JP2008193580A - Information processing apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To prevent the deterioration of an image read by an optical reading process by using information recorded in an original even when the capacitance of information recordable in the original is limited. <P>SOLUTION: The information processing apparatus includes: an image reading unit 101 for acquiring a document image from an original; a character layout information reading unit 104 for acquiring character layout information; a character recognizing unit 103 which performs character recognition processing on the acquired document image to acquire text data; and a character data constitution unit 109 which reconstitutes the acquired text data on the basis of the acquired character layout information to acquire restored electronic data. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to an information processing apparatus and a paper-like recording medium, and more particularly to a technique for preventing deterioration of a character image due to optical reading.

  When a digital image (read image) of a document is acquired by optically reading the document, image quality may be deteriorated due to blur or noise on the read image. Also, the quality of the read image may be lowered due to the document being inclined during reading.

  Several methods have been proposed for preventing such deterioration of the read image quality. For example, a copying apparatus is connected to a server that stores electronic data corresponding to a document to be read, and a code (bar code or two-dimensional bar code) printed on the document to be read is added to or embedded in the document. By acquiring an identifier (ID) for specifying a document read from the RFID chip, and reading out electronic data corresponding to the document from the server, it is possible to prevent deterioration of the read image. ing. However, in a copying apparatus that is separated from a server that stores electronic data corresponding to a document, there is a problem in that the server cannot be accessed and deterioration of a read image cannot be prevented.

  Further, as shown in FIG. 26, electronic data corresponding to a document is recorded on an RFID chip added or embedded in the document, and when copying, printing is performed based on a read image optically read by a scanner. Instead, there is an idea of preventing deterioration of the read image by reading data recorded on the RFID chip and executing printing based on the read data. However, data that can be recorded on an RFID chip of a size that is not inconvenient to add or embed in a document is currently several hundred bytes, which is insufficient to record electronic data corresponding to information for one page of a document. is there.

  Further, as a method of recording electronic data corresponding to an original so as to be machine-readable, the original is encoded and replaced with a pattern such as a barcode or a two-dimensional code as shown in FIGS. A method of acquiring electronic data by decoding a pattern read by a reading means (such as a scanner) has also been proposed. However, in a capacity that can be replaced with a barcode or a two-dimensional code, an electronic device corresponding to a general document is used. The current situation is that it is insufficient to record data.

  In addition, as shown in FIG. 29, there is a proposal of encoding electronic data corresponding to a document and embedding it as a background dot pattern for printing, but in order to embed a background dot pattern so as not to interfere with the interpretation of characters in the document. However, since the area where the pattern can be arranged is limited, a large storage capacity cannot be obtained.

  The recording method as described above is limited to the embedding of information that requires only a small storage capacity such as reading a slip by a machine or recording a FAX number in inventory management or distribution management. It is.

Here, when the user copies a printed material in which the original document portion on which the original document has been printed and the code information portion obtained by encoding the original document are printed on the margin of the original document portion in the printed material, The machine obtains code information from the printed material, decodes the code information to obtain digital information, generates image data of the original document from this digital information, and performs copying, thereby repeatedly copying the image. There is a technique that can provide an image forming system that can avoid deterioration and maintain the same image quality as the original document or electronic document information indefinitely (see, for example, Patent Document 1).
JP 2003-244424 A

  However, the technique disclosed in Patent Document 1 requires a recording unit having a very large capacity because all of the electronic data corresponding to the document is recorded in the recording unit provided in the document. However, the above-described problems are not solved.

  Therefore, an object of the present invention is to prevent deterioration of a read image due to optical reading by using information recorded on a document even when the capacity of information recordable on the document is limited.

  According to the first aspect of the present invention, a reading unit that acquires a document image and character layout information from a document, and a character recognition processing unit that executes text recognition processing on the document image acquired by the reading unit to acquire text data. An information processing apparatus comprising: reconstruction means for reconstructing text data acquired by the character recognition processing means based on character layout information to obtain restored electronic data.

  The invention according to claim 2 executes a character recognition process based on the character layout information acquired by the reading means for the document image acquired by the reading means and the document image acquired by the reading means from the document. An information processing apparatus comprising character recognition processing means for acquiring text data.

  According to a third aspect of the present invention, in the information processing apparatus according to the second aspect, the character recognizing unit acquires the template generating unit that generates a character template based on the character layout information acquired by the reading unit, and the reading unit acquires the character template. A character recognition processing unit that performs character recognition processing by collating a document image with a template generated by a template generation unit for each character is characterized.

  According to a fourth aspect of the present invention, in the information processing apparatus according to the third aspect, the character recognition unit further includes a character separation unit that cuts out characters from the document image based on the character layout information acquired by the reading unit. Is characterized in that a character recognition process is performed by collating the character cut out from the document image by the character separation unit with the template generated by the template generation unit.

  According to a fifth aspect of the present invention, in the information processing apparatus according to the third or fourth aspect, in addition to the document image and the character layout information from the original, the reading unit is a predetermined character included in the document image and the original of the character. Character recognition auxiliary information associated with position information indicating the position is acquired, and the character recognition means refers to the character recognition auxiliary information based on the position of the character to be processed in the character recognition process in the character recognition unit. When the information corresponding to the position is included in the character recognition auxiliary information, the character corresponding to the position included in the character recognition auxiliary information is output instead of the character recognition processing result.

