KR101015233B1 - Method, apparatus and computer-readable recording medium for recognizing character included in hangul document - Google Patents

Abstract

A method, apparatus, and computer readable recording medium for performing character recognition on a Hangul document represented by an image are disclosed. According to the present invention, there is provided a method for recognizing a character included in an Hangul document in an image format, the method comprising: (a) extracting one or more independent components that are not bonded to another component from a character included in an image; b) recognizing a component corresponding to a phoneme among the extracted independent components, (c) performing a primary character recognition based on the recognized phoneme, and (d) the Hangul character from the primary character recognition result Identifying document information for a document, and (e) performing secondary character recognition on characters that are not subject to primary character recognition based on the identified document information. do. According to the present invention, it is possible to improve the accuracy of character division and the speed of character recognition.

Character segmentation, character segmentation, character recognition, character recognition, document information

Description

METHOD, APPARATUS AND COMPUTER-READABLE RECORDING MEDIUM FOR RECOGNIZING CHARACTER INCLUDED IN HANGUL DOCUMENT}

The present invention relates to a method, an apparatus, and a computer-readable recording medium for performing character recognition on a Hangul document represented by an image. More specifically, in performing character recognition on a document, first, recognizable characters are recognized first. By identifying document information such as average height, width, and line spacing for each type of characters, and performing character segmentation on the basis of this, the accuracy of character segmentation can be improved and the speed of character recognition can be improved. An apparatus and a computer readable recording medium.

Recently, with the rapid spread of digital storage media, digitalization of documents that existed on the ground has been actively developed. This phenomenon is accelerating with the development of Optical Character Recognition (OCR), which is a technology for automatically recognizing characters appearing on the ground.

In order to recognize the characters included in the Hangul document represented by the image, in the past, a model was formed and recognized in units of Hangul characters. In other words, each character included in the document is compared with the Korean character model of 2350 characters, and the Korean character with high similarity is output as a result of recognition of the character. However, according to this, there is a problem that the speed of character recognition is very slow because a number of comparison process has to be performed, and the recognition rate decrease may occur frequently due to misrecognition because the difference between 2350 Korean characters is not obvious. There was only.

The method developed to solve this problem is the phoneme unit recognition method. In other words, the characters included in the document are not recognized in units of letters, but are recognized in units of phonemes and combined again to perform character recognition. According to this, since each phoneme included in the letter needs to be corresponded to about 24 standard phonemes, the phonemes can be clearly distinguished and the recognition rate of the character recognition can be improved.

A representative method of these methods is a method of dividing characters included in a document from each other, performing a phoneme division on the divided characters, and recognizing the divided characters.

However, according to this method, if an error occurs in the character division step, which is the first step of character recognition, there is a problem that can lead to misrecognition of the entire document. In addition, there is a problem in that the performance and speed of the overall character recognition are reduced by applying an unnecessary phoneme segmentation algorithm to a phoneme that can be easily segmented without performing the phoneme segmentation algorithm.

Therefore, it is necessary to develop a technology that can improve the performance and speed of the overall character recognition by increasing the accuracy of character division in the character recognition process and preventing the unnecessary segmentation algorithm from being performed.

Accordingly, an object of the present invention, in order to solve all the problems of the prior art as described above, the character recognition that can be easily recognized in performing the character recognition first recognize the document information through this, and then based on this character segmentation By performing the method and the final character recognition, the accuracy of character division can be improved to reduce the error rate of the overall character recognition.

In addition, another object of the present invention is to avoid unnecessary phoneme segmentation process by recognizing the phonemes that are not in contact with other phonemes and can be easily divided without applying a separate algorithm, thereby improving the speed of character recognition. It is to make it possible.

The characteristic structure of this invention for achieving the objective of this invention mentioned above, and realizing the characteristic effect of this invention mentioned later is as follows.

According to an embodiment of the present invention, a method for recognizing a character included in an Hangul document in an image format, comprising: (a) extracting one or more independent components that are not joined to another component from a character included in an image; (B) recognizing a component corresponding to a phoneme among the extracted independent components, (c) performing primary character recognition based on the recognized phoneme, (d) recognizing the primary character Identifying document information on the Hangul document from a result; and (e) performing secondary character recognition on characters not subject to the primary character recognition based on the identified document information. A method is provided.

