JPS62298885A - Optical character read system - Google Patents

Abstract

PURPOSE:To make a high-accuracy read with the read character kind of only numbers and to detect and correct a defective document by reading a symbol 'Yen'as a head character at the time of reading a money amount field. CONSTITUTION:Numbers and the symbol 'Yen' are set in a character kind selec tion table 17 and a one-character recognition part 16 recognizes the head charac ter read by a scanning part; when the character is the symbol 'Yen', a read control part 12 erases the 'Yen' from the character kind selection table 17 to leave only numbers. Then, numbers are read one after another and it is checked whether or not a final number is read at every time a number is read; when so, a minus symbol is added to the character kind selection table 17.

Description

[Detailed Description of the Invention] 3. Detailed Description of the Invention [Field of Industrial Application] The present invention relates to an optical character reading method, and in particular, to highly accurate reading of an amount field in which it is mandatory to enter a ¥ symbol. Optical character reading method % type % [Conventional technology] Traditionally, in businesses that handle amounts, such as banking, a ¥ symbol is written at the beginning of the amount entered on a form to prevent falsification of the amount. . Therefore, in the optical character reading device, it is necessary to recognize two characters by mixing numbers and the ¥ symbol in the read character types.

However, when the ¥ symbol is mixed, the number of character types that can be read increases accordingly, and the accuracy of reading numbers becomes lower than when only numbers are used. Therefore, for example,
In the method described in Publication No. 103488, by reading a specific symbol, the type of characters to be read after that is switched and read, and by separately providing a character type specification line corresponding to the written character type, the type of characters to be read can be changed. A method of switching and reading is shown. In other words, in the former method, the characters to be read in the character entry field are specified as numbers, the character type mixed in these is specified as "Katakana", and once the reading character type switching symbol is read, from then on, it is changed to "Katakana". When the character type switching symbol is read again, numbers are read from then on. In addition, in the latter method, a reading character type switching specification field is provided in addition to the character entry field, and by first reading the reading character type switching specification field, the reading character type for each character in the character entry field on the next line is determined. be. Since a reading character type designation line corresponding to the reading line is provided separately, two special character type switching calls are not required, and a plurality of character types can be easily switched.

[Problem that the invention seeks to solve]

However, among the conventional methods, in the former method, that is, the method of dynamically switching read character types using special symbols, it is necessary to always read the special symbols for switching the read character types mixed with the read character types. Further, no consideration is given to the reduction in reading rate due to the mixture of special symbols in the readable character types.

In addition, in the latter method, that is, a method of separately providing a character type specification line for switching the reading character type, a character type specification line is provided,
Another problem is that it takes time and effort to fill in the character type specification line.

The purpose of the present invention is to improve these conventional problems, and to make it possible to read numbers with high precision when the ¥ symbol is required to be entered in the amount field. An object of the present invention is to provide an optical character reading method that can also check the presence or absence of symbols.

[Means for solving problems]

In order to achieve the above object, the optical character reading method of the present invention controls character reading using a format program that provides information such as the character entry position on the form and the readable character type in the entry field. Check whether information indicating that the field has a specific symbol before and after the number is added to the above format program. is characterized in that it is read using a readable character type that includes numbers and the specific symbol, and after reading the first specific symbol, it is read using a readable character type that includes only numbers. In addition, if a specific symbol is added only before a number, all fields are read using a character type that includes numbers and specific symbols, and then all fields are read using a character type that only includes numbers, and then the results of both readings are combined. There are certain characteristics.

[For production]

In the present invention, we focus on the rule that the ¥ symbol is only written at the beginning of the amount, and only the first character is read as a mixture of the ¥ symbol and numbers, and the next characters and subsequent characters are read as only numbers. . In other words, it is first checked whether information indicating that it is an amount field is added to the format program, and if it is not added, normal reading is performed, and if it is added, the ¥ symbol is mixed in the read character type. to perform reading. and.

After reading the ¥ symbol for the first time, the Y symbol is excluded from the character types to be read, and only numbers are read. As a result, characters other than the first character and the last character are read by specifying the character type using only numbers, so that a highly accurate reading rate can be obtained.

〔Example〕

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

FIG. 2 is a block diagram of a reading device to which the optical character reading method of the present invention is applied. In order for an optical character reading device to read characters written on a form, information such as the position where characters are written on the form and the type of character to be read in the entry field is required. These are communicated to the device by the format program.

The format program 11 in FIG. 2 provides the reading control unit 12 with information such as the size of the form to be read, the thickness of the form, the writing position of the characters to be read, the frame size of the writing field, and the type of characters to be read. The reading control section 12 has a scanning section 13.
．． Photoelectric conversion section 14. Pre-processing section 15. and controls the single character recognition unit 16.

The scanning unit 13 conveys the form while irradiating the form with light. Further, the photoelectric conversion unit 14 receives reflected light from the form and converts it into an electrical signal.

The reflected light from the form reflects the shading information of the pattern of characters and figures on the form, and the electrical signal is a signal corresponding to the form of the pattern (hereinafter referred to as single character). Next, the preprocessing unit 15 performs a process of binarizing the electrical signal and storing the result in an internal pattern memory.

