JPS5951028B2 - character reading device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a character reading system, and more particularly to an apparatus for optically reading characters formatted by printing, typing, etc.

Conventionally, devices for reading typed characters or handwritten characters are
Most of them employ complicated scanning means using flying-spot scanning tubes or registers with a huge number of stages, making them extremely expensive.

In addition, as a means of identifying characters, means such as pattern matching are used, so when the original character pattern is converted into an electrical signal and stored in a shift register etc., the character pattern may be tilted, vertically, horizontally If the characters are moved or are irregularly reduced or enlarged in the vertical or horizontal direction, the characters cannot be identified. Therefore, it is necessary to perform complex preprocessing.

However, even with such preprocessing, if the relative inclination of a character to the vertical direction or the scanning speed in the horizontal direction changes significantly, it becomes difficult to identify the character. Therefore, precise scanning such as a flying point scanning tube is required, and the paper feeding mechanism also requires a precise mechanism that keeps the feeding speed and traveling position constant, resulting in extremely It becomes expensive.

The present invention improves the conventional drawbacks as described above.
The goal is to be able to distinguish characters even when there are large variations in horizontal scanning speed and relative slope between the characters and the vertical scan, and to be able to read the characters even when the sensor is manually moved. An object of the present invention is to provide a low-cost character reading device that enables reading of characters.

The present invention calculates the appearance frequency distribution of character line segments projected onto the vertical axis of the read pattern every time one scan in the vertical direction is completed, normalizes the frequency distribution, and calculates the appearance frequency distribution of character line segments in the vertical axis direction. This method is characterized by determining the position distribution and identifying scanned characters.

Characters printed or handwritten on paper are scanned 5 by a sensor having photoelectric conversion elements arranged in a row (this includes scanning at an angle within a permissible range, and the same applies hereinafter).
, the horizontal scan movement proceeds from right to left or left to right across the character, and each vertical scan is sampled at regular time intervals.

The photoelectric conversion element of the sensor will be referred to as a cell, and the area where data is compressed to the resolution required for recognition will be referred to as a unit area. When the sampled signal is scanned once in the vertical direction, the number of occurrences of a character line segment is counted for each unit area, and the vertical Count the number of scans.

The ratio of the number of appearances for each unit area to the maximum value of that number of appearances, and the ratio of the number of vertical scans from the time of appearance of each unit area to the maximum value of the number of vertical scans are calculated, and each It divides into multiple sections and identifies scanned characters based on the vertical arrangement order of the sections. By normalizing the ratio of the number of appearances or vertical scan times for each unit area and the respective maximum values, scanned characters can be identified without the need for preprocessing and regardless of variations in horizontal scan speed. However, the scanned characters can also be identified if the vertical scan and the characters are tilted relative to each other.

In addition, the number of appearances and the number of vertical scans are counted for each unit area, the ratio of each to the maximum value is determined, normalized, and characters are identified from each vertical arrangement. No storage device is required to store information.

FIGS. 2 and 3 show one embodiment of the present invention, and are block diagrams of a character reading device in which a scanner 12 is held in the hand and is moved horizontally to scan. lamp 13
A and 13b illuminate the paper 11, and the character pattern on the paper 11 is imaged via the lens system 14 onto a sensor 15 having photoelectric conversion elements arranged in a row.

Since the light reflected from the background area of the paper 11 and the light reflected from the character area are different, the corresponding signal levels from each cell are different. The signals obtained in each cell of the sensor 15 are applied to a control and binarization circuit 16,
White and black are determined by level determination. For example, the signal corresponding to the background area of the paper 11, that is, white, is
The signal corresponding to the character area, that is, black, is output as ``1'' to the subsequent stage. When the signal corresponding to each unit area is sequentially inputted from the control and binarization circuit 16, the power counter control circuit 25 causes the preset counter 26 to input the signal of the shift register 27 when the signal is black, that is, "1". The output is preset, 1 is added to the contents of the preset counter 26, and the result is stored in the shift register 27.

