JP5217814B2 - Image processing apparatus, image processing method, control program, and recording medium - Google Patents

Description

  The present invention relates to an image processing apparatus, an image processing method, a control program, and a recording medium, and more particularly to processing for storing character information displayed as an image as character information to be searched.

  In recent years, computerization of information has been promoted. As one aspect of information digitization, there is a process of acquiring an image including a sentence by scanning or the like and recognizing characters displayed on the image as character information. Such a process is called OCR (Optical Character Recognition).

  When the OCR process is performed on an image including a sentence composed of a plurality of lines, a process for extracting a character line area is required as a process prior to the execution of the OCR process. As a method for extracting such a character line region, for example, there is a method of extracting lines by connecting respective characters. As another method, a method of extracting lines by binarizing an input sentence image and performing logical sum reduction and determining an aspect ratio of a connected component included in the reduced image is proposed. (For example, refer to Patent Document 1).

A method for extracting a character line region that is difficult to extract under a single extraction condition in an image having a complicated document layout has also been proposed (for example, see Patent Document 2). In the method disclosed in Patent Document 2, a plurality of character line regions are extracted based on a plurality of different extraction conditions. Then, the character line area to be adopted is determined based on the character recognition results for the plurality of extracted character line areas.
JP-A-5-266075 JP 2006-244309 A

  Conventionally, the character recognition technology as described above has been a problem of mainly reproducing an image including characters as character information, that is, improving reproducibility. However, in recent years, with the importance of information retrieval technology, there is a demand for storing image information including characters as a character retrieval target by character recognition technology. In this case, the precision of extracting characters included in the image based on the input search word is more important than the above reproducibility.

  The precision can be improved by accurately extracting a character line area included in an image and accurately executing character recognition. On the other hand, when the character line area is not accurately extracted, the text included in the image cannot be accurately recognized. As a result, since characters included as an image cannot be stored as character information, a correct search result cannot be obtained even if a character search is executed, and the relevance rate decreases.

  In other words, when an image including characters is stored as a search target for character search by performing character recognition, the reproducibility and the relevance rate are not traded off, and the relevance rate is given priority.

  The present invention has been made in consideration of the above circumstances, and an object of the present invention is to improve the accuracy of character search when an image containing characters is stored as a character search search target by character recognition. .

In order to solve the above-described problem, the present invention provides an image processing apparatus that extracts character regions from an image including character regions, generates character information, and stores the character information as search target information. A first character region extraction unit that extracts a character region from the acquired image based on a first region extraction method for extracting the character region and generates first character information; A second character region extraction unit that extracts a character region from the acquired image based on a second region extraction method different from the first region extraction method and generates second character information; the first character information and the second text information see contains an information storage unit for storing a recording medium as the search target information, the first character region and the second to the first character region extracting unit has extracted The second character extracted by the character area extraction unit A storage target exclusion unit that excludes the character information displayed in the overlapping area from the storage target as the search target information when the area overlaps, If the entire first character area is included in the second character area, the first character information is excluded from the storage target .

According to the present invention, in the above-described image processing apparatus, the first character area extracted by the first character area extraction unit and the second character area extracted by the second character area overlap. A storage target exclusion unit that excludes the character information displayed in the overlapping area from the storage target as the search target information, and the storage target exclusion part includes: The entire range in the horizontal direction is included in the entire range in the horizontal direction of the second character area, and the direction of the character line in the first character area and the direction of the character line in the second character area are predetermined conditions. If the condition is satisfied, the character information displayed in the overlapping area of the first character information is excluded from the storage target.

According to the present invention, in the above-described image processing apparatus, the storage target exclusion unit is configured such that the direction of the character line in the first character area and the direction of the character line in the second character area are horizontal. The character information displayed in the overlapping area of the first character information is excluded from the storage target.

According to the present invention, in the above-described image processing apparatus, the first character area extracted by the first character area extraction unit and the second character area extracted by the second character area overlap. A storage target exclusion unit that excludes the character information displayed in the overlapping area from the storage target as the search target information, and the storage target exclusion part includes: The entire range in the vertical direction is included in the entire range in the vertical direction of the second character area, and the direction of the character line in the first character area and the direction of the character line in the second character area are predetermined conditions. If the condition is satisfied, the character information displayed in the overlapping area of the first character information is excluded from the storage target.

According to the present invention, in the above-described image processing device, the storage target exclusion unit is configured such that the direction of the character line in the first character area and the direction of the character line in the second character area are vertical. The character information displayed in the overlapping area of the first character information is excluded from the storage target.

The present invention includes the dictionary information acquisition unit that acquires dictionary information and the dictionary information acquired by searching the first character information or the second character information in the image processing apparatus described above. A dictionary search unit for extracting words; and a storage target exclusion unit for excluding the first character information or the second character information from the storage target as the search target information according to a search result by the dictionary search unit. It is characterized by having.

Further, the present invention is an image processing apparatus described above, the storing object excluding unit of the first character information and the second character information, the dictionary search unit search results extracted word all by What is a single character word is excluded from the storage target.

