JPH01217686A - Character reader - Google Patents

Abstract

PURPOSE:To improve a reading accuracy in a character recognizing processing by executing the character recognizing processing based on the character type respectively corresponding to a specific code stored in a second memory means and a code stored in a first memory means. CONSTITUTION:A control part 17 has a character type designating area 20 as a storing area for FC information and further has an a area 21 for a character code for storing the character code designated by the code of the character type designating area 20. At the time of recognizing the character, the character recognition processing based on the character type corresponding to the specific code stored in the second memory means 2 and the code stored in the first memory means 20 is executed by a recognition control means 17. Thereby, for instance, only a specific character type designated by a general character type and the specific code can be limited to a range to be read. Accordingly, the specific character type used for a document to be read is designated and an unused specific character type or the like can be eliminated from the range to be read, so that an erroneous reading ratio is lowered to realize the improvement of the reading accuracy.

Description

DETAILED DESCRIPTION OF THE INVENTION [Object of the Invention] (Industrial Application Field) The present invention relates to a character reading device that performs character recognition processing based on format control information.

(Prior Art) Conventionally, optical character reading devices (OCR) optically scan characters recorded on a form to be read, and the character pattern obtained by this scanning is recognized by a character recognition unit. Ru. The character recognition unit performs a process of matching a character pattern with dictionary information consisting of a group of standard character patterns prepared in advance, and performs a process of identifying a character pattern to be read.

By the way, in OCR, before starting the reading process, format control information (FC information) recorded on, for example, a control sheet is provided from the outside (for example, a user). F
The C information is various reading control information necessary to ensure high reading accuracy by limiting the reading range, such as the pitch between two characters of the character type to be read, the font, the reading line position of the form, and the size of the form. Among these, the FC information regarding the character type is particularly important information that limits the range of dictionary information necessary for character recognition processing in order to limit the character type to be read.

Conventionally, when setting character types (character set) based on FC information, for example, numbers (N), alphabets (A)
, Kana character (K), and symbols (special character types) such as r, J, r+J, etc., the corresponding character type is designated as a code. Specifically, for example, if you specify the symbol (N), the character recognition unit will recognize numbers and r, J, r+
Dictionary information corresponding to a symbol such as J is searched, and the character pattern to be read is recognized using this dictionary information. With such a method, the range to be read can be limited to only numbers and symbols, for example, but if there are many types of symbols, the range to be read is not so limited. In particular, the types of symbols that are actually used are often limited. For this reason, if symbols that are not used on a form are also designated by FC information, the rate of misreading will increase during character recognition processing, resulting in a relative decrease in reading accuracy.

(Problems to be Solved by the Invention) Conventionally, when setting the character type of FC information, character types such as symbols other than those to be read are also specified, and the range to be read becomes relatively wide. Therefore, there is a drawback that the rate of misreading increases during character recognition processing, and the reading accuracy decreases relatively.

An object of the present invention is to provide a character reading device that can accurately specify a character type that matches a reading target range when setting the character type of FC information and improve reading accuracy in character recognition processing. be.

[Structure of the Invention] (Means and Effects for Solving the Problems) The present invention provides a character reading device that performs character recognition processing based on format control information such as character types and dictionary information. A first memory means for storing a code indicating a specified range of each special character type consisting of symbols, and a second memory means for storing a specific code for specifying only a specific character type within the specified range of general character types or special character types. It is a device comprising means. During character recognition processing, the recognition control means executes character recognition processing based on character types corresponding to the specific code stored in the second memory means and the code stored in the first memory means. It turns out.

With such a configuration, it is possible to limit the reading target range to, for example, general character types and specific character types specified by a specific code. Therefore, it is possible to specify a specific character type used in a form to be read and exclude special character types that are not used from the reading range, thereby reducing the rate of misreading and improving reading accuracy.

(Example) The present invention will be described in detail below with reference to the drawings. FIG. 1 is a block diagram showing the configuration of an OCR according to the same embodiment. As shown in FIG. 1, OCR includes a quantization unit 10. Examination section 11. Preprocessing normalization unit 12. Feature extraction unit 13. Recognition unit 14. It is equipped with an output section 15 and a dictionary memory IB. The quantization unit 10 converts two images from an image signal obtained by scanning a document by a scanning system (not shown) (having a photoelectric conversion circuit).
This is a circuit that generates character pattern data (hereinafter referred to as character pattern) consisting of digitized signals. The cutoff section 11 is a circuit that detects and cuts out a character pattern for one character from a group of character patterns for one line obtained by the quantization section 10, for example. The preprocessing normalization unit 12 is a circuit that removes noise from the examined character pattern and normalizes the position, size, etc. The feature extraction unit 13 is a circuit that generates and outputs feature information necessary for identifying character patterns. Recognition part 1
4 is feature information corresponding to the character pattern and dictionary memory I
Matching processing with dictionary information stored in advance in B is executed, and the recognition result (corresponding character code) is output to the output unit 15. The output unit 15 is, for example, a circuit that outputs recognition results to a host computer.

