than in the main scanning direction.
It consists of CD pixels, and includes an A/D conversion means that digitizes the electrical signals detected by each CCD pixel of the CCD linear image sensor means into image data; The CCD linear image sensor means further includes image quality compensation calculation means for performing calculation processing based on the frequency characteristic of light reception or the reciprocal of its approximation function and compensating for the deterioration in image quality of image data, and the CCD pixels of the CCD linear image sensor means have a light reception sensitivity of An image reading device characterized in that it is configured to change stepwise or continuously in a sub-scanning direction.