JPH0222428B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a character reading method in which a sheet of paper on which characters, symbols, etc. are written is scanned using a scanner, and information such as characters, symbols, etc. is optically read.

Conventionally, optical character reading devices that optically read characters written on paper while holding a scanner in the hand and moving it around detect the character area in the vertical and horizontal directions and detect the characters when they reach a predetermined position. At the same time, characters are identified and the identification results are output. However, generally the predetermined position is one
Since there is only one location, if the influence of noise is large at that location, it may be difficult to read letters, etc.

FIG. 1 is a block diagram of a conventional optical character reading device, in which 1 is a scanner, 2 is a hand, 3 is paper with numbers written on it, 4 is a lamp that illuminates the paper 3, 5 is a lens system, 6 is a two-dimensional sensor, 7 is a binarization circuit, 8 is a cutting circuit, 9 is a cutting flag, 10 is a timing circuit, 11 is a cutting judgment circuit, and 12 is an identification circuit.

Inside the scanner 1, there is a lamp 4 that illuminates the paper 3, a lens system 5 that images the light reflected from the paper onto the light receiving surface of the two-dimensional sensor 6, and a binary system that binarizes the output signal of the two-dimensional sensor 6. A digitization circuit 7 is provided, and an operator holds the scanner 1 in his/her hand 2 and moves it horizontally over a sheet 3 to read characters.

When a one-dimensional sensor arranged vertically is used, the scanner 1 is moved horizontally by hand 2, so the speed of movement in the horizontal direction is not constant, and the number of horizontal scans varies depending on the operation. Even when reading the same character, the horizontal scanning data differs for each operation, making it very difficult to identify the character.
For this reason, in the invention, a two-dimensional sensor 6 is used in which one character can be seen in the field of view, and almost the same scanning data can always be obtained for one character.

As shown in FIG. 2, the two-dimensional sensor 6 has n×m photoelectric conversion elements arranged on a light-receiving surface. It outputs a signal corresponding to the photoelectric conversion element arranged in the Bn column, then outputs a signal corresponding to the photoelectric conversion element arranged in the B1 to Bn columns of the L2 row, and then sequentially continues with the L3 row,
Finally, by outputting the signals corresponding to the photoelectric conversion elements arranged in the B1 to Bn columns of the Lm row,
Scanning of one screen is completed. Note that after the scanning of one screen is completed, scanning of the next screen is started again, but during this period a certain period of time (for example, the clock signal is
A flyback period of one cycle of CK is provided.

The scanning time for one screen from the L1 row, B1 column to the Lm row, Bn column is very fast, for example, 0.0001 to 0.0002 seconds, so it is much faster than the time it takes for the hand 2 to move the scanner 1 laterally. The two-dimensional sensor 6 of the scanner 1 scans the same character many times while passing through one character. Therefore, by starting the later-described identification process on the first screen when the character reaches a predetermined position on the two-dimensional sensor 6, the same character will not be read over and over again.

The binarization circuit 7 converts the signal applied from the two-dimensional sensor 6 into a binary signal corresponding to the character area and the background area, and sequentially corresponds to columns B2, B3...Bn from column B1 of row L1. Then outputs the binary signal corresponding to the B1, B2, ...Bn columns of the L2 row, then the L3 row, and finally the Lm row of the B1, B2, ...Bn columns, and outputs the binary signal corresponding to the B1, B2, ...Bn columns of the Lm row. 8 and identification circuit 12.

The extraction circuit 8 determines whether or not the character has reached a predetermined position on the light-receiving surface of the two-dimensional sensor 6 based on the binary signal from the binarization circuit 7 . For example, as shown in FIG.
When the number "2" is imaged on the light receiving surface of
When all the binary signals corresponding to column B1 indicate a background area (hereinafter abbreviated as white), and part of the binary signals corresponding to column B2 indicate a character area (hereinafter abbreviated as black). , when the character reaches a predetermined position, the output signal VH is set to "1" and is applied to the cutting flag 9 and the determination circuit 11. If the moving speed of the hand is slow, the state in which all of the B1 column is white and the B2 column is black may continue multiple times, resulting in the same character being read multiple times. To prevent this, a cutting flag 9 is provided. “1” in cutting flag 9
is set, it is assumed that a character has been read once, but will not be read in the same state the next time. A signal RES is output to indicate when a character leaves the field of view and the next character comes, and the cutting flag 9 is reset to read the next character. This operation will be explained next.

During scanning, for example, when all the binary signals corresponding to column B3 or column B4 become white, the cutting circuit 8 sets its output signal RES to "1" and resets the cutting flag 9. It is something.