  According to a sixth aspect of the present invention, in the information processing apparatus according to any one of the first to fifth aspects, the character layout information includes: character font information indicating a character font; character size information indicating a character size; It is characterized by including.

  According to a seventh aspect of the present invention, in the information processing apparatus according to the sixth aspect, the character layout information further includes character modification information indicating a modification applied to the character.

  The invention according to claim 8 is the information processing apparatus according to claim 6 or 7, wherein the character layout information further includes range information indicating a range of the character string.

  The invention according to claim 9 is the information processing apparatus according to claim 5, wherein the predetermined character included in the character recognition auxiliary information is a character that may be erroneously recognized by the character recognition processing unit. To do.

  According to a tenth aspect of the present invention, in the information processing apparatus according to the fifth or ninth aspect, characters that may be erroneously recognized by the character recognition processing unit are classified according to the font of the character or the presence or absence of decoration for the character. It is characterized by being set.

  According to an eleventh aspect of the present invention, in the information processing apparatus according to any one of the first to tenth aspects, the reading unit optically scans a document to obtain a document image and character layout information. Features.

  According to a twelfth aspect of the present invention, in the information processing apparatus according to any one of the first to tenth aspects, the reading unit includes a first reading unit that optically scans a document to obtain a document image; And a second reading unit that acquires character layout information from a recording unit attached to the document.

  According to a thirteenth aspect of the present invention, there is provided an information processing apparatus for printing on a paper-like recording medium having a document image and character layout information and outputting the document image and the character layout information by reading means of another information processing apparatus. Text data is acquired by executing character recognition processing on the document image acquired by the reading means by the character recognition processing means of the other information processing apparatus, and the character recognition processing is performed by the reconstruction means of the other information processing apparatus An information processing apparatus for printing and outputting on a paper-like recording medium characterized in that restored electronic data is obtained by reconstructing text data acquired by the means based on character layout information.

  According to a fourteenth aspect of the present invention, there is provided an information processing apparatus for printing on a paper-like recording medium having a document image and character layout information, and outputting the document image and character layout information information by reading means of another information processing apparatus. The character recognition processing is executed based on the character layout information acquired by the reading unit on the document image acquired by the reading unit by the character recognition processing unit of the other information processing apparatus, and the text data is acquired. This is an information processing apparatus that prints and outputs a characteristic paper-like recording medium.

  According to the present invention, even when the amount of information that can be recorded on a document is limited, it is possible to prevent deterioration of a read image due to optical reading by using the information recorded on the document.

  Next, the configuration of the first exemplary embodiment of the present invention will be described with reference to the drawings.

  Referring to FIG. 1 showing a configuration example of the information processing apparatus according to the first embodiment of the present invention, the information processing apparatus includes an image reading unit 101, an image processing unit 102, a character recognition unit 103, and a character layout information reading unit 104. , Character layout information decoding unit 105, typeface / arrangement information extraction unit 106, typeface / arrangement information conversion unit 107, typeface / arrangement information recording unit 108, character data configuration unit 109, image data configuration unit 110, data recording unit 111, recording It comprises a medium 112, a data transfer unit 113, and a data receiving unit 114, and document data can be automatically constructed from information of a read image and machine reading data recording means. A printing unit 115 for printing such a document is also provided.

  The original means a paper-like recording medium on which a document image and character layout information are recorded (the same applies to the second and third embodiments of the present invention described later). The character layout information may be printed and recorded on a paper recording medium, or may be electronically recorded on an RFID or the like added to or embedded in the paper recording medium. The document image means printed document data. The document data is data in the electronic world, and includes at least document data such as graphics, images, characters, and tables, and character layout data in the document data.

  The character layout information means information in a state where character layout data is recorded on a document. In the present embodiment, the character layout data includes character font information (information indicating character font), character size information (information indicating character size), character of character string data (text data) in the document data. Typeface information (information indicating the typeface of the character), placement information (information indicating the position where the character string is placed), range information (information indicating the range of the character string; character string data in the range included in this information (If the font font information, the character size information, the character font information, and the arrangement information are set in common, the capacity required for recording can be reduced). The same applies to the embodiment).

  In the present embodiment, text data means data obtained by performing character recognition processing on a read image. In the present embodiment, the restored electronic data means information obtained by reconstructing text data based on character layout information.

  The read image means an image obtained by optically reading a document.

  The machine-readable data recording means attached to the paper-like recording medium records information related to character layout such as character font, line spacing and alignment, such as character font and size. For example, as shown in FIG. 4B, which shows an example of information to be recorded in the machine reading data recording means of the present embodiment, information such as the number of lines, the number of digits, the font at that position, the size, etc. Record.

  As the machine reading data recording means, an RFID chip as shown in FIG. 26 showing an example of the machine reading data recording means, a barcode pattern as shown in FIG. 27 showing an example of the machine reading data recording means, and machine reading data. A two-dimensional code pattern as shown in FIG. 28 showing an example of recording means, a background dot pattern as shown in FIG. 29 showing an example of machine reading data recording means, and the like can be considered. The present invention is not limited to these as long as the machine can easily read and write information. When recording information on RFID, an antenna for reading and writing information is required. When printing a print code pattern in which character layout information is embedded, the information is read simultaneously with the reading of the original by the scanner, and the information is read simultaneously with the printing of the document by the printer.