Also, when the signal is white or “o'', the preset
The counter 26 is preset with the output of the shift register 27, and the contents of the reset counter 26 are controlled to be stored as they are in the shift register 27. Also,
The counter control circuit 25 receives a one-character scanning end signal GAP from the gap flag 21, which will be described later.
After clearing the contents of the preset counter 26, when the signal is "1", 1 is added, and when the signal is "O", the contents of the preset counter 26 are directly stored in the shift register 27. shift register 2
7 has the number of stages corresponding to the number of cells of the sensor 15 or the number of unit areas corresponding to the data compressed in the vertical direction. The contents of the shift register 27 represent the number of times a character line segment in the horizontal direction appears in each unit area. Also, when storing the contents of preset counter 26 into shift register 27, the contents of preset counter 20 are sent to comparison circuit 28. The comparison circuit 28 is
The content A of the storage device 29 is compared with the content B input from the preset counter 26, and content B is stored in the storage device 29 only when B>A. When one vertical scan is completed, the signal SEND from the control and binarization circuit 16 is
The contents of the storage device 29 are transferred to the maximum value storage device 30 at the timing of , and then the contents of the storage device 29 are cleared.

That is, the maximum value storage device 30 stores the maximum value of the count of the number of appearances of black for each unit area stored in the shift register 27. (The normalization circuit 31 calculates the ratio between the contents of the shift register 27 and the contents of the maximum value storage device 30, and divides it into a plurality of parts (hereinafter referred to as normalized data).The scanner 12 is moved manually. However, in the present invention, the appearance of black Since the number of times black appears in each unit area is normalized using the maximum value of the number of times, the normalized value is not affected greatly even if the horizontal scanning speed or relative angle is different. The contents divided into multiple parts are sent to the binarization circuit 32.Binarization circuit 32
Then, the normalized data is compared with a certain threshold value SHL, and the binary information, that is, the pattern information projected on the vertical axis, is divided into a character area and a background area of the paper 11. For example, in FIG. 1, the sensor field of view 2 is a sheet of paper 1 on which characters 3, 4, and 5 are written.
FIG. 5 shows the normalized data at the T2 position and the binarized normalized data when moving from the above position t1 to the T2 position.

The background area and character area after binarization are each W
night) and B (black). From this binary data, the unit area number CTA at the top end of the character is extracted by the cutting circuit 33.
is detected, character area CHH is determined, and signal POF is output. When the output of the binarization circuit 32 is B, it means that a line segment has appeared, and when it is W, it means that no line segment has appeared. The counting of the number of vertical scans from the first appearance of black for each unit area will be explained. Black in the output signal of the binarization circuit 32, that is, "1" indicates that black appears in the unit area, and conversely, white, that is, "0" indicates that black does not appear in the unit area. Therefore, if the binarization signal from the binarization circuit 32 has a clock count of ゜゜1゛, the number of scans since the first appearance of black can be obtained.Counter control circuit 3
5 receives the binarized signals corresponding to each unit area of the binarization circuit 32 sequentially and performs the same operation as the counter control circuit 25 described above.

Similarly, preset counter 36, shift register 37, comparator circuit 38, storage device 39 and maximum value storage device 40 each correspond to the aforementioned preset counter 2.
6, shift register 27, comparison circuit 28, storage device 2
9 and the maximum value storage device 30,
The maximum value storage device 40 stores the maximum number of vertical scans from the first black appearance point for each unit area stored in the shift register 37. This method counts the number of vertical scans from the first appearance of a character line segment in each unit area, and detects the position of the first character line segment in each unit area. register 37
The contents are as follows. The normalization circuit 41 includes the shift register 3
The ratio between the contents of 7 and the contents of the maximum value storage device 40 is calculated, divided into a plurality of parts, and a signal HPi is output. Next, a method for detecting the completion of scanning one character will be described.

Gap detection flags 18 and 19 are set by the SEND signal generated after vertical scanning is completed, and gap detection flag 18 is cleared when black (character line segment) is detected between vertical character cutting signals POF. , a gap detection flag 19 is a flip-flop that is cleared when black (character line segment) is detected within one vertical scan.

After one vertical scan, immediately before generating the SEND signal, the gap detection flag 18 is set at the clock CKI timing.
19 is added to the OR circuit 20, and the OR signal is stored in the gap flag 21.
Transfer the contents of flag 21 to gap flag-22. In other words, the gap flag 21 is a flag that is set when no character line segment is detected in the vertical character extraction section POF, or when no character line segment is detected at all within one vertical scan.・Flag 22
is just before. This flag stores the contents of the gap flag 21 during vertical scanning. The output signal GAP of the gap flag 21 is output to the counter control circuit 25, and is ANDed with the signal SEND by the AND gate 23 to add the contents of the space counter 2.4. When the content of the space counter 24 exceeds a preset constant, it is considered that one line of characters has been scanned, a space signal SP is output, and the initial state is set. Space counter 24 is cleared when black (character line segment) is detected. Gap flag 22 outputs signal BFGAP and stores signal GAP. At this time, when BFGAP is "0" and GAP is "1", it is assumed that scanning of one character has been completed, and the character is identified.In Fig. 1, the sensor field of view 2 moves to position T. When the character area shown in FIG. “
1'', and the gap flag 21 is set to "1".