In addition to the signals RES and VH, the output signal a from the timing circuit 10 is added as "1" to the cutting flag 9 during the flyback period. When the signal VH is "1" when the signal a is input, the cutting flag 9 is set to "1", and thereafter the output signal BF from the cutting flag 9 is held at "1". Also, the signal
When RES becomes "1" and the output signal a of the timing circuit 10 is input to the cutout flag 9, the contents of the cutout flag 9 are reset, and the subsequent output signal BF from the cutout flag 9 is set to "0". That is. In addition, the cutout determination circuit 11 is configured so that the signal BF is “0”.
And when the signal VH is "1", assuming that a character has arrived at the predetermined position, the output signal b is set to "1",
It instructs the identification circuit 12 to start identification,
Thereby, the identification circuit 12 performs character identification based on the binary signal applied from the binarization circuit 7. In this way, when the signal VH becomes "1",
The cutout flag 9 is set at the rising edge of the signal a, and the output signal BF is set to "1" to prevent character identification from occurring after that, in order to prevent the same character from being identified multiple times. This is to make it so.

The conventional optical character reading device shown in FIG.
As described above, when a character reaches a predetermined position on the light-receiving surface of the two-dimensional sensor 6, the character is identified once. There is small dirt attached to the screen #n, #(n+1), #n, #(n+1), shown in FIGS.
Assuming that image information of #(n+2) is output, there is a drawback in that the number "2" is not identified. That is, in screen #n shown in FIG.
The binary signals corresponding to columns 1 and B2 all indicate white. Also, in screen #(n+1) shown in B of the same figure, since Noise I is in column B1, the binary signals corresponding to column B1 are , not everything indicates white,
Also, in screen #(n+2) shown in C of the same figure,
Since the numbers are in column B1, the binary signal corresponding to column B1 is
This is because not all of the numbers indicate white, and therefore the cutout circuit 8 cannot detect that the number has reached the predetermined position and cannot set the signal VH to "1".

Therefore, for example, when reading the 3-digit number ``123'', if it is not possible to detect that ``2'' has reached the predetermined position for the above reason, the read result will be the 2-digit number ``13''. As a result, there was a drawback that a major error occurred.

The present invention has been made to improve the above-mentioned drawbacks, and its purpose is to detect that a character has reached a predetermined position even when it is not possible to detect that the character has reached a predetermined position due to the influence of noise.
The purpose is to enable character identification. Examples will be described in detail below.

FIG. 4 is a block diagram of an optical character reading device implementing the method of the present invention, in which 13 is a character cutting circuit, 14 is a white detection flag, 15 is a black detection flag,
16 and 17 are AND gates, and 18 is an OR gate, and the same reference numerals as in other FIG. 1 represent the same parts.

When you hold the scanner 1 in your hand and move it horizontally over the paper 3, the binarization circuit 7
Binary signals corresponding to the character area and the background area are output from. When all the binary signals corresponding to the B1 column indicate white and a part of the binary signals corresponding to the B2 column indicate black, the cutout circuit 13 selects a character to be placed at a predetermined position on the two-dimensional sensor 6 in the same way as described above. (As shown below, all the binary signals corresponding to column B1 show white,
The state in which a part of the binary signal corresponding to column B2 shows black is called the character image reaching a predetermined position), and the output signal VH0 is set to "1". When all binary signals indicate white, the signal RES is set to "1" and the cutout flag 9 is set to "1".
to reset. When the binary signals corresponding to the B1 and B2 columns all indicate white, the signal c is set to "1", and otherwise, the signal c is set to "0". Further, when any of the binary signals corresponding to columns B1 and B2 includes a binary signal indicating black, the signal d is set to "1", and otherwise, d is set to "0".

Now, for example, from the two-dimensional sensor 6,
If the image information of screens #n, #(n+1), and #(n+2) shown in FIG.
The output signals VH0, c, and d are shown in Fig. 5A, respectively.
The results are shown in B and C. In FIGS. 5A to 5J, the upper level is "1" and the lower level is "0".

The white detection flag 14 and the black detection flag 15 are activated by the cutting circuit 3 at the rising edge of the signal a from the timing circuit 10, that is, between the screens.
Take in signals c and d from , and signals c and d are “1”
Then, the output signals e and f of the white detection flag 14 and black detection flag 15 are set to "1" by the next signal a of the timing circuit 10, and if the signals c and d are "0", the next signal a is set to "1". In this case, the output signals e and f of the white detection flag 14 and the black detection flag 15 are set to "0". Signals c and d are shown in Fig. 5B,
If the change occurs as shown in C, the output signals e and f of the white detection flag 14 and black detection flag 15 become as shown in D and E in the figure. That is, white detection flag 1
The output signal e of 4 indicates the state of the signal c one screen before, and the output signal f of the black detection flag 15 indicates the state of the signal d one screen before. The AND gate 16 is connected to this white detection flag 14.
The AND gate 17 takes the logical product of the output signal e of the output signal e and the output signal d of the extraction circuit 13 and adds it to the OR gate 18. The logical product is calculated and added to the OR gate 18. In this case, the output signals VH1, VH2 of the AND gates 16 and 17 and the output signal VH of the OR gate 18 are as shown in F, G, and H of the figure, respectively.

Cutout flag 9 is output signal VH of OR gate 18
When becomes “1”, it is set between screen #(n+1) and screen #(n+2), and its output signal
BF is set to "1" as shown in the figure. Further, when the output signal VH of the OR gate 18 becomes "1" when the output signal BF of the cut flag 9 is "0", the cutout determination circuit 11 sets the output signal b to "1" as shown in FIG. Then, the discrimination circuit 12 is instructed to start discrimination, and thereby the discrimination circuit 12 discriminates characters based on the binary signal applied from the binarization circuit 7.

