JPH01109955A - Picture reader - Google Patents

Abstract

PURPOSE:To obtain a picture with high quality at a low illuminance by increasing the aperture area of a photodetector with low sensitivity and using a picture processing means so as to reduce the mismatching between the resolution of the photodetector and the resolution of other element output. CONSTITUTION:In selecting the aperture width of a CCD line sensor for blue color with the lowest sensitivity among the sensors for red,green and blue colors twice in the subscanning direction, for example, the sensitivity is doubled but the rate of fog is a multiple of 4/3. The output signal rom the red CCD is given to a large buffer 210 and the output signal from the green CCD is fetched in a small buffer 211, retarded and sent to a masking circuit 212. On the other hand, the output signal from the blue CCD is subjected to edge emphasis processing by an edge emphasis filter 213 and fed to a masking 212. Then the color mixture is eliminated and extracted both for the red and green pictures. Thus, the picture with high quality is obtained under a low illuminance.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an image reading device, and particularly to an image reading device using a plurality of photoelectric conversion elements such as a CCD line sensor.

[Conventional technology]

Conventionally, the image sensor used for image reading is C
Silicon crystal type such as OD, bipolar type and CDS,
There was a thin film type made of amorphous silicon, etc., and the optical system configuration was either a reduced type or a 1x type. There are two types of configurations of color image reading devices: a color separation method that uses a single image sensor and switches the light source or color filter, and a simultaneous reading color separation method that does not switch. As the simultaneous reading color separation method, there are two methods: one in which a plurality of image sensors for each separated color are arranged in parallel, and the other in which a stripe type filter is configured in the L-line image sensor and the color separation signals are read out in a time-division manner. Considering the essential performance of an image reading device, a thin film type with a high readout speed is suitable as a high-speed type, and a 1x type with a wide light-receiving area for the same reading and reading resolution is suitable as a high-sensitivity type. In the case of a color image reading device, a high-sensitivity type is required due to the reduction in the amount of incident light due to the color separation filter and the spectral sensitivity characteristics of the image sensor itself, and in order to achieve high-speed reading using a light source within the practical range. It is suitable to use a stripe filter constructed of a silicon crystal type filter of equal size. However, in the case of the silicon crystal type, it is difficult to make a long type that covers 297 mm of the longitudinal width of A4 paper with one chip due to manufacturing constraints, so it is difficult to make a long type with one chip that covers the longitudinal width of A4 paper. In recent years, devices with this structure have emerged for high-speed reading.

However, when multiple image sensors are connected to one in the main scanning direction, for example, blue (B), green (G), red (R
), the distance between each pixel is 1 / 16 mm X l / 3 = 2
Although it is 0.8 μm, the alignment accuracy between each image sensor is sufficient, and there is no problem with the positional accuracy of the read image, and even higher resolution is possible, but the image due to unevenness of characteristics between each image sensor Discrepancies in read density between sensor chips, especially color disparities between image sensor channels in the case of color, pose a problem. The causes of density deviation and color deviation between image sensor chips are l)
Examples include variations in sensitivity and dark current output between image sensor chips, and 2) variations in characteristics of signal processing circuits between chips or color filters.

On the other hand, by using a multiple line sensor in which multiple image sensors are arranged in the sub-scanning direction and each image sensor is coated with, for example, R, G, and B color separation filters, it is possible to eliminate the density deviation and color deviation between the image sensor chips. It is possible to make it smaller.

[Problem to be solved by the invention] However, even in the above conventional example, the sensitivity is different for each color, so the light amount of the light source is considered according to the element with low sensitivity, and the element of the color with high sensitivity is used for sensitivity correction. By installing a filter, the output level was leveled.

Furthermore, leveling of the output level has been achieved by increasing the amplifier gain of the color elements with low sensitivity and lowering the amplifier gain of the color elements with high sensitivity.

Moreover, it has been realized by incorporating both of the above two methods.

However, according to the first conventional method, the amount of light from the light source is wasted on highly sensitive elements, resulting in a waste of power and requiring a large amount of power consumption.

Furthermore, the power of the lamp for document illumination was high, which caused the problem of temperature rise inside the machine.

Further, according to the second conventional method, since the output level of the color element with low sensitivity is low, there is a drawback that a high-quality signal cannot be obtained, such as an inability to maintain S/N ratio.

Therefore, as a third method, a compromise between the first and second methods is actually used, but it is not a good solution.

[Means for solving problems]

The present invention increases the aperture area of the light receiving element with low sensitivity (by increasing the aperture area), obtains an image signal with a good S/N with a small amount of light, and also prevents mismatching of the resolution with that of the output of other elements by increasing the aperture area. The purpose of this is to reduce the amount of light by using an image processing means and obtain a high-quality image at low illuminance.

〔Example〕

FIG. 12 is a cross-sectional view of an image reading device representing an embodiment of the present invention, in which 101 is a document holder, 102 is a document, and 103 is a document holder.
is a halogen lamp for document illumination, 104 to 106 are mirrors, 107 is an imaging lens, lO8 is a COD for image reading,
109 is a motor for driving the optical system. Optical system 103.
104 sequentially scans the document in the X direction, and sequentially scans the document image in the C direction.
The image is formed on CD108.