  A processing operation when the document data of this embodiment is printed will be described with reference to FIG. When the information processing apparatus receives a print command (S201), it extracts character layout information relating to the font, character size, and arrangement in the document from the document data (S202). When the character layout information is embedded in the print code pattern, the character layout information is encoded to generate a print code pattern (S203). When the document is printed, the print code pattern is printed together with the document data. When the machine-readable data recording means is an RFID chip, as shown in FIG. 3, the extracted character layout information is converted into a pre-defined recording format, and preparations for writing into the RFID are made. Next, the restored electronic data as image data to be printed is converted and generated from the document data (S204), and printing is started (S205).

  FIG. 3 showing another processing operation when document data is printed is basically the same as FIG. 2 (S301 to S304), but writing to the RFID chip is performed immediately before printing (S301 to S304). S305) Although the order is the same, depending on the apparatus configuration, writing to the RFID chip may be performed immediately after printing, or recording to the chip may be performed in parallel with printing. good.

  The font information is not Ming Dynasty or Gothic. Corresponds as I say half-width M for Mincho, half-width G for Gothic, half-width I for italic, half-width B for bold. If the symbols are determined, the number of characters can be reduced, so that the amount of information related to the character layout can be reduced. It is possible to reduce the amount to be recorded by recording the size of the most fonts and characters once and recording the information of the font or size different from that.

  The printed matter (FIG. 4A) is read by the image reading unit 101 and the character layout information reading unit 104, and character recognition processing (= OCR processing) is performed as shown in FIG. The restored electronic data is constructed by combining the text data extracted by the above and the character layout information such as the font, size, and color of the characters recorded in the machine-readable data recording means (FIG. 4C). The restored electronic data is printed (copied), saved as a word processor data including information such as text data and character fonts, or transferred to another device (see FIG. 4). (D)).

  The flow is shown in FIG. First, when the user selects a mode such as “copy”, “send”, or “save” (S501), regardless of which mode is selected, the restored electronic data starts to be configured (S503).

  In the restoration electronic data construction step, restoration electronic data is constructed from the scanned scanned image (S5031). Character layout information such as the font, character size, and character arrangement recorded in the machine reading recording unit is decoded (S5032), and the optically read image is subjected to character recognition processing (S5033) to obtain text data. Is generated. The restored electronic data is generated by combining the text data generated by the character recognition and the character layout information from the machine reading data recording unit (S5034).

  Next, the selection of the mode is judged (S502). If the "copy" mode is selected, the number of sheets, the scale, etc. are further input (S504, S505), and the combined restoration is performed according to the input conditions. Electronic data is printed (S506). When the “transmission” mode is selected, the transmission destination is further input (S507, S508), and the restored electronic data is transmitted to the designated transmission destination (S509). If the “save” mode is selected, a file name, a drive, a medium, a folder, etc. for saving data are input (S510, S511), and the synthesized restored electronic data is saved (S512).

Also, when copying in a system accessible to the read image as the original data, the original data is accessed from the ID recorded in the RFID chip or embedded in the printed print code pattern. Then, the original data may be printed. By doing so, the character recognition process becomes unnecessary, so that a copying operation can be performed at a higher speed and more reliably without image deterioration.
In the “copy” mode, the read font and arrangement information may be written again in the RFID chip or the print code pattern. When the machine-readable data recording means is a print code pattern, it can be copied without degrading the print code pattern if the once decoded data is encoded and printed.

  According to the present embodiment, not all document data is recorded in the recording unit provided in the document, so that necessary information can be left in a limited capacity. By combining the character layout information stored in the storage means provided on the paper-like recording medium and the text data obtained by the character recognition process, it is possible to prevent image deterioration of characters due to copying.

Next, a second embodiment of the present invention will be described.
In normal character recognition, after a document is optically read, character recognition processing is performed to obtain character string information. In the present embodiment, the text data means data obtained by performing character recognition processing on the read image based on the character layout information (the same applies to the third embodiment of the present invention described later). is there). In the present embodiment, the restored electronic data means text data. (The same applies to the third to seventh embodiments of the present invention described later).

  A general procedure of conventional character recognition processing is as shown in FIG. First, characters are input as image data by photoelectric conversion (S601). Thereafter, processing such as quantization is performed to convert analog data into digital data. The preprocessing is a process for facilitating character recognition, in which noise that enters when a character is read and image binarization are performed (S602). In the character separation step, segmentation is performed to divide the characters into one by one, and normalization is performed to align the character size to a predetermined size (S603).

  The character data subjected to the character separation process is sent to the next feature extraction unit. The feature extraction unit extracts the feature of the character shape (S604). Since the number of dimensions is high as data just by normalization, the data is compressed into data having a lower number of dimensions than the character represents. As a feature extraction method, as shown in FIG. 7 showing an example of feature extraction, a method based on a core line in which characters are thinned (close to one dot width), and based on an outline as shown in FIG. 8 showing an example of feature extraction. There are methods (by straight lines, curves, relative positions, etc.).

  Next, the character data from which the features have been extracted is sent to the identification unit and collated with a prepared character graphic pattern (a collection of character graphic patterns is a standard character dictionary in the figure) (S605). . In type recognition, there is also a method of identifying an input pattern by comparing it directly with a reference character figure without performing feature extraction. A post-processing step (S606) is provided, and it is attempted to find a misrecognized part by confirming whether a character string that has been recognized by a morphological analysis based on a morphological analysis dictionary is generally formed as a sentence. There is also a method.

  Character recognition, which has conventionally been performed only with image pattern recognition processing, is used to improve the accuracy and speed of character recognition by using print code patterns and character layout information recorded in an RFID chip.