Further, according to the present invention, in the above-described image processing apparatus, character information displayed in the character area extracted by the first character area extraction unit or the second character area extraction unit is stored as the search target information. A storage target exclusion unit that is excluded from the target, and the storage target exclusion unit is displayed in the character area based on a relationship between a line direction of the character area and a width of the character area in a direction parallel to the line direction. The character information is excluded from the storage target.

Further, the present invention is an image processing apparatus described above, the storage exempt unit, when the a width of the character region in the row direction parallel to the direction of the character region width is equal to or smaller than the predetermined range, Character information displayed in the character area is excluded from the storage target.

Further, according to the present invention, in the above-described image processing apparatus, character information displayed in the character area extracted by the first character area extraction unit or the second character area extraction unit is stored as the search target information. A storage object exclusion unit for excluding the object from the object, wherein the storage object exclusion unit stores the character information displayed in the character area based on a relationship between a line direction of the character area and an aspect ratio of the character area It is excluded from the object.

Further, the present invention is an image processing method for extracting the character region from an image including the character region, generating character information, and storing it as search target information, wherein the image acquisition unit acquires the image, 1 of the character area extraction unit to generate a first character information by extracting a character region from the acquired image based on the first area extracting method for extracting the character area, second character region extraction unit, wherein generating the second character information by extracting a character region from the acquired image on the basis of a different second area extracting process and the first region extraction method, the information storage unit, The first character information and the second character information are stored in the recording medium as the search target information , and the storage target exclusion unit includes the first character region extracted by the first character region extraction unit and the first character region. The second character area extracted by the second character area extraction unit is In the case of duplication, the character information displayed in the overlapping area is excluded from the storage target as the search target information, and the entire first character area is included in the second character area. The first character information is excluded from the storage target .

The present invention also provides a control program that causes an information processing apparatus to operate as an image processing apparatus that extracts character areas from an image including character areas, generates character information, and stores the information as search target information. acquiring, and generating a first character information by extracting a character region from the acquired image based on the first area extracting method for extracting the character region, the first generating a second character information by extracting a character region from the acquired image on the basis of a different second area extracting process from the region extraction method, the first character information and the second letter and storing the information in the recording medium as the search target information, in the first of the first character region generating a second character information extracted in the step of generating the character information If the extracted second character area overlaps, the step of excluding the character information displayed in the overlapped area from the storage target as the search target information, and the first character When the entire area is included in the second character area, the information processing apparatus is caused to execute the step of excluding the first character information from the storage target .

Further, the present invention is a recording medium, wherein the control program described above is recorded in a format readable by the information processing apparatus.

  ADVANTAGE OF THE INVENTION According to this invention, when storing the image containing a character as a search object of a character search by recognizing a character, it becomes possible to improve the precision of a character search.

Embodiment 1 FIG.
Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. In the present embodiment, an image processing apparatus that scans the current to generate image information, performs character recognition on the image information, and stores the image information and the character information in association with each other will be described as an example.

  FIG. 1 is a block diagram illustrating a hardware configuration of the image processing apparatus 1 according to the present embodiment. As shown in FIG. 1, the image processing apparatus 1 according to the present embodiment includes the same configuration as an information processing terminal such as a general server or a PC (Personal Computer). That is, the image processing apparatus 1 according to the present embodiment includes a CPU (Central Processing Unit) 10, a RAM (Random Access Memory) 20, an engine 30, a ROM (Read Only Memory) 40, and an NVRAM (Non-Volatile Random Access Memory) 50. And the I / F 60 are connected via a bus 90. Further, an LCD (Liquid Crystal Display) 70 and an operation unit 80 are connected to the I / F 60.

  The CPU 10 is a calculation unit and controls the operation of the entire image processing apparatus 1. The RAM 20 is a volatile storage medium capable of reading and writing information at high speed, and is used as a work area when the CPU 10 processes information. The engine 30 is an image processing engine that executes an image processing function such as a scanner or a printer. The ROM 40 is a read-only nonvolatile storage medium and stores a program such as firmware.

  The NVRAM 50 is a non-volatile storage medium capable of reading and writing information, and stores information such as an OS (Operating System), various control programs, application programs, and operation parameters of the application programs. The I / F 60 connects and controls the bus 90 and various hardware and networks. The LCD 70 is a visual user interface for the user to check the state of the image processing apparatus 1. The operation unit 80 is a user interface for a user to input information to the image processing apparatus 1 such as a keyboard and a mouse.

  In such a hardware configuration, a program stored in a storage medium such as ROM 40, NVRAM 50, HDD (Hard Disk Drive) (not shown) or an optical disk is read into RAM 20, and CPU 10 operates according to the program read into RAM 20. Thus, a software control unit is configured. A functional block that realizes the functions of the image processing apparatus 1 according to the present embodiment is configured by a combination of the software control unit configured as described above and hardware.

  FIG. 2 is a block diagram illustrating a functional configuration of the image processing apparatus 1 according to the present embodiment. As shown in FIG. 2, the image processing apparatus 1 according to the present embodiment includes a controller 100, an ADF (Auto Document Feeder) 101, a scanner unit 102, a paper discharge tray 103, a network I / F 104, a display panel. 105 and a search target DB 106.