Next, the image reading COD will be explained with reference to FIG. 1. 2 is a COD chip, and there are three independent line sensors 3. 4.5 is formed,
Each COD has a pixel pitch of 10μ and 5000 effective pixels. The intervals between each CCD are 180 μm. A red (R) filter is stained on CCDa, a green (G) filter is stained on C0D4, and a blue (B) filter is stained on CCD5.

FIG. 4 shows the spectral distribution of each color filter.

Moreover, FIG. 6 shows the spectral sensitivity characteristics of the COD excluding the filter section.

Further, FIG. 5 shows a spectrum distribution diagram of a halogen lamp. Now, as shown in FIG. 4, each color separation filter of the CCD 108 has a fairly high spectral transmittance even in a wavelength range exceeding 700 nm. In other words, it has a fairly high spectral transmittance even for near-infrared and infrared light, which is invisible light and has no relation to the human eye's ability to discern the hue of an image. Furthermore, as can be seen from the fact that the CCD 108 has a finite sensitivity up to a wavelength of about 11,000 nm, it also has a fairly high sensitivity to near-infrared and infrared light, which are the invisible lights. On the other hand, the halogen lamp used as the lamp 103 emits a considerable amount of near-infrared and infrared light, as shown in FIG.

Therefore, COD is also sensitive to near-infrared and infrared light, causing color noise. Therefore, in the embodiment shown in FIG. 12, a near-infrared removal filter (not shown) is installed in the optical path (for example, in or near the lens).
By inserting an infrared removal filter (not shown),
Prevents color noise from occurring. FIG. 7 shows the transmittance of the near-infrared cut filter, and FIG. 8 shows the transmittance of the infrared cut filter.

As explained above, halogen lamp spectral distribution Figure 5, near-infrared cut filter characteristics Figure 7, infrared cut filter characteristics Figure 8, color filter characteristics Figure 4, CO
By calculating the product of the D spectral distribution in FIG. 6, the output of each color is expressed.

In other words, the overall sensitivity characteristic is as shown in FIG. 9. Therefore, R:
The G:B sensitivity ratio is 5:4:1, and the blue sensitivity is extremely low.

Therefore, in this embodiment, as shown in FIG. 1, for example, the C0D5 aperture for B is doubled in the sub-scanning direction compared to the C0D3.4 aperture for R and G. The case where the sensitivity ratio of B is set to 5:4:2 as shown in FIG. 10 is shown.

Also, ND filters may be inserted into R and G (not shown), R
:G:B output ratio is set to 2.5:2:2 and extracted as a signal.

One pixel of s:ccp as shown in FIG. 11(a) has a certain aperture length in both the main scanning direction and the sub-scanning direction, and is indicated by lengths a and b, respectively, in the figure. When a COD with an aperture length of b in the sub-scanning direction moves the document by C in the sub-scanning direction and reads one pixel of the document, the HX(b+c) of the document is obtained as shown in FIG. 11(b). The area of 1
It will be read as pixels. Here, if the reading movement amount C in the sub-scanning direction is the scan length when reading at the same magnification, then
As shown in FIG. 11(C), an image read in a document area 81 in a certain scanning line is recorded by the printer using pixels Pi, and an image read in a document area S2 in the next scanning line by the same pixels is recorded by the printer. The image is recorded by a printer using a pixel P2, and each pixel P1 and P2 commonly includes a portion corresponding to the aperture area of the COD indicated by diagonal lines as a blur.

Here, the blur rate per pixel of the recorded image is b/(
b+c). Therefore, the aperture area of the COD is doubled in the sub-scanning direction. That is, if b=2a and c=a, the blur ratio will be 4/3 times.

Therefore, by doubling the aperture width of the CCD line sensor for B, which has the lowest sensitivity among R, G, and H, in the sub-scanning direction, the sensitivity could be doubled, but the blur rate was also increased by 4/3. Ta.

Here, the processing of the read signal will be further explained with reference to FIG. Each CCD line sensor 201, 202, 203 for R, G, B is connected to a timing signal generator 2-5.
is driven in the same phase by the output of the amplifier 204.
．． Amplify by 205.206. these amps 20
4 to 206 perform amplification in order to match the sensitivities of R, G, and B and for signal processing. The output signal of COD is approximately 0.
2V, which is multiplied by about 10 to 2■. Next, it is converted into 8-bit digital data using an A/D converter. Then R, G as shown in Figure 1.

B The sensors are spaced apart by 180 μm in the sub-scanning direction, and the reading interval of one pixel is 10 μm, so a delay of 18 lines occurs for each sensor.

Therefore, the output signal from the R COD is taken into the large buffer 210 consisting of 36 lines of memory, and the output signal from the G CCD is taken into the small buffer 211 consisting of 18 lines of memory and delayed.
12, where color correction processing is performed.

Since the output signal from the B COD has a larger aperture width than the other two CODs as described above, it is sent to the means for correcting blur, that is, the edge emphasis filter 213.