T, in the previous position, the gap flag 21 is “0''
Therefore, at T2, GAP=゛゛1'', BFGAP="
o'゛, it is assumed that scanning of one character has been completed, and the identification operation is performed thereafter. The space counter 24 ends a line with a number of times much greater than the number of vertical scans between characters. When the scanning of one character is completed, the signal HBi output from the normalization circuits 31 and 41 is
and HPi are the read-only storage device 4 of FIG.
2 and 43, and then the read-only storage device 42
The contents read from the storage devices 44 and 43 are stored in the storage devices 44 and 4.
5 is written. The address signals of the read-only storage devices 42 and 43 are each composed of a signal HBi, an output Ci of the storage device 44, a signal HPi, and an output Di of the storage device 45. The storage devices 44 and 45 store the character area C.
Divide HBi and HPi in HH into identification character groups Ci,
Di is input to the identification circuit 46, and the identification circuit 46 outputs the result of identification, Dwr. For example, when scanning the number "1" in the 0CR-A font from left to right, the ratio of the number of times black appears in each unit area and the maximum number of vertical scans from the first appearance of black to the respective maximum values is 0. ~7 (0 for 8 types
Figure 4a shows the values HBi and HPi, which are divided into different groups (not shown), and Figures 5 and 6 show the patterns when the values are projected onto the vertical axis.

In Fig. 4a, arrow X indicates the scanning direction, and black and white of the black and white pattern data obtained by scanning are indicated by '' and .
'', and HBi and HPi are divided into eight types from 0 to 7.

That is, when the maximum value of the number of appearances and the number of vertical scans is 7, the ratio is shown for each unit area. For HBi, the larger the number, the more times black appears, and for HPi, the larger the number, the more forward the position of the line segment is, and the values are as shown in FIG. 4c. In the first vertical scan S1 in FIG. 4a, unit areas of n and n+1 (referred to as cells)
Since black ("1") appears in the cell, +1 is added to the n and n+1 stages of the shift register 27 corresponding to the cells n and n+1, and 1 is stored in the maximum value storage device 30.Next, 1 is stored in the maximum value storage device 30.
In vertical scanning of 2, +1 is added to the n and n+1 stages of the shift register 27 corresponding to the n and n+1 cells,
The maximum value storage device 30 is updated to 2. In addition, in S2, when the shift register 27 outputs the NNn+1st stage, the binarization circuit 32 outputs "1", so the output 3
7, the nth and n+1st stages are also added +1, and the maximum value storage device 4
1 is set to 0. It becomes black ('1'), and +1 is added to the n, n+1, and mth stages of the shift register 27, and the result is stored. The same applies to the shift register 37. S2
No black appears in the vertical scan of 3, the gap flag 21 is set, and the scan of one character is completed. At this point,
What is set in the maximum value storage device 30 is the content of the m-th stage of the shift register 27 corresponding to the m cell, and this is set as Bmax. The ratio between the content of each stage of the shift register 27, which is the black count value appearing in each cell, and Bmax is determined. for example,. N. N. of the shift register 27 corresponding to cell n+1. The contents of the n+1st row are Bn. Bn+1
Then, Bn/Bmax. This is to find Bn+1/Bmax. Furthermore, Bn/BmaxlBn+1/Bm
The value of ax is divided into 8 parts to form HBi. Similarly, the maximum value storage device 40 which is the maximum value of the shift register 37 is n
and n and n of the shift register corresponding to cell n+1
The content of the +1st row is ゛, and it is assumed to be Pmax. N. n+1
The contents of the nth and n+1st stages of the shift register 37 corresponding to the cell of Pn. pn+1, Pn/Pmax. pn+
Find 1/Pmax. This is divided into 8 parts and called HPi. Next, a method for selecting a character group will be explained. A part of the state transition diagram of the character group classification circuit unit M (not shown) consisting of the read-only storage device 42 and the storage device 44 and the character group are shown in the seventh figure.
In addition, FIG. 8 shows a part of a state transition diagram of a character group classification circuit section N (not shown) consisting of a read-only storage device 43 and a storage device 45, and a character group.

The alphanumeric characters in the circles in Figures 7 and 8 are codes that indicate changed states, and the numbers near the arrows indicate that the state will change to the state indicated by the arrow when the values of HBi and HPi appear. .