  The controller 100 includes a main control unit 111, an engine control unit 112, an input / output control unit 113, an image processing unit 114, and a character recognition control unit 120. In FIG. 2, the electrical connection is indicated by solid arrows, and the flow of paper is indicated by broken arrows. The controller 100 is realized by the CPU 10 operating according to a program loaded in the RAM 20 shown in FIG.

  The network I / F 104 is an interface used when the image processing apparatus 1 communicates with other devices such as a client terminal. The network I / F 104 is realized by the I / F 60 shown in FIG. The display panel 105 is an output interface that visually displays the state of the image processing apparatus 1, and is also an input interface when the user directly operates the image processing apparatus 1 as a touch panel. The display panel 105 is realized by the LCD 70 and the operation unit 80 shown in FIG.

  The search target DB 106 is a database that stores information to be searched for characters. The search target DB 106 is realized by information stored in a nonvolatile storage medium such as the NVRAM 50 shown in FIG. 1 or an HDD (not shown).

  The main control unit 111 plays a role of controlling each unit included in the controller 100, and gives a command to each unit of the controller 100. The engine control unit 112 serves as a driving unit that controls or drives the scanner unit 102 and the like. The input / output control unit 113 inputs information input via the network I / F 104 or the display panel to the main control unit 111. In addition, the input / output control unit 113 transmits information to other devices connected to the network via the network I / F 104 in accordance with instructions from the main control unit 111. Further, the input / output control unit 113 controls display on the display panel 105 in accordance with a command from the main control unit 111.

  The image processing unit 114 processes image data input from the scanner unit 102 under the control of the main control unit 111 and generates image information. The image information is information stored in a HDD (not shown) as a result of the scanner operation or transmitted to a client terminal or the like via the network I / F 104.

  The character recognition control unit 120 analyzes image information input as a registration target in the search target DB 106 according to the control of the main control unit 111 and executes character recognition. Then, the character recognition control unit 120 associates the input image information with the character information generated by the character recognition and stores them in the search target DB 106 in accordance with the result of the executed character recognition. The function provided by the character recognition control unit 120 is one of the gist of the present embodiment.

  When the image processing apparatus 1 operates as a scanner, the input / output control unit 113 performs main operations in response to a user operation on the display panel 105 or a scan execution instruction input from an external client device or the like via the network I / F 104. A scan execution signal is transferred to the control unit 111. The main control unit 111 controls the engine control unit 112 based on the received scan execution signal.

  The engine control unit 112 drives the ADF 101 and conveys the document to be imaged set on the ADF 101 to the scanner unit 102. In addition, the engine control unit 112 drives the scanner unit 102 and images a document conveyed from the ADF 101. If no original is set on the ADF 101 and the original is directly set on the scanner unit 102, the scanner unit 102 images the set original under the control of the engine control unit 112. That is, the scanner unit 102 operates as an imaging unit.

  In the imaging operation, an imaging element such as a CCD included in the scanner unit 102 optically scans the document, and imaging information generated based on the optical information is generated. The engine control unit 112 transfers the imaging information generated by the scanner unit 102 to the image processing unit 114. The image processing unit 114 generates image information based on the imaging information received from the engine control unit 112 according to the control of the main control unit 111.

  Image information generated by the image processing unit 114 is stored in a storage medium attached to the image forming apparatus 1 such as an HDD. The image information generated by the image processing unit 114 is stored in the HDD or the like as it is according to a user instruction or transmitted to an external device via the input / output control unit 113 and the network I / F 108.

  1 and 2 exemplify a case where the image processing apparatus 1 is configured by a single apparatus. However, the functions shown in FIGS. 1 and 2 can be realized by dividing the apparatus into a plurality of devices connected by a network. In such an image processing apparatus 1, as described above, the gist of the present embodiment is the character recognition process performed by the character recognition control unit 120. The character recognition control unit 120 according to the present embodiment will be described in more detail with reference to FIG.

  FIG. 3 is a block diagram illustrating details of the character recognition control unit 120 according to the present embodiment and a connection relationship between the character recognition control unit 120, the main control unit 111, and the search target DB 106. As illustrated in FIG. 3, the character recognition control unit 120 according to the present embodiment includes an input information acquisition unit 121, a character information recognition unit 122, and an information registration unit 123. The information insertion control unit 120 according to the present embodiment is configured by the CPU 10 operating according to a program loaded in the RAM 20 shown in FIG.

  The input information acquisition unit 121 acquires image information to be registered in the search target DB 120 as a search target as input information. The input information is image information generated by scanning processing by the scanner unit 102 or image information input via the network I / F 104. The input information acquisition unit 121 acquires the input information from the main control unit 111. The input information acquisition unit 121 inputs the acquired input information to the character information recognition unit 122.

  The character information recognition unit 122 recognizes character information included in the input information acquired by the input information acquisition unit 121. The character information recognition unit 122 performs character recognition after extracting a character line area where a character line is displayed from image information acquired as input information. The character information recognition unit 122 has a plurality of methods for extracting the character line region, and performs character recognition after extracting a character line based on each method. The processing by the character information recognition unit 122 will be described in detail later.