The edge enhancement filter 213 is a Laplacian filter using a three-line delay buffer, and as shown in FIG. 3, performs edge enhancement processing by performing quadratic differentiation in the sub-scanning direction. In other words, the delay buffer memory 302 for 3 lines addressed by the output of the address counter 301 that identifies the pixels for 1 line processes the current line's image signal 303.1 the previous image signal 304.2 the previous line's image signal. Image signals for three lines of 305 are output. These image signals are delayed pixel by pixel by the latch 306.

The pixel signals 313 and 314 of the lines before and after the pixel of interest in the same main scanning positional relationship are multiplied by (-1), for example, in the multiplier 315, and the result of doubling the pixel of interest signal 307 is added in the adder 316, and the pixel of interest is Second-order differential signal 3 in the sub-scanning direction for
Get 17.

By adding this second-order differential signal 317 to the pixel of interest in an adder 318, an edge-enhanced image signal is obtained. In other words, a signal with less blur can be obtained. The actual strength of edge enhancement is determined depending on the difference in aperture width. This data is sent to the masking circuit 212 in FIG. 2 to remove color turbidity due to filter characteristics, and can be taken out as data for each color to be recorded on the printer by the UCR and selector circuits.

Furthermore, although this embodiment has been described with reference to the case where the aperture width of the sensor differs in the sub-scanning direction, similar effects can be obtained by performing similar processing in the main-scanning direction as well.

[Other Examples]

In the above embodiment, it was shown that by increasing the CCD aperture area of blue, the blur caused by the edge enhancement can be reduced, but it is also possible to smooth the red and green signals and make them blurred to the same extent as blue. That is, smoothing filters 214 and 216 using 3-line delay buffers are installed in each R and G signal line as shown in FIG. The signal is multiplied by 1/4, for example, by a multiplier, and the result of multiplying the target pixel signal by 1/2 is added by an adder to obtain a signal smoothed in the sub-scanning direction for the target pixel.These signals are sent to the masking circuit 212. The data can be sent to the printer and taken out as data for each color to be recorded in the printer using the UCR and selector circuits.In FIG. 13, the same reference numerals are given to the same functional parts as in FIG.

Furthermore, as shown in Fig. 14, smoothing filters 214 and 216 are respectively inserted into the red and green signals, and the CCD
By inserting an edge enhancement circuit 213 into the signal from the blue signal with a large aperture area, the red and green signals are smoothed, the blur is removed from the blue signal, and the same level of resolution is obtained and color shift is reduced. It is also possible. In FIG. 14, functional parts similar to those in FIGS. 2 and 13 are given the same reference numerals.

Furthermore, changing the strength of edge enhancement according to the magnification in the sub-scanning direction is also effective in improving image quality.

For example, as shown in FIG. 11(a), if the aperture widths of the CCD in the main scanning direction and the sub-scanning direction are a and b, respectively, and the document reading movement amount in the sub-scanning direction is C, as described above,
The rate of blur per pixel of the recorded image is b/(b+c). Therefore, the recording magnification in the sub-scanning direction is set to 100% and 400%.
Comparing the blur ratio in (Figure 11(d)), it is 100
Assuming that the original reading movement amount at % is c=b, at 400%, c=l/4 b, so the blur ratio is 815 times higher at 400%.

Therefore, as shown in FIG. 15, the output of the rotary switch 320 for setting the sub-scanning magnification in percent units is input as an address, and the corresponding sub-scanning edge emphasis signal multiplication coefficient is output. For example, as shown in Figure 16, the magnification is 10.
Assuming that the multiplication coefficient at 0% is 1 and the coefficient at 800% magnification is 2, the coefficients from 100% to 800% are determined by connecting with a straight line.

Although this embodiment has been described using R, G, and B sensors as an example, it is of course applicable to cases where other color filters are used.

[Effects of the Invention] As described above, according to the present invention, it is possible to compensate for the lack of sensitivity by increasing the aperture area, and to obtain a high-quality video signal with a good S/N ratio at low illuminance. Further, resolution mismatching caused by increasing the aperture area can be prevented by applying filters with different frequency characteristics, making it possible to obtain a high quality video signal.

[Brief explanation of the drawing]

Fig. 1 is a diagram showing an example of a multi-line COD, Fig. 2 is a signal processing circuit diagram of this embodiment, Fig. 3 is a diagram showing an example of an edge emphasis filter circuit, and Fig. 4 is a diagram showing an example of an edge emphasis filter circuit.
The figure is a CCD filter characteristic diagram, Figure 5 is a lamp spectral characteristic diagram, Figure 6 is a CCD spectral sensitivity characteristic diagram, Figure 7 is a diagram showing the transmittance of a near-infrared cut filter, and Figure 8 is an infrared diagram. Figure 9 is a diagram showing the transmittance of the cut filter, Figure 9 is a diagram showing the overall sensitivity characteristic of the conventional example, Figure 10 is a diagram showing the overall sensitivity characteristic of the present example, Figure 11 is a diagram for explaining blur,

Claims (1)

[Claims] Image reading comprising: a plurality of line sensors having a plurality of photosensitive element rows with different aperture widths for reading image information; and filter means for filtering with a filter having different spatial frequency characteristics according to the aperture widths. Device.