In Figure 4 a, HBi is 5, 5, 1, 1, 1 from the top of the letters.
...is appearing. From Figure 7, first starting from CO, the first HBi is 5, so the state is C
Changes from O to C2. Next, since HBi is 5, C4
Then, when HBi is 1, the state changes to C8. Similarly, the state changes to Cl3 and C3l, and C3l becomes the final state. C3l is a character group including the character "1", and C3l is output from the storage device 44. The same is true for HPi; from Figure 8, HPi is 7, 7, 4, 4, 4.
...... appears, so the status is DO, Dl, D2
, D3, D6, D9, and D3l, and D3l is output from the storage device 45 as a character group including the character "1".
The identification circuit 46 outputs the character "1" from C3] and D3l as the identification result. In the present invention, classification is made based on the ratio of the number of times black appears in each unit area and the maximum number of vertical scans from the first black appearance, so the horizontal scanning speed may differ, or the permissible Even if the characters to be read and the row of unit areas are scanned at a relative angle within the range, they can be identified.

For example, as shown in Figure 4b, the horizontal movement speed, that is, the horizontal scanning speed is slow, so the number of vertical scans is greater than in Figure 4a, and the rows of characters and unit areas appear to be tilted. Even if black and white pattern data is obtained by scanning, it can be identified in the same manner as in FIG. 4a. In vertical scanning, sampling is performed at a constant period, so if the speed at which the paper or the photoelectric conversion element array is moved in the horizontal direction differs, the vertical scanning for scanning one character differs. In the present invention, the number of appearances of black and the number of times of scanning from the first black are counted for each unit area, and normalization is performed using the ratio to the maximum value of each. Even if the horizontal speed is fast and the number of scans is small, or even if the speed is slow and the number of scans is large, the number of scans from the first black in the black appearance frequency band for each unit area and the maximum of each The ratio remains almost constant. Therefore, the influence of fluctuations in scanning speed is small. Furthermore, even if the characters and the photoelectric conversion element rows are tilted, the number of times black appears in each unit area is approximately the same. Therefore, when scanning is performed with a tilt, the number of scans from the first black for each unit area will be slightly different, but by taking the tilt into consideration when dividing the normalized value from the maximum value, , the influence of the tilt can be reduced within the permissible range. Furthermore, by considering the slope in the state transition of the character group segmentation circuit N, the influence of the slope can be reduced. In addition, a fixed value is determined that is smaller than the minimum number of vertical scans per character at the predetermined maximum horizontal scanning speed, and if the scanning of one character is completed in the number of vertical scans less than the fixed value, The data is treated as noise, and the output HBi of the normalization circuit 31 is
This will be indicated and no further identification will be made.

If paper or a scanner is mechanically moved horizontally at a nearly constant speed, count the number of times black appears in each unit area and the number of vertical scans from the first appearance of black, and use only that value. It is also possible to output HBi and HPi by dividing the data into HBi and HPi.

In addition, in the embodiment, the division of the unit area is performed by photoelectric conversion element rows, and the case is shown in which they are arranged in one row in the vertical direction, but it is also possible to have multiple rows or a two-dimensional arrangement. It can also be constructed in combination with optical fibers.

Of course, it is also possible to scan using a flying point scanning tube or a vidicon. Furthermore, the target characters are
It is possible to read and identify formalized numbers, symbols, alphabetic characters, etc.

In the embodiment, it has been explained that the appearance frequency distribution of character line segments projected onto the vertical axis and characters scanned from the front shape are identified, but in the present invention, character line segments do not appear in each unit area. The back shape of the character is calculated by counting the vertical scanning circuits from when the last character line segment appears until the end of one character scan, and from the back shape and the appearance frequency distribution of the character line segment, or the character line It is also possible to identify scanned characters from the appearance frequency distribution of minutes, front shape, and back shape.

As explained above, the first invention includes a first counting means (in the embodiment, a counter control circuit 25, a preset Counter 26 and shift register 27
), and a first selection means (consisting of a read-only storage device 42 and a storage device 44 in the embodiment) for selecting a first character group based on the count value of the first counting means. and the number of vertical scans for each unit area of the sensor from the time the first character line segment appears to the end of one character scan, or from the time the last character line segment appears to the end of one character scan. second counting means (in the embodiment, a counter control circuit 35, a preset counter 36, a shift register 37) for counting the number of vertical scans of the
), and a second selection means (consisting of a read-only storage device 43 and a storage device 45 in the embodiment) that selects a second character group based on the count value of the second counting means. and an identification means for identifying scanned characters, symbols, etc. based on the first and second character groups selected by the first and second selection means, and an identification means for identifying the scanned characters, symbols, etc. as electrical signals. Since there is no need to store character patterns, there is no need for a large-scale storage device, and there is an advantage that the character reading device can be constructed economically.