  As shown in FIG. 9, which is a diagram showing the flow of processing of the restored electronic data configuration part, the restored electronic data is configured. The difference from FIG. 5 is the portion related to the structure of the restored electronic data (S903), and the operations before and after that are the same.

  In the step of constructing the restored electronic data (S903), when the original is read (S9031), the character layout information is read from the machine reading data recording means and decoded (S9032). First, character size and font information are used for character segmentation. If the font used is a monospaced font, character separation can be performed by separating the information at a predetermined interval using information on the font and character size without performing complicated pattern recognition processing. Yes (S9033). For this reason, there is no need to misread “Ten” or “Two” again or again, and it takes time to separate characters.

  If half-width characters are mixed in a full-width character, a process that can detect a shift in the cutout of the character after that half-width character (if the match rate with the reference character continuously decreases in character determination, If character recognition is performed with the character at the position where the matching rate starts to decrease as half-width characters), even if half-width characters are mixed, character separation can be performed accurately.

  Proportional fonts cannot be cut out at a fixed pitch, but the character spacing is the same (in the case of English fonts, it may be the spacing of specific parts), so if the character size is determined , Character spacing is also determined. For this reason, it is only necessary to cut out characters so as to have a fixed character interval, so that a simple pattern recognition process can be performed compared to the conventional method. From the above points, it can be seen that the accuracy and speed are improved as compared with the case where the character is cut out only by the pattern recognition process without knowing the font or character size in advance.

  After the character segmentation is finished, in the conventional method, as shown in FIG. 6, character feature extraction is performed, candidate characters are narrowed down, and finally character discrimination is performed by comparison with a standard character figure. . In the character discrimination step (S9037), the print code pattern and the character layout information recorded on the RFID chip are also used. Even if no character is recorded, the typeface and the size of the character can be known, so the standard character (reference template) is limited to a specific typeface and size, and character matching is performed for identification. For example, suppose that it is understood from the printed print code pattern or the typeface information recorded on the attached RFID chip that the character to be character-recognized is printed with 11 points of Gothic type.

  Taking the case of reading the character “eiga” and determining the character as an example, as shown in FIG. 10, the read character image and the Gothic font, a candidate character with 11 points (for example, “eiga”) A template of “water”, “ice”, etc.) is generated and contrasted (S9036). If the printed character is Mincho, it can be understood from the font information recorded in the machine-readable data recording means, so a Mincho template can be generated and compared. Select the character with the highest match rate as the target result. Since the font and character size of the template shape can be specified, when the character recognition can be performed correctly, the shape matching rate becomes high. Since a template image having the same size as the read character may be generated and compared, the normalization process that has been necessary in the past is unnecessary. As a result, character recognition accuracy and processing speed can be improved.

  It is known that the accuracy of character recognition can be improved by fixing the character size and typeface to one type. However, since an appropriate template is prepared according to the printed character, Even if multiple types of fonts are used, the same character recognition accuracy as when a predetermined font is used can be obtained (unless it is a special font that is not prepared on the information processing apparatus side).

  Further, since the font and position of the character are known when the character is identified one character at a time, unlike the procedure described in the first embodiment, the character recognition result and the character layout information are newly changed after the character recognition is completed. It is also possible to adopt a processing procedure that does not synthesize.

  Subsequent processing is the same as steps S5033 and S5034 shown in FIG. 5 of the first embodiment.

  According to the present embodiment, the accuracy of character recognition can be improved because the shape of the template is specified when the font and size are limited. In many types of character recognition processing, the result of character recognition is determined by including a step of comparing a read image with a character template that serves as a character recognition reference. If the font of the character that is printed and the character that is easy as a character template are different, the matching rate is low even if the characters are the same, and it is effective because the accuracy of character recognition cannot be increased. is there.

Next, a third embodiment of the present invention will be described.
In the present embodiment, the character layout information means information in a state where character layout data is recorded on a document. In the present embodiment, the character layout data includes character font information (information indicating character font), character size information (information indicating character size), character of character string data (text data) in the document data. Typeface information (information indicating the typeface of the character), placement information (information indicating the position where the character string is placed), range information (information indicating the range of the character string; character string data in the range included in this information If character font information, character size information, character typeface information, and arrangement information are set in common, the recording capacity can be reduced.) Furthermore, character decoration information (decoration applied to characters) (For example, underline or strikethrough).
Similar to recording the size of the font or character in the machine reading data recording means, the portion to be underlined or strike-through is recorded in the machine reading data recording means and printed.

  In order to perform character recognition at the location where there is an underline or strikethrough, a character template with a strikethrough or shading added to the original character is prepared and compared with the scanned image.

  For example, as shown in FIG. 11, even if an image with a strikethrough is compared with a template of characters without a strikethrough, the shape changes and the matching rate decreases. Therefore, when identifying a template, a reference template is created by adding a strikethrough to the original character shape as shown in FIG. 12, and is compared with the read image. By comparing with a strike-through template, it is not necessary to reduce the accuracy of character recognition even for a strike-through character.

  According to the present embodiment, even if there is an underline or strikethrough, a character template with an underline or a character template overlapping with a strikethrough can be prepared, so that the accuracy of character recognition can be improved. In a sentence, some lines may be underlined, or one or two lines may be superimposed on a character as a strikethrough to leave a history of document creation. This is effective because character recognition may not be performed correctly. Strikethrough parts are difficult to separate from characters and strikethrough, and even when compared with normal (not overlapping with strikethrough) character templates, characters that are underlined or strikethrough Since the rate is low, the accuracy of character recognition is reduced, which is effective.