  The information registration unit 123 associates the image information acquired by the input information acquisition unit 121 with the character information generated by the character information recognition unit 122 through character recognition, and registers them in the search target DB 106.

  Next, the operation of the image processing apparatus 1 according to this embodiment will be described. FIG. 4 is a flowchart showing the operation of the image processing apparatus 1 according to the present embodiment. As shown in FIG. 4, first, the input information acquisition unit 121 acquires image information as input information from the main control unit 111 (S401). That is, the input information acquisition unit 121 functions as an image acquisition unit. FIG. 5 shows an example of an image acquired by the input information acquisition unit 121 in S401 of the present embodiment. The input information shown in FIG. 5 is image information in which sentences are displayed in black characters on a white background.

  As described above, the image information acquired by the input information acquisition unit 121 in step S401 is input by scanning with the scanner unit 102 or inputting information via the network I / F 104. The input information acquisition unit 121 inputs the acquired input information to the character information recognition unit 122.

  When the input information as shown in FIG. 5 is acquired, the character information recognition unit 122 selects one method from a plurality of line extraction methods (S402). Then, the character information recognition unit 122 executes line extraction based on the line extraction method selected in S402 (S403).

  First, in the process of S402, a case will be described in which black pixels are connected to generate a circumscribed rectangle of black pixels, and a method of extracting the rows by connecting the generated circumscribed rectangles in the horizontal direction is selected. In this case, the character information recognition unit 122 recognizes continuous black pixels in the input information in the process of S403. Then, the character information recognition unit 122 extracts a circumscribed rectangle of the recognized black pixel. FIG. 6 is a diagram illustrating an aspect of rectangle extraction processing by the character recognition unit 122.

  When the circumscribed rectangle is extracted as illustrated in FIG. 6, the character information recognition unit 122 extracts the character line by connecting the extracted circumscribed rectangle in the horizontal direction. FIG. 7 shows an example of the result of extracting horizontal writing lines based on the circumscribed rectangle extraction result shown in FIG. When the extraction of the character line is completed, the character information recognition unit 122 performs character recognition on the extracted character line (S404). In step S404, the character information recognition unit 122 extracts text data corresponding to the character image by feature extraction and pattern dictionary collation, and executes character recognition. In the character recognition process, a known technique can be used, and detailed description thereof is omitted in this embodiment.

  FIG. 8 is a diagram showing character information generated as a result of performing character recognition on the extracted character line as shown in FIG. As shown in FIG. 8, when character recognition is performed on the extracted character line as shown in FIG. 7, “character recognition technology is one of the technologies for digitizing paper data. Will not be smaller. ”Will be generated. Thus, first, the character information recognition unit 122 functions as a first character region extraction unit.

  When character recognition is performed and character information as shown in FIG. 8 is generated, the character information recognition unit 122 selects all executable line extraction methods and confirms whether the processes of S403 and S404 have been performed (S405). ). If not all line extraction methods have been selected (S405 / NO), the character information recognition unit 122 repeats the processing from S402.

  FIG. 9 is a diagram illustrating an example of a row extraction result when a vertically written row extraction method is selected in S402. In this case, the character information recognition unit 122 connects the circumscribed rectangles extracted as shown in FIG. 6 in the vertical direction, and extracts character lines as shown in FIG. 9 (S403). When the extraction of the character line is completed, the character information recognition unit 122 performs character recognition on the extracted character line in the same manner as described above (S404).

  FIG. 10 is a diagram showing character information generated as a result of performing character recognition on the extracted character line as shown in FIG. As shown in FIG. 10, when character recognition is performed on the extracted character line as shown in FIG. The text information “is in a sentence” is generated. Thus, as a result of selecting another line extraction method in S402, the character information recognition unit 122 functions as a second character region extraction unit.

  As described above, by repeating the processes of S403 and S404, the same image area is recognized as a different character line area, and different character information is generated. When all of the executable line extraction methods are selected and the processing of S403 and S404 is completed (S405 / YES), the character information recognition unit 122 displays the image information acquired from the input information acquisition unit 121 and FIGS. 10, the generated character information is input to the information registration unit 123.

  The information registration unit 123 registers the information input from the character information recognition unit 122 in the search target DB 106 (S406), and ends the process. That is, the information registration unit 123 functions as an information storage unit that stores information in the search target DB 106. In S <b> 406, the information registration unit 123 registers the image information acquired by the input information acquisition unit 121 in S <b> 401 and the character information generated by the character information recognition unit 122 in association with each other in the search target DB 106. FIG. 11 shows an example of information stored in the search target DB 106.

  As shown in FIG. 11, the search target DB 106 stores information relating to image storage paths and character information 1, 2, 3,. The image storage path is information on the storage path of the image information acquired by the input information acquisition unit 121 in S401. Character information 1, 2, 3,... Is character information generated by the character information recognition unit 122. When the search target DB 106 is searched, character information 1, 2, 3,... Is a character search target. As described above, the character information recognition unit 122 extracts a character line region based on a plurality of different line extraction methods, and generates a plurality of character information. Therefore, as shown in FIG. 11, a plurality of character information is stored in association with the image information.