According to another embodiment of the present invention, an apparatus for recognizing a character included in an Hangul document in an image format, the independent component for extracting one or more independent components that are not bonded to other components of the character included in the image An extraction unit, an independent component recognition unit for recognizing a component corresponding to a phoneme among the extracted independent components, and performing a primary character recognition on the basis of the recognized phoneme, and from the result of the primary character recognition, Provided is a device including a document information acquisition unit for acquiring document information and a final character recognition unit for performing secondary character recognition on characters that are not subject to the primary character recognition based on the identified document information.

According to still another embodiment of the present invention, an apparatus for recognizing a character included in an Hangul document in an image format, comprising: one or more independent character recognitions for characters included in an image, which are not bonded to other components; A document information acquisition unit for acquiring document information about the Hangul document based on a result of performing the second character recognition, and a second character for characters that are not subject to the first character recognition based on the identified document information An apparatus is provided that includes a final character recognition unit that performs recognition.

On the other hand, in addition to this, there is further provided a computer readable recording medium for recording another method, system and computer program for executing the method for implementing the present invention.

According to the present invention, since document information such as line spacing, font size, etc. is obtained through characterization that can be easily recognized, and character segmentation is performed based on the document information, the accuracy of character segmentation can be improved. Can be improved.

In addition, according to the present invention, since unnecessary phoneme segmentation is not applied to the easily recognized phoneme, it is possible to improve the speed of the overall character recognition.

DETAILED DESCRIPTION The following detailed description of the invention refers to the accompanying drawings that show, by way of illustration, specific embodiments in which the invention may be practiced. These embodiments are described in sufficient detail to enable those skilled in the art to practice the invention. It should be understood that the various embodiments of the present invention are different but need not be mutually exclusive. For example, certain features, structures, and characteristics described herein may be implemented in other embodiments without departing from the spirit and scope of the invention in connection with an embodiment. It is also to be understood that the position or arrangement of the individual components within each disclosed embodiment may be varied without departing from the spirit and scope of the invention. The following detailed description, therefore, is not to be taken in a limiting sense, and the scope of the present invention, if properly described, is defined only by the appended claims, along with the full range of equivalents to which such claims are entitled. Like reference numerals in the drawings refer to the same or similar functions throughout the several aspects.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those skilled in the art can easily implement the present invention.

1 is a diagram illustrating a configuration of an apparatus for performing character recognition on a document represented by an image according to an embodiment of the present invention.

As shown in FIG. 1, the character recognition apparatus 100 of the present invention includes an image information recognizer 110, an independent component extractor 120, an independent component recognizer 130, a document information acquirer 140, The character divider 150, the final character recognizer 160, the controller 170, and the communicator 180 may be included. According to an embodiment of the present invention, the character recognition apparatus 100 may include an image information recognizer 110, an independent component extractor 120, an independent component recognizer 130, a document information acquirer 140, and a text component. The installment unit 150, the final character recognition unit 160, the control unit 170, and the communication unit 180 may be program modules in which at least some of them are communicated with the character recognition apparatus 100 (however, in FIG. 1, image information). Recognition unit 110, independent component extraction unit 120, independent component recognition unit 130, document information acquisition unit 140, character divider 150, final character recognition unit 160, control unit 170 and The communication unit 180 is illustrated as being included in the character recognition apparatus 100). Such program modules may be included in the phoneme segmentation apparatus 100 as an operating system, an application program module, and other program modules, and may be physically stored on various known storage devices. In addition, these program modules may be stored in a remote storage device that can communicate with the phoneme splitting apparatus 100. Meanwhile, such program modules include, but are not limited to, routines, subroutines, programs, objects, components, data structures, etc. that perform particular tasks or execute particular abstract data types, which will be described later, according to the present invention. For reference, in the present specification, the character recognition apparatus 100 may include an image information recognizer 110, an independent component extractor 120, an independent component recognizer 130, a document information acquirer 140, and a character divider 150. ), The final character recognition unit 160, the control unit 170 and the communication unit 180 is assumed, but is not necessarily limited to this, the image information recognition unit 110, independent component extraction unit 120, independent Various modifications may be assumed, such as the component recognition unit 130 may be included in the document information acquisition unit 140.

The character recognition apparatus 100 according to an exemplary embodiment of the present invention may be directly included in an image data generating apparatus such as a scanner or a camera, or may be connected to the image data generating apparatus.

Character recognition apparatus 100 according to another embodiment of the present invention is a digital device such as a personal computer (for example, desktop computer, notebook computer, tablet computer, palmtop computer, etc.), workstation, PDA, web pad, mobile phone, etc. It may be included in, or communicate with a digital device through a communication network. Here, the communication network may be configured without regard to communication modes such as wired and wireless, and may include a local area network (LAN), a metropolitan area network (MAN), and a wide area network (WAN). It may be configured with a variety of communication networks.