Next, a fourth embodiment of the present invention will be described.
Most of the manuscripts are recognized correctly, but there are characters with a large number of strokes and characters with similar shapes (number 0 and alphabet O, hiragana and katakana). There are characters that are likely to be unrecognizable or misrecognized due to, e.g.

  The original means a paper-like recording medium on which character recognition auxiliary information is recorded in addition to a document image and character layout information (the same applies to fifth to seventh embodiments of the present invention described later). is there). The character recognition unit records in the recording means a list (misrecognized character list) in which characters that may be erroneously recognized during character recognition processing are listed in advance. Checks whether there is any character in the misrecognized character list in the character string data included in the document data. Record. That is, the character recognition auxiliary information is information recorded on a document, and is information indicating a character code and a position of a character that may be erroneously recognized in the character recognition process. Similar to the character layout information, the character recognition auxiliary information may be printed and recorded on a paper-like recording medium, or may be electronically recorded on an RFID or the like added to or embedded in the paper-like recording medium.

  In the present embodiment, the text data means data obtained by subjecting a read image to character recognition processing based on character layout information and character recognition auxiliary information (fourth to fourth of the present invention described later). The same applies to the seventh embodiment).

  Referring to FIG. 13 showing the operation of the present embodiment, when a print command is received (S1301), there is no character that is difficult to recognize in the printed text before and after extracting the character layout information (S1302). (S1303). If there is a character in the document that is difficult to recognize, the character position and character code are used as auxiliary information for character recognition, and the character code information such as a print code pattern or RFID chip is recorded. Record in recording means.

  Referring to FIG. 14 showing the process of reading the printed document and reproducing the document data, the character layout information and the character recognition auxiliary information are read (S14032) and recorded in the print code pattern or RFID chip as character recognition is difficult. With respect to the detected character, text data is created using the recorded character and the position information included in the character recognition auxiliary information without using the character recognition process (S14036). Assuming that character recognition is not difficult, character recognition processing is performed for characters not recorded in the print code pattern or RFID chip (S14035), and text data is generated. These processes are performed for all characters (S14037). Characters that are difficult to recognize are recorded on a print code pattern or an RFID chip, so that the recorded characters are not erroneously recognized.

  Returning to FIG. 13, when character recognition (including data insertion of the machine reading data recording means) is completed, the restored electronic data is converted and generated (S1305), written to the RFID chip (S1306), and printed. Start (S1307).

  According to the present embodiment, it is possible to reduce the possibility that a read character is erroneously recognized and erroneously printed due to a large number of strokes or a similar character. If there are characters with a large number of strokes or characters that are similar in shape, the machine may be difficult to recognize characters or misrecognize, it takes time to recognize characters, or character recognition fails, This is effective because the document data is composed or printed with wrong characters.

Next, a fifth embodiment of the present invention will be described.
If the font type and size of characters are frequently changed, the character layout information increases, and only information other than the character layout can be recorded. Therefore, as shown in FIG. 15, a list of character recognition difficulty levels is prepared, and according to the print code pattern in which information is embedded and the free space in the storage part of the RFID chip, the character The position information is recorded. For example, Hiragana Hebepe and Katakana Hebepe, Hiragana Nori and Katakana Li, Katakana Ta and Kanji Evening are highly likely to be misrecognized. However, among them, Hiragana Hebepe and Katakana Hebepe are the most difficult, so if there are "he" and "ta" in the sentence, "he" is recorded with priority. When the capacity of the machine-readable data recording means is insufficient, the low-priority characters (characters with a low possibility of character recognition being listed) are not recorded in the machine-readable data recording means. To.

  Since the difficulty level of character recognition for each character varies depending on the character recognition method and the like, it is necessary to rank them in advance.

  According to the present embodiment, it is possible to prevent failure of character recognition with a limited recording capacity by recording only characters that are likely to be erroneously recognized. Since the print code pattern and the capacity of the RFID chip are limited, it is effective because it may not be possible to record all information related to characters that are difficult to recognize in addition to the character layout information.

Next, a sixth embodiment of the present invention will be described.
The character recognition auxiliary information in the present embodiment is not different from the character recognition auxiliary information in the fourth embodiment described above, but the misrecognized character list recorded by the character recognition unit is different. The character recognition unit records, in the recording unit, a list (character recognition character list) in which characters that may be erroneously recognized during character recognition processing are listed in advance for each font. The character recognition unit acquires the character of the character string data included in the document data and the font of the character, checks whether the character is included in the misrecognized character list prepared for each font, and enters the misrecognized character list. For the included characters, the character code and position are recorded on the document as character recognition auxiliary information. The character recognition auxiliary information is information recorded on a document and is information indicating a character code and a position of a character that may be erroneously recognized in the character recognition process.

  Since the order of difficulty varies depending on the type of font, a different difficulty level table (misrecognized character list) is prepared for each font as shown in FIG. Depending on the font used, characters that are likely to fail character recognition are preferentially recorded in the machine reading recording unit. Considering the case where Gothic and Mincho are compared, for example, hehepe (Hiragana) and Hebepe (Katakana), ta (Katakana) and evening (Kanji) are more distinctive than the Gothic Is difficult. On the other hand, it is easier to distinguish the Mincho type than the Gothic type for-(long sign) and one (Kanji), to (Katakana) and 卜 (Kanji). Therefore, if the characters on the same page are printed with a mixture of Mincho and Gothic, and if there is one Mincho (Kanji) and one Gothic (Kanji) in the sentence, Mincho A Gothic character (kanji) is given priority over a physical character (kanji), and is recorded in the recording unit for machine reading.