  8 and 10, it is clear that the line extraction method according to FIG. 8 is accurate and the line extraction method according to FIG. 10 is inaccurate, and can be easily determined. However, it is difficult for the information processing apparatus to make such a determination. Further, the above determination becomes more difficult for a manuscript in which vertical writing, horizontal writing, and images are mixed as shown in FIG.

  Various methods for making the above determination have been proposed, but their accuracy is not sufficient. Accordingly, there are many cases where an erroneous character line area is employed, and in such a case, erroneous character information as shown in FIG. 10 is employed. As a result, since the accurate character information shown in FIG. 8 is not a search target, a search omission occurs, and the relevance rate decreases.

  In the present embodiment, as shown in FIG. 11, all character information generated by applying a plurality of line extraction methods is stored as a search target. Thereby, registration of accurate character information is not mistakenly excluded, and it becomes possible to improve the search relevance rate when registering search target information.

  As described above, when the image processing apparatus 1 according to the present embodiment stores an image including characters as a character search target by recognizing the characters, it is possible to improve the character search relevance rate. Become. Further, according to the above embodiment, extra character information as shown in FIG. 10 is also registered as a search target. However, as shown in FIG. 10, character information extracted by an incorrect line extraction method is information that does not hold as a sentence or a word. Therefore, the search noise does not increase during the search, and no problem occurs for the purpose of registering the search target information.

  In the above embodiment, as an example of the type of the line extraction process selected in S402, as described in FIGS. 7 and 9, the line extraction process having different line directions has been described as an example. In addition, in order to extract accurate character information, it is preferable that the number of types of line extraction processing that can be selected by the character information recognition unit 122 in S402 is as many as possible.

  For example, in the above-described embodiment, as an example of the process of S403 in FIG. 4, a character line is extracted by connecting black pixels in a monochrome image. In addition, when the input image is a color image, a pixel set may be generated by connecting pixels having similar colors, and a character line may be extracted by connecting the generated pixel set. That is, as a variation of the line extraction method, a mode in which a method for recognizing lines constituting characters is different can be considered.

  Further, in the above embodiment, for example, a character direction can be considered in addition to the row direction described in FIGS. FIG. 13 is a diagram illustrating an example of a layout on a paper surface as seen in a design drawing or the like. Among the character information included in FIG. 13, the character information “90 mm in height” has the character direction rotated by 270 °. Thus, as a variation of the line extraction method, there can be considered a mode in which the direction of recognizing the upper and lower sides of the character is different.

  The line direction of the character information “height 90 mm” is vertical, but is a special line direction that is read from the bottom to the top. That is, depending on the layout format, there are various directions in which characters are read in addition to the line direction (hereinafter referred to as the character reading direction). Therefore, as a variation of the line extraction method, a mode in which the direction in which characters are read is different can be considered.

  Similarly, even when a row is extracted as shown in FIG. 9, it is possible to read from left to right instead of reading from right to left. That is, depending on the layout format, there are various directions for reading a line (hereinafter referred to as a line reading direction). Therefore, as a variation of the line extraction method, a mode in which the direction in which the line is read is different can be considered.

  In order to correspond to the various layout formats as described above, variation information of the line extraction method using the line direction, the character direction, the character reading direction, the line reading direction, and the like as parameters is registered in the character information recognition unit 122. Is preferred. In S402, the character information recognition unit 122 can perform character recognition in various modes by selecting one of the registered line extraction method variations.

  FIG. 14 shows an example of variation information of the line extraction method registered in the character information recognition unit 122. As illustrated in FIG. 14, the association information registered in the character information recognition unit 122 includes information on a line direction, a character direction, a character reading direction, and a line reading direction. The row direction is indicated as “vertical” or “horizontal”. The character direction is indicated by the rotation angle of the character such as “0 °” and “270 °”. If it is a character as shown in FIG. 5, it is “0 °”, and if it is “270 °”, it indicates that it is rotated like the character “height...” In FIG. In the character reading direction and the line reading direction, the direction in which the character or line is read is indicated by an arrow.

  In the example of FIG. 14, for example, the variation information with ID “001” indicates that the row direction is vertical, and the character is not rotated like the character “height...” In FIG. The character reading direction is from top to bottom, and the line reading direction is from right to left.

Embodiment 2. FIG.
In the first embodiment, as illustrated in FIG. 4, an example in which the precision is improved by registering all character information generated based on a plurality of line extraction methods in the search target DB 106 has been described. As described above, the character information generated and registered by the wrong line extraction method does not adversely affect the search result. However, an extra storage area is required due to character information registered by an incorrect line extraction method. In the present embodiment, an example will be described in which the extra storage area is reduced by reducing the character information registered by the wrong line extraction method while maintaining the search precision. In addition, the welfare which attaches | subjects the code | symbol same as the code | symbol demonstrated in Embodiment 1 shows the same or an equivalent part, and abbreviate | omits description.

  The image processing apparatus 1 according to the present embodiment has a configuration that is substantially the same as the mode described in FIGS. 1, 2, and 3. The image processing apparatus 1 according to the present embodiment is different in information registration processing in the search target DB 106 by the character recognition control unit 120. With reference to the figure, the information registration process to the search target DB 106 by the character recognition control unit 120 will be described.