Furthermore, the preprocessing section 15 cuts out a pattern for one character from the pattern memory and sends it to the one character recognition section 16.

The single character recognition unit 16 performs feature extraction on the input cover pattern for one character, performs matching with a category-only dictionary set in the character type selection table 17, and performs single character recognition. In this case, the fewer character types are set in the one character type selection table 17, the more accurately the characters can be read. The character type selection table 17 is set with the various information described above by the reading control unit 12, and includes character types.

FIG. 3 is a diagram showing the amount field read by the optical character reading device of the present invention, and FIG. 4 shows the reading results when the amount field of FIG. 3 is read by the conventional method and the method of the present invention. FIG.

First, the reading operation of the field 31 shown in FIG. 3 will be described in detail. Figure 3: To prevent falsification of amounts,
A field that requires adding the ¥ symbol 34 to the beginning of the amount and adding the 1 symbol 36 to the end of the amount (
(hereinafter referred to as the amount field). In addition,
32.degree. 33 in FIG. 3 is a blank field, and as a result of reading, it is determined that the field is blank because there is no shading.

When reading the monetary amount field 31 shown in FIG. Make them recognize each other. The reading results are as shown in FIG. 4(a): space, space, ¥ symbol, l, 2, 3, . It is judged as unknown, ¥ symbol, - symbol, or space.

In other words, the character 35 shown in FIG.
Since it resembles a symbol and does not resemble the shape of a "4", it cannot be distinguished from the ¥ symbol and is either unreadable or misread as a ¥ symbol, as shown in determination result 41.

Therefore, in this embodiment, when reading an amount field, information indicating that it is an amount field is added to the format program, and the reading control unit 12 checks whether or not the above information is added to the format program. do. If it is not added, the normal number reading process is performed, and if it is added, the reading process in this embodiment is performed. In other words, if the first read character type is the \ symbol, the character type selection table is used. The ¥ symbol is removed from 17 and only numbers are read after that to improve reading accuracy. As a result, as shown in Fig. 4 (b)
As shown in ), [4], which is similar to the ¥ symbol, can also be determined as "4", as shown in the reading result 42.

FIG. 1 is a flowchart of one-field recognition control processing showing one embodiment of the present invention. Figure 1 shows that information indicating the amount is added to the format program.
It shows one field recognition operation when the reading control unit 12 makes a determination. First, the reading control unit 12 sets numbers and the ¥ symbol in the character type selection table 17 (step 21).
. The reading control unit 12 controls the preprocessing unit 15 to cut out one character pattern from the left end, and the one character recognition unit 16 cuts out one character pattern.
Character recognition is performed (step 22). Next, the number of read characters in the read field is checked, and if the number of characters actually read satisfies the number of read characters, the one field recognition process is ended (step 23). At this point, the first character is still being read, so proceed to the next step. Next, it is checked whether the recognition result in step 22 is a ¥ symbol or not. If it is not a ¥ symbol, the process returns to step 22 to cut out the next character and recognize that character.

In the case of the field shown in FIG. 3, characters 32 and 33 are blank, and when character 34 is read, it is determined that it is a ¥ symbol, and the process proceeds to the next step (step 24).

Here, although it is omitted in the flow, if the ¥ symbol is not read, it violates the definition of the amount field that "the first character is the ¥ symbol", so an error is output. flag. By decoding this error flag, processing is performed to request intervention by an operator or the like. Further, this process can be skipped and reading can be continued based on instructions from a formatting program or the like.

Returning to Figure 1, we read the first character and found that it was ¥
If it is a symbol, the reading control unit 12 deletes the ¥ symbol from the character type selection table 17 set in step 21, leaving only numbers (step 25). Next, check whether the read character is the last character excluding spaces (step 26), and if it is the last character, step 28)
Proceed to. If there are still characters to be read, the character 3 shown in FIG.
5 is read (step 27), and the process returns to step 26 to check whether it is the last character. Here, the character 35 in FIG. 3 is very similar to the ¥ symbol, and in the conventional method, the ¥ symbol is added to the character type selection table 17, so the reading result 41 in FIG. As shown in the figure, it is either unreadable or misread as a \ sign. However, in this embodiment, since the character type selection table 17 contains only numbers, the character is correctly read as "4", which is the closest character shape among the numbers, as shown in the reading result 42 of FIG. 4(b).

FIG. 5 is an explanatory diagram of the reading results of an optical character reading method showing another embodiment of the present invention, and FIG. 6-11 is an operational brooch yard of the reading method of FIG.

FIG. 5(1) is a diagram showing the amount field in this embodiment. In this case, a ¥ symbol is added at the beginning, and no symbol is written at the end of the number. In this way, when a symbol is not written after a number, a method is used in which one reading result is synthesized from two reading results without switching the reading character type midway. Figure 5 (2) is the result of reading with a character type that includes numbers and the ¥ symbol, Figure 5 (3) is the result of reading with a character type that contains only numbers, and No. 5 @ (4) is the result of reading with (2). This is the result of combining and (3).