  When the capacity of the recording unit is insufficient, characters of low priority are not recorded as in the fifth embodiment.

  According to the present embodiment, the easily distinguishable character and the character that is not so differ depending on the font. Therefore, the failure of character recognition is reduced by changing the priority order to be recorded in the machine reading recording unit for each font. be able to. Even if the characters are the same, the difficulty of character recognition differs depending on the type of font. Therefore, when the difficulty level is uniformly set, the difficulty level setting may not be suitable depending on the type of font, which is effective.

Next, a seventh embodiment of the present invention will be described.
The character recognition auxiliary information in this embodiment is not different from the character recognition auxiliary information in the fourth and sixth embodiments described above, but is different in the misrecognized character list recorded by the character recognition unit. The character recognition unit records, in the recording unit, a list (misrecognized character list) in which characters that may be erroneously recognized during character recognition processing are classified according to cases where there is no character decoration information. The character recognition unit acquires the character of the character string data included in the document data and the decoration information of the character, and determines whether the character is included in the misrecognized character list prepared separately for cases where there is no decoration. When the character is included in the misrecognized character list, the character code and position are recorded on the document as character recognition auxiliary information.

  Even if it is known that there is a strikethrough, a list of characters that are easily affected by machine character recognition due to the strikethrough is created by changing the difficulty level of the character recognition depending on the presence or absence of the strikethrough. The priority order recorded in the recording unit such as the RFID chip is made different depending on the presence or absence of the strikethrough.

  For example, a misrecognized character list is created as shown in FIG. The characters gate and 閂 are registered with low character recognition difficulty because there is no strikethrough or when there is a strikethrough but vertical writing, so the character recognition difficulty is registered low, but when strikethrough is drawn in horizontal writing Since it becomes difficult to identify, the character recognition difficulty is registered high. Characters that are likely to be mistaken, such as cows and noon, are set to be more difficult in the case of vertical strikeout ants.

  According to the available capacity of the machine reading data recording means, the data is recorded in the machine reading data recording means in descending order of character recognition difficulty.

  Also, depending on the type of font and the size of the character, the influence of the strikethrough is small, so even if there is a strikethrough, the character recognition difficulty setting may be changed depending on the font and size. For example, as shown in FIG. 18, when the font is a regular font, the horizontal line in the foreground does not overlap with the strikethrough in the case of a single strikethrough. The difficulty does not change. Therefore, depending on the capacity of the recording unit, it is possible to adjust so that the recording is not performed because the priority is lowered. Therefore, as shown in FIG. 19, the setting of the difficulty level depending on the presence or absence of a strikethrough may be changed for each font.

  According to the present embodiment, in the case of characters having different legibility due to the presence or absence of strikethrough or underline, the priority order of characters to be recorded in the machine reading data recording means limited by the presence or absence of strikethrough or underline Because it changes, the limited storage capacity can be used effectively. When a part of the horizontal line constituting the character is a character close to the position of the strike-through line, the strike-through line and the line constituting the character overlap or approach each other, which makes it difficult to recognize the character. For example, as shown in FIG. 30, the gate and the fence overlap the strikethrough and the one in front of the fence. For this reason, even if it is known that there is a strikethrough, it is difficult to determine whether the character is a gate or a trap. Similarly, it is difficult to determine whether a character is 弋 or 弌 when 取 り 消 し and 弌 overlap each other in the strikethrough and the previous one. In addition, in the case of vertical writing, the strikethrough line passes through the center of the character, so it is known that characters such as 1 and 10 or large and wood have strikethrough lines as shown in FIG. Is effective because it is difficult to determine.

Next, an eighth embodiment of the present invention will be described.
As shown in FIG. 20, a document having a center-aligned line or a left-justified line does not record all the arrangement information on the print code pattern or the RFID chip, but at the time of character recognition. As shown in FIG. 21, the arrangement information is detected by replacing the blank part with the character space from the character arrangement. Unless the layout is special, if the character extraction accuracy of the character recognition process can be obtained, the character layout information of the document can be obtained from the character arrangement.

  Even if there is a center-aligned line, it does not detect that the line is center-aligned, and recognizes that there is a space of multiple characters from the beginning of the line. Strictly speaking, the original document data is not reproduced, but in many cases there is no problem. The amount of information related to characters that are difficult to recognize can be recorded by reducing the amount of information related to character arrangement.

  According to the present embodiment, since character arrangement information can also be detected during character recognition, it is not necessary to record character arrangement information in advance. Therefore, other information can be recorded in the recording unit. If the layout of each line changes frequently, such as when the layout of each line in the document is center-aligned, left-justified, or right-justified, the information related to the text layout becomes large, as described below. This is effective because a large amount of information for improving the accuracy of character recognition cannot be recorded.

Next, a ninth embodiment of the present invention will be described.
As in normal copying, the document data is configured so that the copied image is printed as it is without recognizing the read image, and the read image is recognized so that image deterioration of the character portion does not occur. Then, prepare a printing mode so that the user can select it.