  FIG. 15 is a flowchart showing the operation of the character recognition control unit 120 according to this embodiment. As shown in FIG. 15, from S1501 to S1505, processing is executed in the same manner as S401 to S405 described in FIG. When the character recognition is completed for the entire line extraction method (S1505 / YES), the character information recognition unit 122 deletes the character information generated by the misrecognition (S1506). The processing in S1506 is processing for excluding character information generated by misrecognition from being stored in the search target DB. That is, the character information recognition unit 122 functions as a storage target exclusion unit.

  Details of the processing executed by the character information recognition unit 122 in S1506 will be described with reference to FIG. FIG. 16 is information indicating a list of means (hereinafter referred to as erroneous detection area deletion means) used when the character information recognition unit 122 deletes character information generated by erroneous recognition in S1506. As shown in FIG. 16, the character information recognition unit 122 according to the present embodiment includes an overlapping area deleting unit and a dictionary search deleting unit.

  First, the overlapping area deleting unit will be described. The overlapping area deleting means deletes overlapping character areas based on a predetermined condition when a plurality of character areas extracted by different line extraction methods overlap. With reference to FIGS. 17 (a) to 17 (d), a description will be given of the overlapping area deleting sun by the overlapping area deleting means.

  FIG. 17A is a diagram illustrating a case where the character area B is included in the character area A. FIG. In the case shown in FIG. 17A, the character information recognition unit 122 deletes the character area B indicated by diagonal lines if the line direction and the character direction of the character area A and the character area B are the same. Only character information is registered in the search target DB 106. If the line direction and the character direction of the character area A and the character area B are the same, the character information included in the character area B is also included in the character area A. Therefore, by registering the character information of the character region A, the character information of the deleted character region B is also registered, so that the matching rate does not decrease.

  FIG. 17B is a diagram illustrating a case where a part of the character area C and a part of the character area D overlap. In the example of FIG. 17B, the horizontal width of the character area C is wider than the horizontal width of the character area D, and the horizontal range of the character area D is included in the horizontal range of the character area C. In this case, if both the line direction of the character area C and the character area D are horizontal directions, the character information recognizing unit 122 deletes all of the character area C and a part in which the hatched area is deleted from the character area D. Register with.

  If the line directions of the character area C and the character area D are both horizontal, by registering all the character information of the character area C, all of the overlapping character information indicated by hatching is included. Therefore, by registering the entire character region C, the deleted hatched character information is also registered, so the precision does not decrease.

  FIG. 17C is a diagram illustrating a case where a part of the character area E and a part of the character area F overlap. In the example of FIG. 17C, the vertical width of the character area E is wider than the vertical width of the character area F, and the vertical range of the character area F is included in the vertical range of the character area E. Yes. In this case, the character information recognizing unit 122 determines that the character area E and the character area F are all deleted from the character area E and the area indicated by the diagonal lines if the line direction of the character area E and the character area F is the vertical direction. Register with.

  If the line direction of the character area E and the character area F are both vertical, by registering all the character information of the character area E, all of the overlapping character information indicated by diagonal lines is included. Therefore, by registering the entire character area E, the deleted hatched character information is also registered, so the precision does not decrease.

  FIG. 17D is a diagram illustrating a case where a part of the character area G and a part of the character area H overlap. In the example of FIG. 17D, the character region G and the character region H are overlapped with the vertical and horizontal directions shifted from each other. In this case, the character information recognition unit 122 registers the character information of the character region G and the character information of the character region H, respectively.

  In the case of FIG. 17D, for example, a case where the character information of the entire character region G and the portion where the hatched portion is deleted from the character region H is registered. In this case, even if the line direction of the character region G and the character region H is both horizontal, by deleting the hatched portion from the character region H, the sentence between the hatched portion and the region on the right side thereof is cut. . On the other hand, even if the line directions of the character region G and the character region H are both vertical, by deleting the hatched portion from the character region H, the sentence between the hatched portion and the region below it is cut. . Accordingly, in the overlap mode as shown in FIG. 17D, that is, in a mode where the horizontal direction and the vertical direction of one region and the other region are shifted and overlapped, the overlap region is not deleted.

  As described above, in the overlapping area deleting unit according to the present embodiment, the entire vertical or horizontal range of one area is included in the vertical or horizontal range of the other area, and the row direction is predetermined. Only when the condition is met, the overlapping area is deleted. In other words, in the present embodiment, the overlapping area is deleted only when the character line is not divided even if it is deleted. As a result, it is possible to delete only the unnecessary overlapping area and reduce the storage area required while maintaining the matching rate during the search.

  Next, dictionary search deletion means will be described. The dictionary search / deletion means performs a dictionary search on character information generated as a result of character recognition, and deletes character information in which no word is hit. Therefore, the character information recognition unit 122 according to this embodiment is configured to be accessible to the dictionary information DB 107 as shown in FIG. That is, the character information recognition unit 122 functions as a dictionary information acquisition unit and a dictionary search unit. The dictionary search / deletion means executes a dictionary search for each character area as shown in FIG.