Hereinafter, assuming that the original document in the form of an image shown in FIG. 2 is text recognized by the text recognition apparatus 100 of the present invention, each component included in the text recognition apparatus 100 will be described in detail.

The image information recognizing unit 110 according to an embodiment of the present invention receives image information including a character from a digital device such as a scanner or a camera, and performs a function of recognizing the received image.

The independent component extracting unit 120 according to an embodiment of the present invention performs a function of dividing and extracting an independent component that can be divided in text information included in the recognized image without using a separate algorithm. For example, FIG. 3 is a diagram illustrating an example of extracting an independent component that can be divided into a part of a document (the document of FIG. 2) recognized by the image information recognizing unit 110. It means a component that does not form at least one contact point in the relationship, that is, a component that does not contact other components. Components that are not in contact with other components can be easily extracted without separation from other components.

Referring to FIG. 3, it can be seen that the extraction is performed by the independent component extractor 120 for a part of the document of FIG. 2. For example, since the letter “t” and the letter “?” Included in the letters “Pew” are not joined to each other, they may be extracted as the independent components by the independent component extracting unit 120. On the other hand, in the letter "ker", since "k" and "k" are joined, the part "ker" itself may be an independent component extracted by the independent component extracting unit 120.

The independent component recognition unit 130 according to an embodiment of the present invention performs a function of recognizing components extracted by the independent component extraction unit 120. In this case, the independent component recognizer 130 may refer to a database (not shown) that stores information on phoneme, which is the minimum unit for representing phonemes in Korean characters. That is, the independent component recognition unit 130 performs phoneme recognition by comparing each of the independent components extracted by the independent component extraction unit 120 with information on the phoneme stored in advance. As a method of recognition, if the extracted independent component is compared with one or more phonemes stored in the database and there is a phoneme indicating similarity or more than a predetermined value, a method of recognizing the independent component as the phoneme may be used. If there is more than one phoneme indicating a similarity or more than a predetermined value, the phoneme showing the highest similarity may be determined as the recognition result of the independent component. Since the information on the phoneme is stored in the database, the phoneme component of the independent components extracted by the independent component extractor 120 may be recognized by the independent component recognizer 130.

4 shows components extracted by the independent component extractor 120 and recognition results of the independent component recognizer 130. For example, among the independent components extracted by the independent component extractor 120, FIG. In the case of the component "ker", since there is no similar entity among the information on the phonemes stored in the database, it is not recognized. In the case of the components "ㅁ", "ㄷ", and "ㅠ", there are similar entities. Recognized as "ㅁ", "方", "ㅠ" respectively. In the independent component recognition results shown in FIG. 4, the components indicated in black represent components that are not recognized, and the components indicated in red represent components recognized as a phoneme by the independent component recognition unit 130.

Next, the document information acquisition unit 140 according to an embodiment of the present invention performs character recognition by creating a character by combining the components recognized by the independent component recognition unit 130, through which the entire document Function to grasp document information. Here, the document information is a concept including the line spacing in the entire document, the average height and width of each type of letters (for example, font type).

FIG. 5 illustrates a process in which the document information acquisition unit 140 generates a character based on the result recognized by the independent component recognition unit 130 to perform character recognition. Indicates the process of identifying document information.

As shown in FIG. 5, the document information acquisition unit 140 may combine the phonemes recognized by the independent component recognition unit 130 to generate a completed character. For example, if there are results recognized in the order of "o", "-a", and "d", the document information acquisition unit 140 may combine them to produce a character of "". Character recognition is performed in this way, but the components that cannot be combined are left alone. For example, if there is a recognized result in order of "ㅁ", "ㄷ", "ㅠ", the combination of "ㅠ" and "ㅠ" produces the character "Pew", and the combination "ㅁ" The ingredients are just left. When the character recognition is completed, the document information acquisition unit 140 grasps information such as average height and width or line spacing for each type of characters based on the result. Specifically, as illustrated in FIG. 6, the averaged height and width of the characters included in the document may be grasped through the recognized characters. For example, a character recognized as "" can be used to determine the height and width of a character that contains both initial, neutral, and trailing characters, and a character recognized as "pew" can only contain initial and neutral. One can figure out the height and width of characters. In addition, the average line spacing of the entire document can be determined by measuring the interval between recognition results existing on different lines in the entire document. The character height and width or line spacing here may be in pixels.

