JPS61224660A - Picture reader - Google Patents

Abstract

PURPOSE:To obtain reproduction of an original picture with fidelity even when the length of joint is the width of plural picture elements by restoring smoothly the missing of picture data generated at the joint of solid-state image pickup elements. CONSTITUTION:A proportional sharing density operating section 102 operated the density gradient at the 1st picture element and a proportional sharing density operating section 103 operates the density gradient at the 2nd picture element length. A timing generating means 10 generates a timing 113 corresponding to the joint and generates timings 108, 109 sharing the gap of joint inversely proportional to the 1st and 2nd picture element lengths into two. Thus, the sections 102, 103 output picture element densities 110, 111 proportional to the picture element position in the gap respectively, a selection means 104 selects either of the picture element densities 110, 111 by the picture element location divided into two in the timing 112 and outputs a picture data 114. Further, a selection means 105 selects the picture data in the timing 113 and selects an output of a delay circuit 100 in case other than the timing 113 to form a picture signal 115.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of Application in the Rice Noodle Industry] The present invention relates to an image reading device having a line sensor formed by connecting a plurality of solid-state imaging devices such as CCDs in a substantially horizontal direction.

[Summary of the Disclosure] This specification and drawings describe an image reading device having an image input sensor formed by connecting a plurality of image sensors with a gap of one pixel or more, in which one or more pixel lengths adjacent to each other before and after the joint are Focusing on two effective pixel sections with The missing image data that occurs at the joint is interpolated by using the corresponding proportional density as interpolated pixel data, and using the proportional density corresponding to the density gradient of the effective pixel section adjacent to the rear part of the gap as interpolated pixel data for the latter half of the gap. An image reading device that faithfully reproduces an original image is disclosed.

Conventionally, in order to make line sensors longer, there has been an effective means for determining the signal level of data for connection purposes, which corresponds to the gap between line sensors when short sensors are arranged in series in the horizontal direction. There wasn't. Also, as pre-processing for image signal processing,
There has not been a particular method for interpolating image signals with multilevel gradation levels for the purpose of joining, and a method has also been proposed in which the joining is approximated from the density gradient of the preceding and following pixels, but such methods There is no problem when it is a gradation image, but it is not effective for restoring when the original image is a line drawing of characters etc. and the thin vertical lines are used for this connection purpose, so the reproduced image is a lost image compared to the original image. was occurring.

[Problems to be solved by the invention E] The present invention was made in view of the above-mentioned drawbacks of the prior art.
An image reading device with excellent image reproducibility is provided.

[Means for solving the problem] The configuration of the basic embodiment for achieving the above problem is described in the first example.
In the figure, 100 and 101 are delay circuits, 102 and 103
is the delay circuit 100. 104 and 105 are image signal selection means, respectively, and 106 is a timing generation means for generating timing.

"Operation" In the embodiment with the above configuration, the delay circuits 100 and 101
are $1 each. It is assumed that the second pixel length is the second pixel length. The proportional distribution concentration calculation unit 102 calculates the concentration gradient using the first pixel length,
The proportional distribution density calculation unit 03 calculates the density gradient using the second pixel length. The timing generating means lO6 generates a timing 113 corresponding to the seam and a timing 108.1 obtained by dividing the gap between the seams into two by inversely distributing the gap between the first and second pixel lengths.
09 is generated. Therefore, the proportional distribution concentration calculation section 102,
103 outputs pixel densities 110 and 111 that are proportional to the pixel position within the gap, and the timing 112 corresponds to the above 2.
The selection means 104 selects either pixel density 110 or 111 at the divided pixel position, and image data 114 is output. Furthermore, the selection means 105 selects the image data when the timing 113 is reached, and selects the output of the delay circuit 100 as the image signal 115 when the timing is other than the timing 113.

[Examples] Hereinafter, more specific examples will be described according to the drawings. FIG. 5(a) shows the structure of a line sensor used in the image reading device of the embodiment. The line sensor 20 is a chip-shaped solid-state image sensor 2 made of, for example, COD.
0-a, 20-b, and 20-c, and the gap 21 between each solid-state image sensor has a width of m pixels. The image signal obtained while scanning such a line sensor 20 in the main/sub-scanning direction is
As shown in Figure (b), M image signals can be obtained, but of course the image signals of m pixels corresponding to the joints are lost. Therefore, as shown in FIG. 5(C), it is necessary to interpolate pixels (C+ to C-) corresponding to the gap between the image pickup elements.