  For example, as described with reference to FIG. 9, it is considered that the character information as shown in FIG. 10 generated by performing line extraction does not often include words registered in the dictionary. On the other hand, single-character words such as “day”, “now”, and “sentence” are included even in character information generated by an incorrect line extraction method. Therefore, the dictionary search / deletion unit deletes a character area that does not include two or more words among the words registered in the dictionary information DB 107.

  As described above, the dictionary search / deletion unit according to the present embodiment deletes a character area that does not include a word registered in the dictionary. As a result, it is possible to obviously delete unnecessary character areas and reduce the required storage area while maintaining the matching rate at the time of search.

  When the character information recognizing unit 122 deletes the character information of the clearly unnecessary character region by the overlapping region deleting unit and the dictionary search deleting unit, the character information recognizing unit 122 obtains the image information acquired from the input information acquiring unit 121 and the character information not deleted. The information is input to the information registration unit 123. The information registration unit 123 registers information in the same manner as S406 in FIG. 4 (S1507), and ends the process.

  The amount of character information registered in the search target DB 106 is reduced by deleting the character information of the character region that is clearly judged unnecessary by the overlapping region deletion unit and the dictionary search deletion unit. By reducing the amount of character information registered in the search target DB 106, the search target information is reduced. Therefore, as described above, the time required for the search can be reduced by deleting the character information of the character region that is clearly determined to be dependent.

  In the above embodiment, as described with reference to FIG. 16, the overlapping area deletion unit and the dictionary search deletion unit have been described as examples of the erroneous recognition area deletion unit. In addition, any means that deletes a character area that is clearly judged to be unnecessary can be used as a misrecognition area deleting means.

  As an example of the erroneous recognition area deletion means, deletion based on the relationship between the vertical or horizontal width of the character area and the line direction can be considered. For example, in the case of a horizontally written character area, it is considered that the width in the horizontal direction is at least several characters to several tens of characters. Therefore, when the character area is horizontally written and the width in the horizontal direction is less than several characters, for example, when the width is about 1 to 3 characters, the character area is deleted as an erroneously recognized area. Similarly, in the case of a vertically written character area, the width in the vertical direction is considered to be at least several characters to several tens of characters. Therefore, when the character area is vertically written and the width in the vertical direction is less than several characters as described above, the character area is deleted.

  In addition, deletion by the line direction and aspect ratio of the character area can be considered. For example, in the case of a horizontally written character area, it is considered that the width in the horizontal direction is wider than the vertical direction. Accordingly, a character area that is horizontally written and has a width that is wider in the vertical direction than in the horizontal direction is deleted as an erroneously recognized area. Similarly, in the case of a vertically written character area, the width in the vertical direction is considered to be wider than that in the horizontal direction. Therefore, a character region that is vertically written and has a width that is wider in the horizontal direction than in the vertical direction is deleted in the same manner as described above.

It is a block diagram which shows the hardware constitutions of the image processing apparatus which concerns on embodiment of this invention. It is a block diagram which shows the function structure of the image processing apparatus which concerns on embodiment of this invention. It is a block diagram which shows the function structure of the character recognition control part which concerns on embodiment of this invention. 3 is a flowchart showing an operation of the image processing apparatus according to the embodiment of the present invention. It is a figure which shows the example of the input information which concerns on embodiment of this invention. It is a figure which shows the extraction aspect of the circumscribed rectangle of the black pixel which concerns on embodiment of this invention. It is a figure which shows the row extraction aspect which concerns on embodiment of this invention. It is a figure which shows the example of the character information produced | generated by character recognition in embodiment of this invention. It is a figure which shows the row extraction aspect which concerns on embodiment of this invention. It is a figure which shows the example of the character information produced | generated by character recognition in embodiment of this invention. It is a figure which shows the example of the information which the search object DB which concerns on embodiment of this invention has memorize | stored. It is a figure which shows the example of the reading original document concerning embodiment of this invention. It is a figure which shows the example of the reading original document concerning embodiment of this invention. It is a figure which shows the variation information of the line extraction method which the character information recognition part which concerns on embodiment of this invention has memorize | stored. It is a figure which shows operation | movement of the image processing apparatus which concerns on other embodiment of this invention. It is a figure which shows the list of the misrecognition area deletion means which concerns on other embodiment of this invention. It is a figure which shows the deletion aspect of the misrecognition area | region which concerns on other embodiment of this invention. It is a figure which shows the function structure of the character recognition control part which concerns on other embodiment of this invention.

Explanation of symbols

1 Image processing apparatus 10 CPU
20 RAM
30 Engine 40 ROM
50 NVRAM
60 I / F
70 LCD
80 Operation unit 100 Controller 101 ADF
102 Scanner unit 103 Output tray 104 Network I / F
105 Display panel 106 Search target DB
107 Dictionary information DB
DESCRIPTION OF SYMBOLS 111 Main control part 112 Engine control part 113 Input / output control part 114 Image processing apparatus 120 Character recognition control part 121 Input information acquisition part 122 Character information recognition part 123 Information registration part

Claims (13)