  If you are reading from a paper document and creating word processing document data, or if there is a high possibility that the copied document will be further copied, or if you do not like image deterioration due to copying, you can perform the character recognition and document data composition process. When copying in the included mode and wanting to make a quick copy, the optically read image is printed as an image without character conversion, as in a conventional copying machine. When not only the copy mode but also the send mode or save mode is selected, the scanned image can be sent and saved as image data as it is, or the document data composed by character recognition and document data composition processing can be sent Or save it.

  According to the present embodiment, it is possible to select a mode that does not reduce the copying speed by selecting not to include an electronic re-editing step. If new document data is constructed by character recognition processing, it is possible to prevent deterioration in character image quality, but it takes time, so it is inconvenient for users who want to prioritize copying speed over image quality. It is.

Next, a tenth embodiment of the present invention will be described.
In order to clarify that there is a possibility that the data that made up the document data via paper and the scanned image that is the original data may be different due to a failure in character recognition, etc. Is recorded on the display unit or the machine reading recording unit, and when it is stored as electronic data, it is recorded as tag information of the restored electronic data.

  As shown in Fig. 22, this document is printed as headnotes and footnotes such as "This document is printed after character recognition processing" and "Character recognition may fail." However, it indicates that it is not normal copying or printing from the read image that is the original data (when recording on a display unit for visual recognition).

  Alternatively, when printing the restored electronic data, information indicating the unique ID attached to the paper and the copy made by reading the paper with the ID recorded on the RFID chip or the print code pattern, etc. The information is recorded on an RFID chip attached to the copied paper, or the information is embedded in a print code pattern.

  As shown in FIG. 23, the information is printed in a print code pattern. In FIG. 23, when a read image that is original data is printed, for example, “717346” is added as an ID, and the document with the ID is read in an environment where the read image that is the original data cannot be accessed. In the case of copying, data indicating that ID “717346” has been copied into the print code pattern is embedded in the print code pattern. For example, when the above-described processing is performed on a “717346” document, an ID such as “717346C” is attached. Further, when “717346C” is copied and printed by the above-described processing, it is conceivable to make it possible to know the copy of the generation, such as “717346CC”.

  When the restored electronic data is generated from the read image and stored, the ID of the read image that is the original data and information indicating that it has been copied are recorded as tag information of the constructed restored electronic data. In this way, when the document is read in an environment where the scanned image that is the original data can be accessed, or when the document data is copied to a computer connected to the network, the relationship with the scanned image that is the original data Can be confirmed.

  According to the present embodiment, it can be seen that the copied document and the constructed restored electronic data are not original. Further, when the copied document is read or the constructed restored electronic data is used in an environment where the original data can be accessed, it becomes a clue to know the relationship with the read image that is the original data. It must be possible to recognize that documents and restored electronic data copied via printed paper do not match the original manuscript, and the system can access the original data. This is effective because it is desirable that the original is easily associated with the original when the copy of the manuscript and the constructed restored electronic data are handled.

Next, an eleventh embodiment of the present invention will be described.
In addition to the printed sentence, there is a mode in which only the content drawn in the margin is extracted and printed, stored as image data, or transferred.

  At the time of printing, if the arrangement and size of characters are recorded on a print code pattern or RFID chip to be printed, the position where the characters are present in printing can be known. If the range of characters is separated from the read image as shown in FIG. 24 (FIG. 24C), the added part can be extracted (FIG. 24D). If a position such as a figure is also recorded, it is possible to detect a blank area other than the area of the character or the figure at the time of printing.

  According to the present embodiment, it is possible to extract a portion that is considered to be drawn after printing from the printed print code pattern or the character arrangement information stored in the RFID chip. If you only need to write in the margin, or if you want to create a document that hides the original print part, if the method is to take a difference from the read image that is the original data, the read image that is the original data Without it, the part that was originally printed and the part that was added to the margin could not be separated by machine processing, and there is also a method of making a mask by cutting paper etc., but it is troublesome and effective It is.

Next, a twelfth embodiment of the present invention will be described.
As shown in FIG. 25, there is a mode of copying in the form printed first for the case where a drawn memo or the like is to be hidden or a beautiful document without writing is desired.

  A mode is prepared in which the added portion can be removed and copied, and when that mode is selected, the document is printed in a state immediately after printing. Even in a mode other than the copy mode, it can be transmitted and stored as document data at the time of printing without the writing unit after printing.

  In combination with the ninth embodiment, an image obtained by excluding writing after printing from a read image without performing character recognition processing using only the area information on which the original printing before writing was performed. It is also possible to select a mode for copying, transmitting, and storing the image.

  According to the present embodiment, when the same form as that at the time of printing is desired from the character arrangement information, excluding the part that is drawn after printing or dirty, it is possible to print in such a form. . This is effective in the case where an original document without writing or dirt (less) is desired even in an environment where the read image as the original data cannot be accessed.

  Each of the above-described embodiments is a preferred embodiment of the present invention, and various modifications can be made without departing from the scope of the present invention. The present invention can also be applied to scanners, printers, copiers, multifunction devices, and the like.

