JPH0131750B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an information input control method for inputting information such as characters and figures on a sheet.

When an optical character reader (OCR) inputs a character image or a facsimile device inputs a document image to be transmitted, as the size of the target sheet increases, A single one-dimensional semiconductor sensor (hereinafter simply referred to as a "sensor") does not have enough pixels, and it becomes necessary to use a combination of multiple sensors. this is,
The main cause is that the degree of integration of the semiconductor sensor itself has a manufacturing limit, but when a plurality of sensors are used in combination, the following problems occur.

When inputting images by combining multiple sensors,
The problem is the continuity of images input by each sensor. To explain using a diagram, for example, when using two sensors as shown in FIG. 1, there is an image of a straight line AC on the sheet 1, which is
Assuming that input is made by sensors (2) 3, if the light receiving elements of each of the above sensors are not positioned correctly with respect to the reference plate (not shown) on which the sheet 1 is placed, the output of the above sensors will be as shown in Figure 2. As shown in
This results in a lack of continuity. Figure 2 shows the above sensor
(1)2, This is a diagram to explain the state in which the video signal obtained from sensor (2)3 is binarized and written to the memory for each scan, and memory (1)4 is the one for sensor (1)2. The memory (2) 5 shows the output contents of the sensor (2) 3. In addition, the straight line AC in Figure 1
Point B above is sampled by sensor (1) 2 and sensor (2) 3 and stored in the corresponding memory.
It is assumed that the image is decomposed into two adjacent pixels, B ₁ and B ₂ , with pixel B ₁ corresponding to the edge of sensor (1) 2 and pixel B ₂ corresponding to the edge of sensor (2) 3. Therefore, as shown in Fig. 2, there is actually an error in the mounting position of the sensor, so as shown in this figure, for example, pixel B ₁ is spaced by n ₁ pixels in the horizontal direction.
This indicates that B ₂ is shifted by n ₂ pixels, and is also shifted by a relative l scanning line in the vertical direction.

Conventionally, in order to eliminate the image shift as described above,
A method has been adopted in which an adjustment mechanism that can be finely adjusted at least two-dimensionally is provided at the mounting part of the sensor, the position is adjusted while observing the output signal of the sensor, and the sensor is fixed after the adjustment is completed. However, this method has the following drawbacks: First, it is difficult to make adjustments, takes a long time, and has limited accuracy.
Second, even if the sensor position is adjusted and fixed in this way, the sensor position may change due to deformation of the base material that supports the sensor due to aging, etc., and the image information between multiple sensors may change. The drawback was that continuity was lost.

An object of the present invention is to eliminate the above-mentioned drawbacks of conventional image information input devices, maintain continuity of image information between multiple sensors with high precision, and provide image information that can easily cope with changes over time. An object of the present invention is to provide an information input control method for realizing an input device.

The above object of the present invention is to provide an image information input method in which a plurality of one-dimensional sensors are arranged so that their scanning areas partially overlap, and the one-dimensional sensors scan and output image information on a sheet. By scanning a predetermined test pattern arranged so as to be included in the overlapped portion that is scanned in an overlapped manner, the overlap between each one-dimensional sensor in the scanning line direction of the one-dimensional sensor is scanned in an overlapping manner. deviation width,
and a learning mode in which the amount of deviation of the scanning line between the one-dimensional sensors in the direction perpendicular to the scanning line direction is calculated and stored in advance, and a learning mode in which the sheet is scanned by a plurality of one-dimensional sensors and the amount of deviation in the overlapping portion is calculated and stored in advance. By selecting one output from the outputs of the plurality of one-dimensional sensors according to the deviation width obtained in the learning mode, and further delaying the output of the one-dimensional sensor by the deviation amount obtained in the learning mode. This is achieved by providing a correction mode for correcting the output of the one-dimensional sensor.

Embodiments of the present invention will be described in detail below with reference to the drawings.

An embodiment of the present invention is shown in which a photoelectric conversion section in an image information input device is configured with two sensors. That is, as shown in FIG. 1, the outputs of the two sensors, sensor (1) 2 and sensor (2) 3, are amplified by amplifier (1) 6 and amplifier (2) 7, respectively, and then converted to binarization circuit 8, 9 basically quantizes it into black and white binary values,
The correction circuit 10 corrects the deviation between the sensors.