An image processing device that generates character information by extracting the character region from an image including a character region and stores it as search target information,
An image acquisition unit for acquiring the image;
A first character region extraction unit that extracts a character region from the acquired image based on a first region extraction method for extracting the character region and generates first character information;
A second character region extraction unit that extracts a character region from the acquired image based on a second region extraction method different from the first region extraction method, and generates second character information;
An information storage unit that stores the first character information and the second character information in the recording medium as the search target information;
When the first character area extracted by the first character area extraction unit and the second character area extracted by the second character area extraction unit overlap, they are displayed in the overlapping area. A storage object exclusion unit that excludes the character information being stored from the storage object as the search object information,
The image processing apparatus, wherein the storage target exclusion unit excludes the first character information from the storage target when the entire first character area is included in the second character area. When the first character area extracted by the first character area extraction unit overlaps the second character area extracted by the second character area, the first character area is displayed in the overlapping area. A storage target exclusion unit that excludes the character information that is included from the storage target as the search target information,
The storage object exclusion unit includes the entire horizontal range of the first character area included in the entire horizontal range of the second character area, and the direction of the character line of the first character area and the When the direction of the character line of the second character area satisfies a predetermined condition, the character information displayed in the overlapping area of the first character information is excluded from the storage target. to claim 1 Symbol mounting image processing apparatus. When the direction of the character line of the first character area and the direction of the character line of the second character area are horizontal directions, the storage target exclusion unit duplicates the first character information. characterized in that it excludes the character information displayed in a region are from the storing object, the image processing apparatus according to claim 2 Symbol placement. When the first character area extracted by the first character area extraction unit overlaps the second character area extracted by the second character area, the first character area is displayed in the overlapping area. A storage target exclusion unit that excludes the character information that is included from the storage target as the search target information,
The storage object exclusion unit includes the entire vertical range of the first character area included in the entire vertical range of the second character area, and the direction of the character line of the first character area and the When the direction of the character line of the second character area satisfies a predetermined condition, the character information displayed in the overlapping area of the first character information is excluded from the storage target. The image processing apparatus according to any one of claims 1 to 3 . When the direction of the character line of the first character area and the direction of the character line of the second character area are vertical, the storage object exclusion unit includes the duplicate of the first character information. characterized in that it excludes the character information displayed in a region are from the storing object, claim 4 Symbol mounting image processing apparatus. A dictionary information acquisition unit for acquiring dictionary information;
A dictionary search unit for searching for the first character information or the second character information and extracting a word included in the acquired dictionary information;
And a storage target exclusion unit that excludes the first character information or the second character information from the storage target as the search target information according to a search result by the dictionary search unit. The image processing apparatus according to any one of 1 to 5 . The storage object exclusion unit excludes, from the storage object, all of the first character information and the second character information whose words extracted as a result of the search by the dictionary search unit are single-character words. characterized in that, the image processing apparatus according to claim 6 Symbol mounting. A storage target exclusion unit that excludes the character information displayed in the character region extracted by the first character region extraction unit or the second character region extraction unit from the storage target as the search target information;
The storage object exclusion unit excludes character information displayed in the character area from the storage object based on a relationship between a line direction of the character area and a width of the character area in a direction parallel to the line direction. wherein the image processing apparatus according to any one of claims 1 to 7. The storage object exclusion unit stores the character information displayed in the character area when the width of the character area and the width of the character area in the direction parallel to the line direction is equal to or less than a predetermined range. characterized in that it excluded from, 8 Symbol claim mounting image processing apparatus. A storage target exclusion unit that excludes the character information displayed in the character region extracted by the first character region extraction unit or the second character region extraction unit from the storage target as the search target information;
The storage object exclusion unit excludes character information displayed in the character area from the storage object based on a relationship between a line direction of the character area and an aspect ratio of the character area. The image processing apparatus according to any one of 1 to 9 . An image processing method for generating character information by extracting the character region from an image including a character region and storing it as search target information,
The image acquisition unit acquires the image,
A first character region extraction unit that extracts a character region from the acquired image based on a first region extraction method for extracting the character region to generate first character information;
A second character region extraction unit extracts a character region from the acquired image based on a second region extraction method different from the first region extraction method to generate second character information;
An information storage unit stores the first character information and the second character information in the recording medium as the search target information,
When the storage target exclusion unit overlaps the first character region extracted by the first character region extraction unit and the second character region extracted by the second character region extraction unit, the duplication If the character information displayed in the area is excluded from the storage target as the search target information, and the entire first character area is included in the second character area, the first character information Is excluded from the storage target. A control program that causes an information processing device to operate as an image processing device that extracts character regions from an image including character regions, generates character information, and stores the information as search target information.
Obtaining the image;
Extracting a character region from the acquired image based on a first region extraction method for extracting the character region to generate first character information;
Extracting a character region from the acquired image based on a second region extraction method different from the first region extraction method to generate second character information;
Storing the first character information and the second character information in the recording medium as the search target information;
The first character area extracted in the step of generating the first character information and the second character area extracted in the step of generating the second character information are overlapped. Removing the character information displayed in the area from the storage target as the search target information;
A control program for causing the information processing apparatus to execute a step of excluding the first character information from the storage target when the entire first character region is included in the second character region. . Recording medium, wherein the information processing apparatus control program according to claim 1 2 was recorded in a readable format.

Images