Next, the character dividing unit 150 according to an embodiment of the present invention performs a function of dividing characters not yet recognized based on the document information acquired by the document information obtaining unit 140.

7 shows characters further divided by the character divider 150. In FIG. 7, a letter denoted by a red square represents a character recognized by the independent component extractor 120, the independent component recognizer 130, and the document information acquirer 140, and the character represented by a blue square represents a character divider ( 150 to indicate that the character is additionally divided.

In short, since the character segmentation according to the present invention is performed based on document information obtained through the above series of processes, the accuracy can be considerably increased. That is, the conventional character recognition through the phoneme segmentation is performed through the process of character segmentation, phoneme segmentation, and phoneme recognition, so that the character segmentation is performed based on the information of the entire document, and thus the accuracy of the character segmentation. Has a problem that the error rate of the overall character recognition is inevitably decreased, but according to the present invention, only the phonemes of the independent components are recognized and combined with each other, so that the easily recognizable characters are preferentially recognized. Since the document information is grasped and then the character segmentation is performed based on the document information, the error rate in the character segmentation can be greatly reduced, and the accuracy of the overall character recognition can be improved.

When the character division is additionally completed by the character division unit 150, the final character recognition unit 160 according to an embodiment of the present invention performs character recognition for the additionally divided characters. That is, character recognition is performed on characters represented by blue squares in FIG. 7. As the character recognition method, a conventional mesh recognition or a character recognition method through phoneme division may be used. Mesh recognition refers to a method of recognizing a whole character by character unit, and character recognition through phoneme division refers to a method of completing character recognition by dividing the phonemes included in a character through a predetermined method and recognizing it. Means. Regarding the separation of phonemes, those skilled in the art to which the present invention pertains to the separation of letters for printed Hangul character recognition disclosed in the well-known technical literature, the Korean Information Science Society, Vol. 17, No. 2, Technology, and the like. An example of a method of character recognition through phoneme division is as follows. First, classify the character type based on the presence or absence of the base, the shape and the approximate position of the vowel, the number of connected components (the connected component means a set of connected one pixels), and the position and size of each connected component. Do it. Then, the phoneme is divided by a predetermined algorithm for each type. Specifically, the phoneme without a linking component is first separated, and if there is a linking component, the phoneme separation is performed using the position and size information of each linking component and the like. When the phoneme division is completed through this process, character recognition is completed by combining the divided phonemes.

On the other hand, the control unit 170 according to an embodiment of the present invention is an image information recognition unit 110, independent component extraction unit 120, independent component recognition unit 130, document information acquisition unit 140, character divider 150, the function of controlling the flow of data between the final character recognition unit 160 and the communication unit 180.

In addition, the communication unit 180 according to an embodiment of the present invention may perform a function of allowing the character recognition apparatus 100 according to the present invention to communicate with an external device.

Embodiments according to the present invention described above may be implemented in the form of program instructions that may be executed by various computer components, and may be recorded in a computer-readable recording medium. The computer-readable recording medium may include program instructions, data files, data structures, etc. alone or in combination. Program instructions recorded on the computer-readable recording medium may be those specially designed and configured for the present invention, or may be known and available to those skilled in the computer software arts. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks and magnetic tape, optical recording media such as CD-ROMs, DVDs, and magneto-optical media such as floptical disks. media), and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, flash memory, and the like. Examples of program instructions include not only machine code generated by a compiler, but also high-level language code that can be executed by a computer using an interpreter or the like. The hardware device may be configured to operate as one or more software modules to perform the process according to the invention, and vice versa.

Although the present invention has been described by specific embodiments and drawings, which are limited to specific matters such as specific components, etc., this is merely provided to help a more general understanding of the present invention, and the present invention is not limited to the above embodiments. Those skilled in the art to which the present invention pertains can make various modifications and variations from this description.

Therefore, the spirit of the present invention should not be construed as being limited to the above-described embodiments, and all of the equivalents or equivalents of the claims, as well as the following claims, I will say.

1 is a diagram illustrating a configuration of an apparatus for performing character recognition on a document represented by an image according to an embodiment of the present invention.

2 is a diagram illustrating an example of a document to be recognized by a character recognition apparatus according to an embodiment of the present invention.

3 is a diagram illustrating an example of a process of extracting an independent component from a document by the independent component extractor 120 according to an exemplary embodiment of the present invention.