First, a procedure for detecting a sensor position error in the learning mode will be described. As a test pattern for learning, a black linear test pattern 11 on a test sheet 12 is used, the end of which is connected to the sensor (1).
2. Set the sensor (2) so that it is located in the overlapping area of 3. The image obtained by the sensor ( ₁ ) 2 is as _shown in _FIG .
Detect X ₁ pixel. Similarly, the pixel output of the sensor (2) 3, as shown in FIG. 4b, is at the pixel _position , Y ₂ is detected.

From this result, for example, if Y ₁ - Y ₂ = l0, the sensor (2) 3 side has a buffer that delays l scans, and furthermore, the 1st to X _1st pixels of sensor (1) 2 are processed as valid pixels. After that, if the X ₂ +1 pixel and subsequent pixels of the sensor (2) 3 delayed by the l scanning amount are connected as effective pixels, a completely smoothly continuous scanned image can be obtained.

Next, the procedure for detecting the upper right end of the linear pattern from the output image of the sensor will be described in more detail. An enlarged view of the sensor output is shown below.
Depending on the inclination of the sensor or the test sheet, the linear pattern will be downward to the right (see FIG. 5a) or upward to the right (see FIG. 5b). In order to detect the upper right end, in any case, after detecting black in each scan, it is sufficient to detect the point of change from black to white and find its rightmost position. For example, in Figure 5a, A, B,
Since C and D are detected sequentially, D is recognized as the upper right edge, and in FIG. 5b, A, B, . . . are detected sequentially and A is recognized as the upper right edge. The position of the upper right end of the linear pattern can be determined for other sensors using exactly the same procedure.

FIG. 6 shows a specific example of an apparatus for carrying out the above-described procedure. Binarized signal (1)21 obtained from sensor (1)2
is input to the set side of the flip-flop 22, and if a black pixel exists within the scan, the flip-flop 22 is set. This flip-flop 22
is reset for each scan by the first pulse 23 of the scan. Binarized signal (1) 21
is further input to an AND gate 25 through an inverter 24, and when a white pixel is detected with the flip-flop 22 set (a black pixel appears), a flip-flop 26 is set. This set state opens the AND gate 27 and generates the X clock (1) 2 for each pixel during scanning.
8 counts up the counter 29, and its output indicates ₁ (remaining number of pixels of X ₁ ) in FIG. 4a at the end of each scan. The output of the counter 29 is
It is stored in the register 30 and compared with the minimum ₁ obtained so far by the comparator circuit 31. If the current _X1 is smaller, the output pulse of the comparator circuit 31 passes through the OR gate 32 and is output to the counter. 2
The output of 9, that is, the current ₁ data is stored in the register 30. The rightmost point position information ₁ data 33 obtained in this way is used as one of the parameters in the subsequent correction process.

Next, the procedure for determining _Y1 in FIG. 4a will be explained.

A Y clock (1) 34 (which may be the same as the first pulse 23 of the above-mentioned scan) generated for each scan causes a counter 35 to count up. On the other hand, each time the comparison circuit 31 detects a new minimum value of ₁ ,
The output pulse generated in the comparison circuit 31 clears the counter 35 and initializes the count by the Y clock (1) 34.
_One data shown in Figure a remains as the output of the counter 35 and is input to the subtraction circuit 36 for later processing.

Using exactly the same procedure as described above, install sensor (2)3.
Processing is also performed on the binary signal (2) 37 from the register 46, and the X ₂ data shown in FIG. 4b is obtained as the output of the register 46. Further, as the output of the counter 51, ₂ data is input to the subtraction circuit 36.
In this way, since _{the 1} and ₂ data are input to the subtraction circuit 36, it is possible to output _2-1 =l as the output ₅₂ with a sign.

_{Next ,} calculate X _{1 ,} The procedure for connecting will be explained based on FIG.

The l data 52 obtained as described above is decoded by the decoder 53 and sent to the sensor (1) 2 side,
Generates one decoded output signal to each sensor (2) 3 side. For example, if l=+1, Y ₁ is better than Y ₂
Since the number of scans is one more, the decoded output is input to one of the AND gates 62 so as to delay the sensor (2) 3 side by one scan, and at the same time input the decoded output to one side of the AND gate 54 so that the sensor (1) 2 side selects no delay. The decoded output is input as an input.