A second circuit is used to restore (interpolate) the lost image data.
In FIG. 2, the counter l is connected to the signal RCLK.
A counter whose clock input is horizontal synchronization signal HSY
After being cleared by NC, counting starts. The counter l counts the number of pixels of the image read by the line sensor 20, and at the end of the effective pixels of one chip of the line sensor 20, the output CYx is set to ``1'' for one pixel, that is, 1 for each gap. There is an "o" between pixels.

At the falling edge of the signal CY, the gap section signal generation circuit 2
It outputs "0"' during a period corresponding to the gap between the line sensors. In addition, the gap 172 section signal generation circuit 3 generates a signal C.
At the falling edge of Yz, O'' is output for an interval of 1/2 the number of pixels of the width of the gap.

The output CY2 of the gap section signal generation circuit 2 is input to the AND gate circuit 10. Image transfer lock ICLK is input to the AND gate circuit 10, and RCLK is output only when CY2 is 1''. RCLK is input as the clock of counter l, so the pixel of interest is E3 +
When the process proceeds as E2 + El and reaches the timing corresponding to one gap, RCL K becomes "o" and the counter l stops. Their timing chart is shown in FIG.

The clock signal RCLK is supplied to the shift calculation circuit 4, 5゜6.7
is also entered. The counter 1 and the line sensor 20 are positioned so that the output Z of the shift calculation circuit 6 becomes the "target pixel °". That is, one image data IMAGE is sequentially shifted in the shift calculation circuits 4, 5, 6.7, but when the pixel of interest reaches the timing of the joint (CY
2 is "o"), RCLK becomes "O" and transfer (shift) is stopped.

Image data IMAGE is transferred to shift calculation circuits 4, 5, 6°7
The data is sequentially transferred to , and subjected to simple arithmetic processing as described below.
A3 from the Y terminals of shift calculation circuits 4, 5, 6.7, respectively.
, A2, At, and Ao are output.

The internal configuration of the shift calculation circuits 4, 5, 8.7 is shown in Figure 4 (&
) as shown. Shift calculation circuit 4, 5, 6
．． Each of the shift registers 13-1 to 13-R is composed of R shift registers 13-1 to 13-R and an averaging circuit 16, and input data is sent to the shift registers 13-1. ...RCLK sequentially up to 13-R
It is transferred in synchronization with and output from the X terminal. That is,
What is output from the X terminal is image data delayed by R shift registers. In addition, the input image data is delayed only by the shift register 13-1 in the shift operation circuit.
Output from the terminal.

Shift registers 13-1 to 13-R in the shift operation circuit
The respective outputs are input to the adder circuit 14, and further multiplied by 1/R by the 1/R multiplication circuit 15, and the average value of the outputs up to the shift registers 13-1-13-R is sent to the Y terminal. Output. This Y output is the shift calculation circuit 4, 5, 6
．． 7 respective outputs A3.

Corresponds to A2, AI, and Ao.

A3 is the average value of each shift calculation circuit.

A2, AI, and Ao are input to an arithmetic circuit 9.8 in order to calculate a data value of interpolated data. The output of the arithmetic circuit 8.9 is input to the selector 11. Selector 1
1, the output CY3 of the interpolation 1/2 interval signal generation circuit 3 is input as a select input, and when CY3 is °°O'', the output value of the ° calculation circuit 8 is selected and output.On the other hand, CY3 is
When ゛l°゛, the output of the arithmetic circuit 9 is selected. These b forces are the correction values of the gap interpolation data. Arithmetic circuit 8.
The operation of 9 will be explained later.

Incidentally, the pixel of interest is normally output from the Z terminal of the shift calculation circuit 6 described above. This pixel of interest is selector 1
2, and this output is selected by the selector 12 while the signal CY2 of the gap section signal generating circuit 2 is at °°1°. When the signal CY2 is °O''', that is, at the timing of a gap, the selector 12 selects and outputs the output of the selector 11, which is the correction value. Image data calculated from previous and subsequent effective pixel data is supplemented as interpolation data.

Incidentally, in the above embodiment, the correction data value is corrected according to the gradient of the data level of the image data before and after the gap, and the calculation of the correction value is performed by the arithmetic circuit 8.
．． 9 and selector 11. Below, we will briefly discuss this correction.

The gap is 01 to C as shown in Figure 3 or Figure 5(e).
The number of shift registers in the shift operation circuit 4゜5.6.7 is Ra, R5, Re respectively.
, 87, the operation in the arithmetic circuit 8 is Cn = At + [2(At - Ao)l(Rs
+1)/2+nl]/ (Re + R7)...Formula ■ (However, n=1...m/2) The calculation in the arithmetic circuit 9 is Cn-A2 + [2(A2-A3)l(R5+1
)/2+m-n)]/ (R4+ Rs)...Formula ■ (However, n = ts/2+1.-...ta) Ao,
AI, A2, and A3 are shift calculation circuits 7
This is the respective average value data output from the Y terminal of ゜6.5.4. As shown in FIG. 6, this is the average value of 87, 26, R5, and 84 image data before and after the m-pixel section of the gap.

As shown in FIG. 6, the first term in equation (2) above is the data level of the reference point, and the second term is the difference between one pixel and the adjacent pixel level, and the slope is 2 (AI - Ao )/(Re
+R7) or 2(A2-A3)/CRa +R
5) and the distance from the gap to the reference point, i.e. (Rs +1
)/2+n or (Rs +1)/2+m-n, and is generally equivalent to the straight line equation where y=b+ax. Therefore, in this embodiment, the image data level at the position of the gap can be estimated and determined according to the data gradients before and after the gap. Also, in order to avoid image data fluctuations such as noise, the slope of the data is set to R7, Re, R5.
, R4 is the number of pixels.

Next, a modification of the calculation of this correction value will be described.

When R in FIG. 4(a) is set to R=1, that is, when the number of shift registers in the shift calculation circuit is one, FIG. 4(b)
) is a simplified circuit. Then, R
7=Re=R5=When R4=1, the image data level of the gap is calculated by calculating the gradient of the data level from two pixel data at the front and rear, and the gap data is obtained.

In addition, as mentioned above, the calculation results of the calculation circuits 8 and 9 are determined by the gap 17.
It is switched by the selector 11 under the control of the output CY3 of the two-section signal generating circuit 3. As a result, the result of calculation using the image data of the sensor chip that is half closer to each other among the m pixels in the gap is used as the correction data.

As explained above, according to this embodiment, even if the sensor is composed of several chips and the gap between the joints is wide, it is possible to easily use a long sensor with a gap by adjusting the number of corrections.

Further, since the slope for correction and the cheater level of the reference point are calculated by calculating the average value of a plurality of pixels, it is less susceptible to the influence of noise and the like.

Furthermore, the density gradient of the effective image data before and after the joint part,
Alternatively, since the image data of the joint part is predicted and determined based on the gradient of the illuminance level, even if a thin line or the like overlaps the joint part, it is possible to restore it to some extent depending on the surrounding image data level.

[Effect of the Invention J As explained above, according to the present invention, by smoothly restoring the loss of image data that occurs at the joints of solid-state image sensors, faithful reproduction can be achieved even if the length of the joint is multiple pixels wide. The original image can be reproduced, which is particularly effective when the original image is a line drawing or the like.

[Brief explanation of drawings]

FIG. 1 is a basic configuration diagram of the embodiment, and FIG. 2 is a circuit block diagram of the embodiment. FIG. 3 is a timing chart of the example circuit. Figure 4(a) is a detailed diagram of the shift calculation circuit, Figure 4(b)
is an example circuit diagram of a simplified shift calculation circuit, Figure 5(a) is a structural diagram of a COD line sensor, and Figure 5(a) is a structural diagram of a COD line sensor.
b) to (C) are diagrams showing the interpolation of the gaps between joints, and Figure 6 is a virtual diagram when the gaps are interpolated with data. A3゜In the diagram, l... Counter, 2... Gap section signal generation circuit, 3... Gap 1/2 section signal generation circuit, 4゜5
．． 6.7... Shift operation circuit, 8.9... Operation circuit, 11.12... Selector, 13-1. ...13-
R...Shift register, 14...Addition circuit, 15.
・Bow/R times multiplication circuit, 16...Averaging circuit, 17...
- Shift register register 20... is a line sensor. Patent Applicant: Canon Co., Ltd. - No. 5 Agent: Yasunori Otsuka, Patent Attorney
E Figure 6

Claims (1)

[Claims] In an image reading device that has a line sensor formed by connecting a plurality of image sensors in a substantially horizontal direction with joints at predetermined gaps, and outputs an image signal from image data of an original image obtained by scanning the line sensor, Focusing on two effective pixel sections having one or more pixel lengths adjacent to each other before and after the joint, the gap at the joint is made inversely proportional to the two effective pixel lengths, dividing the gap into the first half and the second half. In the first half of the gap, a proportional density according to the density gradient of the effective pixel section adjacent to the front part of the gap is used as interpolated pixel data, and in the second half of the gap, the proportional density according to the density gradient of the effective pixel section adjacent to the rear part of the gap is used as interpolated pixel data. An image reading device that restores lost image data within a seam by using proportional density as interpolated pixel data.