4 is a diagram illustrating an example of a process in which the independent component recognition unit 130 recognizes a phoneme with respect to the independent component extracted by the independent component extraction unit 120 according to an exemplary embodiment of the present invention.

5 is a diagram illustrating an example of a process in which character recognition is performed by the document information acquisition unit 140 according to an embodiment of the present invention.

6 is a diagram illustrating an example of a process of acquiring document information by the document information obtaining unit 140 according to an exemplary embodiment of the present invention.

7 is a diagram illustrating an example of character division performed by the character division unit 150 according to an embodiment of the present invention.

<Brief description of the major reference numerals>

110: image information recognition unit

120: independent component extraction unit

130: independent component recognition unit

140: document information acquisition unit

150: character division

160: the final character recognition unit

Claims (21)

A method for recognizing characters included in Hangul documents in image format. (a) extracting one or more independent components of the text contained in the image that are not joined to other components; (b) recognizing a component corresponding to a phoneme among the extracted independent components, (c) performing primary character recognition based on the recognized phonemes; (d) identifying document information on the Hangul document from the first character recognition result, and (e) performing secondary character recognition on characters not subject to the primary character recognition based on the identified document information How to include. The method of claim 1, In step (b), Comparing the one or more independent components with one or more graphemes, and Selecting a phoneme that exhibits a similarity or more as a result of the comparison as a result of recognition of the independent component How to include. The method of claim 2, In step (b), Selecting a phoneme with the highest similarity as a result of the comparison as a result of recognition of the independent component. The method of claim 1, In step (c), Outputting a character that can be produced by combining the recognized phonemes in order as a result of the primary character recognition. The method of claim 1, And the document information is at least one of an average height, width, and line spacing for each type of characters included in the Korean document. The method of claim 1, In step (e), Dividing each character not subject to the primary character recognition based on the identified document information; and Performing the secondary character recognition on the divided character How to include. The method of claim 6, The secondary character recognition is performed using a mesh recognition method. The method of claim 6, Performing the second character recognition, Performing a phoneme division on the divided characters, and Performing character recognition based on the phoneme division result How to include. A device for recognizing characters included in a Korean document in image format, Independent component extraction unit for extracting one or more independent components for the characters contained in the image, which are not joined to other components, Independent component recognition unit for recognizing a component corresponding to the element of the extracted independent components, A document information acquisition unit performing primary character recognition based on the recognized phoneme, and grasping document information about the Hangul document from the primary character recognition result; A final character recognition unit for performing a secondary character recognition for the characters that are not the object of the primary character recognition based on the identified document information / RTI &gt; 10. The method of claim 9, And the independent component recognizing unit compares the one or more independent components with one or more phonemes, and selects a phoneme having a similarity or more as a result of the comparison as a result of recognition of the independent component. The method of claim 10, And the independent component recognizing unit selects a phoneme having the highest similarity as a result of the recognition as a result of recognition of the independent component. 10. The method of claim 9, And the document information acquiring unit outputs, as a result of the primary character recognition, characters that can be formed by combining the recognized phonemes in order. 10. The method of claim 9, And the document information is at least one of an average height, width, and line spacing for each type of characters included in the Hangul document. 10. The method of claim 9, The apparatus may further include a character divider configured to divide each character that is not a target of the first character recognition based on the identified document information. The final character recognition unit, the apparatus for performing the secondary character recognition for the divided characters. The method of claim 14, The second character recognition is performed by a mesh recognition method. The method of claim 14, The secondary character recognition is performed through character recognition based on the phoneme segmentation and the phoneme segmentation result of the divided character. A device for recognizing characters included in a Korean document in image format, A document information acquisition unit which grasps document information on the Hangul document based on a result of performing a primary character recognition through recognition of one or more independent characters that are not bonded to other components of a character included in an image; A final character recognition unit for performing a secondary character recognition for the characters that are not the object of the primary character recognition based on the identified document information / RTI &gt; The method of claim 17, The result of performing the primary character recognition is a character that can be created by combining the recognized independent characters in order. The method of claim 17, And the document information is at least one of an average height, width, and line spacing for each type of characters included in the Hangul document. The method of claim 17, The apparatus may further include a character divider configured to divide each character that is not a target of the first character recognition based on the identified document information. The final character recognition unit, the apparatus for performing the secondary character recognition for the divided characters. A computer readable recording medium having recorded thereon a computer program for executing the method according to any one of claims 1 to 8.

Images