The output of the counter 55, which is counted up by the X clock (1) 28 generated for each pixel in the scan on the sensor (1) 2 side, is
It is compared with the X ₁ data 56, and as long as the former is not greater than the latter, the output is maintained as "1", and during that time, the AND gates 58 and 59 are opened and the X clock (1) 28
is input to the OR gate 60, and the binary signal (1) 21 is input to one of the OR gates 68, forming the first half of the X clock 61 and the binary signal 69.

Similarly, on the sensor (2) 3 side, a counter 6 is incremented by the X clock (2) 44.
4 and the X ₂ data 64 are sent to the comparison circuit 6.
5, and as long as the former is larger than the latter, the output "1" is maintained. As a result, the AND gates 66 and 67 are opened, and the X clock (2) 44 is input to one of the OR gates 60 and the binary signal 2 is input to one of the OR gates 68, forming the second half of the X clock 61 and the binary signal 69. do.

As described above, in the learning mode, the amount of displacement l of the sensor in the vertical (Y) direction and the parameters X ₁ and X ₂ of the amount of sensor overlap are detected, and then in the correction process, these parameters are used to calculate the two This makes it possible to smoothly connect the outputs of two sensors.

In the above embodiment, parameter detection in the learning mode is performed by reading the pattern once, but it is clear that the accuracy can be improved by performing this multiple times. Further, in the above embodiment, a linear test pattern was used, but it goes without saying that it is possible to use a test pattern of other shapes or to use other detection logic for detecting feature points.

Furthermore, although this embodiment has been described using an example of continuous processing of two sensors, it is clear that the present invention can also be applied to continuous processing of three or more sensors. It should also be noted that the quantization of the sensor output is not limited to binary quantization, and may be further quantized to a multilevel level.

As described above, according to the present invention, an image information input device configured to perform one-dimensional scanning using a plurality of one-dimensional sensors when inputting image information such as characters and figures on a sheet. In the information input control method, a predetermined test pattern is arranged so that its characteristic part is included in the overlapping part of adjacent sensors, and the deviation of the image information output of the sensor in the scanning line direction and in the direction perpendicular thereto is measured. Image information output of the adjacent sensor using a learning mode for detecting and storing the amount, a buffer for delaying the image information output of the sensor by a predetermined number of scans, and a means for removing image information output corresponding to the overlapping portion of the sensor. The information input control method, which consists of a correction mode that corrects This has the remarkable effect of making it possible.

[Brief explanation of drawings]

Fig. 1 is a block diagram showing the basic configuration of the image information input device, Fig. 2 is a diagram showing one row of its output,
FIG. 3 is an explanatory diagram of the learning mode according to an embodiment of the present invention, FIGS. 4 a and b are diagrams showing details of the output, and FIGS. 5 a and b are diagrams showing the procedure for determining the upper right end of the test pattern. The explanatory drawings, FIGS. 6 and 7, are block diagrams showing the specific apparatus. 2, 3: Sensor, 6, 7: Amplifier, 8, 9: 2
Value conversion circuit, 10: Correction circuit, 11: Test pattern
pin, 21, 37: binary signal, 22, 38: flip-flop, 25, 41: AND gate, 2
6, 42: Flip-flop, 27, 43: AND gate, 29, 45: Counter, 30, 46:
Register, 31, 47: Comparison circuit, 35, 51:
Counter, 36: Subtraction circuit, 53: Decoder, 5
4, 62: AND gate, 55, 63: Counter, 57, 65: Comparison circuit, 58, 59, 66,
67: AND gate, 60, 68: OR gate.

Claims (1)

[Scope of Claims] 1. An image information input method in which a plurality of one-dimensional sensors are arranged so that scanning areas partially overlap, and the one-dimensional sensors scan and output image information on a sheet, comprising: The overlap between the one-dimensional sensors in the scanning line direction of the one-dimensional sensors is determined by scanning a predetermined test pattern arranged so as to be included in the overlapping portion scanned by the sensors. scanning the sheet with the plurality of one-dimensional sensors; , selecting one output from the outputs of the plurality of one-dimensional sensors in the overlapping portion according to the deviation width, and delaying the output of the one-dimensional sensor by the deviation amount. An information input control method comprising: a step of correcting the output of the information input control method.