The operation will be explained in detail with reference to FIG. First, create a character type selection table 17 that includes a mixture of single digits and ¥ symbols.Step 6
1) The first character is read by the single character recognition unit 16 (step 62). Check if it is the last character (
Step 63). If there are still characters to be read, the process returns to step 62 and the characters are read one after another. In this way, when the amount field in FIG. 5(1) is read, the result shown in FIG. 5(2) is obtained =12-. Since the last character is similar to the ¥ symbol, it is determined that it is illegible or a ¥ symbol. The read result is stored in the first register (step 64).

Next, the character type selection table 17 containing only numbers is read (step 65), and the amount field shown in FIG. 5(1) is read (step 66). The result of reading to the end is shown in Figure 5 (3).
It is. This result is stored in the second register (step 68). Here, the reading control unit 12 knows from the result of FIG. 5 (2) that the first character excluding the blank is a ¥ symbol, and the reading results after that are based on the reading results of FIG. 5 (3). We will use it. That is, until the ¥ symbol is read, the characters within the thick line shown in FIG. 5(2) are used, and after the ¥ symbol is read, the characters within the thick line shown in FIG. 5(3) are used. According to such rules, the first Rujista and the second
The contents of the registers are combined to obtain the read result shown in FIG. 5(4) (step 69).

In the above example, after reading the entire amount field 51 using the character type selection table that includes numbers and the ¥ symbol, the amount field 51 is then read using the character type selection table that only includes numbers. , a number and ¥ for each character
It is clear that mixed reading of symbols and reading of only numbers will give exactly the same result.

FIG. 7 is an explanatory diagram of the operation of an optical character reading system showing still another embodiment of the present invention.

In FIG. 7(,), a 10 symbol for determining polarity is written in front of the number, and in FIG. 7(b), a - symbol for determining polarity is written in front of the number. These are the same symbols for samurai. Even when reading such a numerical field with an optical character reading device, the method shown in the embodiment of FIG. 1 and the embodiment of FIG. 6 of the present invention can be used as is.

It is generally known that the greater the number of character types that can be read, the lower the reading rate. In this embodiment, focusing on the fact that the ¥ symbol is added to the beginning of the number string, reading is performed in a mixed manner until the ¥ symbol is read, and after the ¥ symbol is read, character types containing only numbers are read. By reading, the reading rate is improved. That is, only the minimum necessary fields are read using mixed character types, and the remaining parts, that is, most of the fields, are read at a high precision reading rate.

〔Effect of the invention〕

As explained above, according to the present invention, when reading an amount field, if a ¥ symbol is written as the first character, the number can be read with a character type that only has numbers, resulting in extremely high accuracy. can be read with,
Furthermore, since it is also possible to check whether the first character is a ¥ symbol, it is possible to detect and correct incorrectly filled out forms.

[Brief explanation of drawings]

Fig. 1 is an operation flowchart of an optical character reading system showing an embodiment of the present invention, Fig. 2 is a block diagram of an optical character reading device to which the reading system of the present invention is applied, and Fig. 3 is applied to the present invention. Figure 4 is a diagram comparing the reading result in Figure 1 with the conventional result; Figure 5 is a diagram showing the reading result of an optical character reading method showing another embodiment of the present invention; Figure 6 is a diagram of the amount field; This figure is an operation flowchart of the reading method shown in FIG. 5, and FIG. 7 is an explanatory diagram of an optical character reading method showing still another embodiment of the present invention. 11: Format program, 12: Reading control section, 1
3: Scanning section, 14: Photoelectric conversion section, 15: Preprocessing section, 16
: 1 character recognition section, 17 2 character type selection tables. =16= Figure 1 Start Set a number and \ symbol in the starting character type selection table. Is one character recognized? 7 N □F%, h, 9 Fig. 2 Fig. 3 Fig. 4 Fig. 5 Fig. 6 At the beginning of the figure Chi and ¥ symbol set 1 One character has been recognized or set the character type containing only digits stored in the LUGISTA in the table 1 Character recognition Figure 7

Claims (1)

[Scope of Claims] 1. In an optical character reading method that controls character reading by a format program that provides information such as the position of writing characters on a form and the type of characters to be read in entry fields, specific symbols are placed before and after the number to be read. Check whether information indicating that the field is filled in has been added to the above format program, and if this information has been added, the numbers and the specific symbol are not read until the first specific symbol is read. An optical character reading method characterized by reading a mixture of character types, and after reading the first specific symbol, reading is performed using a numeric character type only. 2. The optical character reading system according to claim 1, wherein the specific symbol added after the numeral is read by reading the last character excluding a blank space using a character type that is a mixture of numerals and specific symbols. . 3. If the above specific symbol is added only before a number, read all fields with a character type that includes numbers and the specific symbol, then read all fields with a character type that only includes numbers, and combine the results of both readings. When obtaining the final reading result, set the priority of both until the specific symbol is read.
2. The optical character reading method according to claim 1, wherein the optical character reading method is switched after reading. 4. Optical characters according to claim 1, characterized in that in reading the mixed character types, if a specific symbol is not written, it is determined as a writing error and an error flag is added to the output code. Reading method.