In this way, in this embodiment, B1 of the current screen,
When the binary signal corresponding to the B2 column includes a binary signal indicating black, and the binary signals corresponding to the B1 and B2 columns one screen before all indicate white, that is, the output signal d of the cutting circuit 13 is “1” and the output signal e of the white detection flag 14 is “1”, the character image is detected by the two-dimensional sensor 6.
When the character image passes through a predetermined position on the light receiving surface, the signal VH applied to the cutting flag 9 and the cutting judgment circuit 11 is set to "1". Even if it is not possible to detect that the character has reached a certain position, the character can be identified. As in the example shown in Figure 3, on screen #n, the B1 and B2 columns are all white, and the B1 and B2
Characters can be identified even if both columns are black.

As described above, when the binary signals corresponding to the B1 and B2 columns of the current screen both include binary signals indicating black, and the binary signals corresponding to the B1 and B2 columns of the previous screen all indicate white, Or vice versa, B1, B of the current screen
A state in which the binary signals corresponding to two columns all indicate white, and the binary signals corresponding to columns B1 and B2 one screen before both indicate black is called a character image passing a predetermined position. That is, if a character image has reached a predetermined position when an arbitrary column has a background area (white) and a character area (black) exists in a part of an adjacent column, the first detection means,
When the arbitrary column and the adjacent column are both a background area or a character area, and in the next scanning screen, the arbitrary column and the adjacent column are both a character area or a background area opposite to the previous front surface, the character image is at a predetermined position. It consists of a second detecting means that detects that the object has passed through.

In addition, in this embodiment, screens #(n+2) and #(n
+1), #n image information is output, that is,
Even when the scanner 1 is moved in the opposite direction to that described above, when the character image passes a predetermined position on the light receiving surface of the two-dimensional sensor 6, the output signal f of the black detection flag 15 and the cutout circuit 13 are detected. Since the output signal VH2 of the AND gate 17 which takes the AND with the output signal c of It is also possible to identify characters. In addition, the influence of small noises as shown on screen #(n+1) can be removed using known methods, so even if characters are identified using screen #(n+1), the recognition of the characters themselves will be affected. However, no error will occur in the identification results.

As explained above, the present invention provides a character recognition start signal (in the embodiment, a signal
When the first detection means such as the cutting circuit 13 that outputs VH0) and the two-dimensional sensor detect that the character image has passed without detecting that the character image has reached the predetermined position, the character is identified. A cutout circuit 13 that outputs a start signal (signal VH1 or signal VH2 in the embodiment), a white detection flag 14, and a black detection flag 1
5. A second detection means consisting of AND gates 16, 17, etc. is provided, and when the first detection means does not output a character identification start signal, the second detection means outputs an identification start signal. Since the system starts identifying characters at the same time, it has the advantage of being able to identify characters, symbols, etc. even when it is not possible to detect that the character image has reached a predetermined position due to the influence of noise, etc. be.

[Brief explanation of drawings]

Figure 1 is a block diagram of a device implementing the conventional method, Figure 2 is an explanatory diagram of a two-dimensional sensor, and Figures 3A-
C is a diagram showing image information output from a two-dimensional sensor, FIG. 4 is a block diagram of an apparatus implementing the present invention, and FIG. 5 is an explanatory diagram of its operation. 1 is a scanner, 2 is a hand, 3 is paper, 4 is a lamp, 5 is a lens system, 6 is a two-dimensional sensor, 7 is a binarization circuit, 8 and 13 are cutting circuits, 9 is a cutting flag,
10 is a timing circuit, 11 is a cutout determination circuit, 1
2 is an identification circuit, 14 is a white detection flag, 15 is a black detection flag, 16 and 17 are AND gates, and 18 is an OR gate.

Claims (1)

[Scope of Claims] 1. A scanner equipped with a two-dimensional sensor that enters reflected light from a sheet of paper on which characters and symbols are written and converts it into an electrical signal is moved, and characters are detected based on the output signal of the two-dimensional sensor. , in a character reading method for identifying symbols, a first detection means outputs a character identification start signal when it is detected that a character image has reached a predetermined position of the two-dimensional sensor; a second detection means that outputs a character recognition start signal when it detects that a character image has passed without detecting that a character has reached a predetermined position of the sensor; When the identification start signal is output, character identification is started and the first
A character reading method characterized in that when a character identification start signal is not outputted from the second detection means, character identification is started when a character identification start signal is outputted from the second detection means. 2. The first detection means detects that an arbitrary column is a background area (white) and a text area (black) exists in a part of an adjacent column; An arbitrary column and an adjacent column are both a background area (white) or a character area (black), and in the next scanning screen, both the above arbitrary column and an adjacent column are a character area (black) opposite to the previous screen, or 2. The character reading method according to claim 1, wherein the character reading method is obtained by detecting a background area (white).