Further, the second invention includes a first counting means for counting the number of appearance of character line segments in each unit area of the sensor for each character, a maximum value of the count value of the first counting means, and each count value. A first method for determining the normalized distribution of each count value based on the ratio of
based on the normalization means (in the embodiment, it consists of a comparison circuit 28, a storage device 29, a maximum value storage device 30, and a normalization circuit 31) and the normalized distribution obtained by the first normalization means. a first selection means for selecting a first character group; and a first selection means for selecting a first character group; a second counting means for counting the number of vertical scans until the end of scanning; and a second counting means for calculating a normalized distribution of each count value based on the maximum value of the count value of the second counting means and each count value. Normalization means (in the embodiment, the comparison circuit 38,
a second selection means for selecting a second character group based on the normalized distribution obtained by the second normalization means; Since the device is equipped with an identification means for identifying the symbol, etc. selected by the first and second selection means, it is possible to use it when the horizontal scanning speed is different or when the characters and the photoelectric conversion element array are tilted. It also has the advantage of being able to identify characters.

[Brief explanation of drawings]

Fig. 1 is a diagram of the relationship between characters on paper and the reading head, Figs. 2 and 3 are block diagrams of a character reading device showing an embodiment of the present invention, and Fig. 4 is a monochrome pattern and normalized data. FIG. 5 is an appearance frequency distribution diagram of character line segments, FIG. 6 is a diagram of the front shape of characters, and FIGS. 7 and 8 are schematic diagrams of internal state transitions of the character group division section. 1, 11: paper, 2: sensor field of view, 3, 4, 5: characters,
12: Scanner, 13a, 13b: Lamp, 14: Lens system, 15: Sensor, 16: [■■: Gap detection flag, 21, 22: Gap flag, 24: Space
Counter, 25, 35: Counter control circuit, 26, 36
: Preset counter, 27, 37: Shift register, 28, 38: Comparison circuit, 29, 39: Storage device, 3
0, 40: maximum value storage device, 31, 41: normalization circuit,
42, 43: Read-only storage device, 44, 45: Storage device, 46: Identification circuit, CHH: Character area, CTA: Unit area number at the top end of character, POF: Vertical character cutting signal, SEND: End of vertical scanning subsequent sending signal, SP: space signal, HBi: normalized distribution of the appearance frequency of character line segments projected onto the vertical axis, HPi: normalized data of the number of vertical scans from the first appearance of black, Ci: storage device 44 Di: output of the character group classification circuit section N of the storage device 45;

Claims (1)

[Claims] 1. Characters, symbols, etc. scanned by a sensor having light receiving elements arranged vertically or horizontally;
In a character reading device for identifying symbols, etc., a first counting means counts the number of times a character line segment appears in each unit area of the sensor for each character; a first selection means for selecting one character group, and a number of vertical scans for each unit area of the sensor from the time when the first character line segment appears to the end of one character scanning, or from the time the last character line segment appears to the end of one character scanning; a second counting means for counting the number of vertical scans up to the time; a second selection means for selecting a second character group based on the counting result of the second counting means; A character reading device comprising: identification means for identifying scanned characters, symbols, etc. based on the first and second character groups selected by the selection means. 2. The character reading device according to claim 1, wherein the identification means uses characters commonly included in the first and second character groups as an identification result. 3 Characters, symbols, etc. are scanned by a sensor having light receiving elements arranged vertically or horizontally, and the scanned characters,
In a character reading device for identifying symbols, etc., a first counting means counts the number of times a character line segment appears in each unit area of the sensor for each character; a maximum value of the count value of the first counting means; a first normalization means that obtains a normalized distribution of each count value based on a ratio with each count value; and a first character group selected based on the normalized distribution obtained by the first normalization means. a first selection means for determining the number of vertical scans from the appearance of the first character line segment to the end of one character scan or the number of vertical scans from the appearance of the last character line segment to the end of one character scan for each unit area of the sensor; a second counting means for counting;
a second normalization means for calculating a normalized distribution of the number of vertical scans based on the maximum value of the count value of the counting means and the ratio of each count value; and a normalization distribution obtained by the second normalization means. a second selection means for selecting a second character group based on the first,
A character reading device comprising: identification means for identifying scanned characters, symbols, etc. based on the first and second character groups selected by the second selection means. 4. The character reading device according to claim 2, wherein the identification means uses characters commonly included in the first and second character groups as an identification result.