It is a figure which shows the structural example of the information processing apparatus of the 1st Embodiment of this invention. It is a figure which shows the operation | movement of the 1st Embodiment of this invention. It is a figure which shows the operation | movement of the 1st Embodiment of this invention. It is a figure which shows an example of the information recorded on the data recording means for machine reading of the 1st Embodiment of this invention. It is a figure which shows the operation | movement of the 1st Embodiment of this invention. It is a figure which shows the general procedure of a character recognition process. It is a figure which shows the method of feature extraction. It is a figure which shows the method of feature extraction. It is a figure which shows the flow of a process of a document data structure part. It is a figure which shows the reconstruction of document data. It is a figure which shows the comparison with a template. It is a figure which shows the comparison with a template. It is a figure which shows the operation | movement of the 4th Embodiment of this invention. It is a figure which shows the operation | movement of the 4th Embodiment of this invention. It is a figure which shows the list | wrist of the difficulty of character recognition. It is a figure which shows a different difficulty table for every font. It is a figure which shows a priority. It is a figure which shows an example of the 7th Embodiment of this invention. It is a figure which shows a priority. It is a figure which shows an example of the 8th Embodiment of this invention. It is a figure which shows an example of the 8th Embodiment of this invention. It is a figure which shows an example of the 10th Embodiment of this invention. It is a figure which shows an example of the 10th Embodiment of this invention. It is a figure which shows an example of the 11th Embodiment of this invention. It is a figure which shows an example of the 12th Embodiment of this invention. It is a figure which shows the RFID chip | tip as a data recording means for machine reading. It is a figure which shows the barcode pattern as a data recording means for machine reading. It is a figure which shows the two-dimensional code pattern as a machine reading data recording means. It is a figure which shows the background dot pattern as a data recording means for machine reading. It is a figure which shows an example in which the strikethrough and the line which comprises a character overlap or approach. It is a figure which shows an example in which the strikethrough and the line which comprises a character overlap or approach.

Explanation of symbols

DESCRIPTION OF SYMBOLS 101 Image reading part 102 Image processing part 103 Character recognition part 104 Character layout information reading part 105 Character layout information decoding part 106 Typeface / placement information extraction part 107 Typeface / placement information conversion part 108 Typeface / placement information recording part 109 Character data structure part DESCRIPTION OF SYMBOLS 110 Image data structure part 111 Data recording part 112 Recording medium 113 Data transfer part 114 Data receiving part 115 Printing part

Claims (14)

Reading means for acquiring a document image and character layout information from a manuscript;
Character recognition processing means for acquiring text data by executing character recognition processing on the document image acquired by the reading means;
Reconstructing means for reconstructing the text data obtained by the character recognition processing means based on the character layout information to obtain restored electronic data;
An information processing apparatus comprising: Reading means for acquiring a document image and character layout information from a manuscript;
Character recognition processing means for executing text recognition processing on the document image acquired by the reading means based on the character layout information acquired by the reading means, and acquiring text data;
An information processing apparatus comprising: The character recognition means includes
A template generation unit for generating a character template based on the character layout information acquired by the reading unit;
3. The character recognition processing unit according to claim 2, further comprising: a character recognition processing unit that performs character recognition processing by collating the document image acquired by the reading unit with the template generated by the template generation unit for each character. Information processing device. The character recognition means further includes
A character separation unit that cuts out characters from the document image based on the character layout information acquired by the reading unit;
4. The character recognition unit according to claim 3, wherein the character recognition unit performs character recognition processing by collating the character cut out from the document image by the character separation unit with the template generated by the template generation unit. Information processing device. The reading unit acquires character recognition auxiliary information in which a predetermined character included in the document image and position information indicating the position of the character in the document are associated with the document image and the character layout information from the document. And
The character recognition means refers to the character recognition auxiliary information based on the position of the character to be processed in the character recognition process in the character recognition unit, and the character recognition auxiliary information includes information corresponding to the position. 5. The information processing apparatus according to claim 3, wherein, in a case where the character recognition is performed, the character corresponding to the position included in the character recognition auxiliary information is output instead of the character recognition processing result.   6. The information processing apparatus according to claim 1, wherein the character layout information includes character font information indicating a character font and character size information indicating a character size.   The information processing apparatus according to claim 6, wherein the character layout information further includes character modification information indicating a modification applied to the character.   The information processing apparatus according to claim 6, wherein the character layout information further includes range information indicating a range of a character string.   The information processing apparatus according to claim 5, wherein the predetermined character included in the character recognition auxiliary information is a character that may be erroneously recognized by the character recognition processing unit.   The information processing apparatus according to claim 5 or 9, wherein characters that may be erroneously recognized by the character recognition processing unit are set according to character font or presence / absence of decoration for characters. .   The information processing apparatus according to claim 1, wherein the reading unit optically scans the document to obtain a document image and character layout information. The reading means includes
A first reading unit that optically scans the original to obtain a document image;
A second reading unit that acquires character layout information from a recording unit attached to the document;
The information processing apparatus according to claim 1, comprising: An information processing apparatus that prints and outputs on a paper-like recording medium having a document image and character layout information,
The document image and character layout information information is acquired by reading means of another information processing apparatus,
Text data is acquired by performing character recognition processing on the document image acquired by the reading unit by the character recognition processing unit of the other information processing apparatus,
A paper-like recording medium, wherein reconstructed electronic data is obtained by reconstructing text data obtained by the character recognition processing means by the reconstructing means of the other information processing apparatus based on the character layout information An information processing device that prints and outputs data. An information processing apparatus that prints and outputs on a paper-like recording medium having a document image and character layout information,
The document image and character layout information information is acquired by reading means of another information processing apparatus,
Character recognition processing is executed on the document image obtained by the reading means by the character recognition processing means of the other information processing apparatus based on the character layout information obtained by the reading means, and text data is obtained. An information processing apparatus that prints and outputs a characteristic paper-like recording medium.

Images