The control unit 17 includes a microprocessor that controls the overall operation of the OCR, and is a circuit that performs registration (setting) processing of FC information related to the gist of the present invention. The control unit 17 includes a storage area for storing FC information given from the outside, and outputs this FC information to the recognition unit 14. The control unit 17 has an FC information storage area as shown in FIG.
) has a character type designation area 20 as shown in FIG. Furthermore, as shown in FIG. 10B or 2C, it has a character code area 21 that stores the character code designated by the code in the character type designation area 20.

Next, the operation of this embodiment will be explained. First, a document to be read is scanned by a scanning system (not shown), and the quantization unit 10
A binary character pattern is generated. This character pattern is subjected to inspection processing by the inspection section 11 into a character pattern for one character, and then subjected to normalization processing by the preprocessing normalization section 12. The feature extraction section 13 generates feature information necessary for character pattern identification and outputs it to the recognition section 14 .

The recognition unit 14 executes a matching process between the feature information and the dictionary information from the dictionary memory 1B, and determines and outputs the corresponding character code as a recognition result. Here, the recognition unit 14
performs character recognition processing based on dictionary information corresponding to the character type specified in the FC information output from the control unit I7.

The control unit 17 sets a designation bit in the corresponding area of the character type designation area 20 shown in FIG. 2(a) based on FC information provided from the outside, for example, from a control sheet.

Here, in the character type designation area 20, rNJ means a number, rAJ means an alphabetic character, and rKJ means a kana character. Furthermore, the "symbols A-Cl" are designated areas provided for each predetermined special character type.Specifically, the "symbols C" are, for example, "period," and "plus ten."
This is an area for specifying 26 kinds of special character types, including the special character types included in each of the symbol A and symbol B. Further, the character type designation area 20 is provided with an area "code" for storing a specific code for designating a specific character type.

Assume now that designation bits are set at rNJ and "symbol C" in the character type designation area 20 by external FC information. As a result, the control unit 17 sends an FC code to the recognition unit for specifying that the range to be read during recognition processing consists of 26 types of special characters designated by the numbers “0 to 9” and the “symbol C”. Output to 14. The recognition unit 14 performs recognition processing based on dictionary information corresponding to the FC code output from the control unit 17.

Here, it is assumed that designation bits are set in rNJ and "code" in the character type designation area 20 by external FC information. The control unit I7 controls the character type specification area 2.
A designation bit of 0 designates a character type consisting of numbers "0 to 9" and symbols such as "period" and "plus ten" designated by "code" as the range to be read. The control unit 17 controls the character code specified by the "code", for example, "period" (2E shown in FIG. 2(b)).
) and "plus ten" character code (2B shown in FIG. 2(C)) are generated and output to the recognition unit 14 as an FC code together with the numerical character code. As a result, the recognition unit 14
performs character recognition processing based on dictionary information consisting only of numbers and symbols such as "period" and "plus+" as the reading range.

In this way, in addition to the codes that specify the general character types rN, A, and Kl and the special character types of symbols A-Cl, a specific code for specifying a specific character type among the general character types or special character types is created. Depending on the setting method, the range of special character types can be limited to an appropriate range, for example. Therefore, for a form that only uses symbols such as "period,""plus+", etc., the corresponding special character type can be Recognition processing is performed based on dictionary information of character types. Therefore, it is possible to significantly reduce the misreading rate in recognition processing.

[Effects of the Invention] As described in detail above, according to the present invention, in a character reading device that performs recognition processing based on format control information, it is possible to accurately specify a character type that matches a reading target range. Therefore, it is possible to reduce the rate of misreading in character recognition processing and improve reading accuracy as a result.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of a character reading device according to an embodiment of the present invention, and FIGS. 2(a) to (c) are for explaining the contents of the storage area provided in the control unit of the embodiment, respectively. It is a conceptual diagram. 14... Recognition unit, 16... Dictionary memory, 17...
control section. Applicant's agent Patent attorney Takehiko Suzue Figure 1 Figure 2

Claims (1)

[Claims] In a character reading device that performs character recognition processing based on format control information such as character types and dictionary information provided from the outside, the format control information specifies each of general character types and special character types consisting of various symbols. a first memory means for storing a code indicating a range; and a memory means for specifying only a specific character type within the specified range of the general character type or the special character type specified by the code stored in the first memory means. A second section that stores a specific code.
memory means, and recognition control means for executing character recognition processing based on character types corresponding to the specific code stored in the second memory means and the code stored in the first memory means, respectively. A character reading device characterized by: