JPH08249425A - Document reading method to output character attribute information - Google Patents

Abstract

PURPOSE: To provide a document reading method to output information relating to the attribute of character such as the form and size of the character, etc., by attaching to a read and recognized character code. CONSTITUTION: This document reading method is constituted in such a way that a character string is extracted by analyzing character image data inputted from an image scanner and labeling is applied (S2), and after that, processing (S5) in which the character is recognized by measuring size, processing (S6) which recognizes by measuring the thickness of a character line following the processing, and processing (S7) which recognizes by judging a character form specification based on the recognition results of above two processing are provided in processes (S3), (S4) in which character recognition is performed by segmenting a single character area, and when no judgement and recognition for the character form specification is performed by such processing, a segmented character is rotated by 90 deg., and after that, the character is recognized by repeating processing (S8), (S9) to recognize by judging the character form specification again, and the information relating to the character attribute recognized in the above processing is outputted by attaching on the character code.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method of reading a document in a document reading device, and in particular, a different character type required in the construction of a database using an existing type print document as an information source or a reprint publication printing of an out-of-print book. The present invention relates to a document reading method having a function of extracting a phrase / sentence region emphasized by and outputting it together with a character code.

[0002]

2. Description of the Related Art A document reading device (OCR) 1 includes an image scanner 21 and a character recognition processor 22 as shown in FIG.
Character reading recognition unit 2 and host computer 3
The character recognition processor 22 receives the character image data obtained by optically scanning the document to be read by the image scanner 21 as an input, and first reads the read data according to the process of the flow shown in FIG. The character image data area to be observed and processed as one character is cut out, the cut-out character image data is analyzed, the characteristic parameter contained in the character image data is extracted, and this characteristic parameter is read for each character in the range to be read. Corresponding to the character parameters that belong to each character prepared in advance to match the characteristic parameters are extracted to recognize the read character, the character classification code assigned to the character is output as character information The action is the basic function. The host computer 2 functions as a man-machine interface for setting the reading conditions of the document to be read, displaying the reading result, and the like, and also constructs a database based on the document information obtained by the reading or writes the document. Also performs editing and proofreading work.

[0003]

In the conventional document reading apparatus described above, it is the converted character code that is output, and information such as the type and size of the character font such as Gothic font or Mincho font is output. Is not output, and information such as font and character size cannot be obtained unless the input image is reviewed by a human again. Therefore, when trying to reproduce the input document as it is, or when extracting headings and bold parts for table of contents and index editing, it is necessary for humans to review and process, and this process will take time. .

Further, when extracting a keyword for constructing a database or searching for a filed character, it is not possible to extract a Gothic typeface or a character emphasized in bold, and after the processing by the document reading device, a human being is read again. Has the problem of having to look at the document and enter these new information. The present invention solves the problem in the document reading device based on the related art that information related to the attributes of characters such as the font and size of characters is lost during the character reading recognition process, and the character attribute information is solved. It provides a means to process the input document so that the input document can be reproduced as it is, and extracts words and sentences that are emphasized based on the character attribute information and output them. It is an object of the present invention to provide a document reading method that enables the realization of a document reading device.

[0005]

In order to solve the above problems, the present invention provides a method for reading a document in a document reading apparatus in which character image data optically scanned by an image scanner to be read is stored in a buffer. After analyzing and extracting the character string and labeling, cutting out a single character area, determining the attribute of the cut out character by collating the character pattern dictionary, and assigning the character code to the process of processing,
A process of measuring and recognizing the size of a character, a process of measuring and recognizing the thickness of a character line subsequent to the process, and a process of recognizing a font type based on the recognition result of the two processes. When the character type judgment cannot be recognized by the above process, the cut character is rotated by 90 degrees, and then the character type is judged again and recognized again to recognize the character. Information relating to the character attributes recognized in the above processing is added to the character code and output.

[0006] Thus, after the attribute determination processing of the cut out characters, a processing for blocking the character range recognized as the same font is provided, and a character block within a predetermined character length that does not belong to the font of the text is output as a keyword. Add processing to do. In addition, after the process of blocking the character range of the same font, this process and the process of analyzing the character pattern data to extract the character string and label it based on the recognition results obtained in the process Is provided and a process for detecting and determining a document-specific attribute consisting of a process for detecting a head of line and a process for detecting a heading of a line is performed, and based on the document attribute determination result, the document attribute information and the character attribute information are converted into character codes. Attach and output.

Further, after the process of determining the attribute peculiar to the document, a process of selectively extracting only the document portion having the designated attribute through the input means is provided, and only the document portion having the designated attribute is output and displayed.

[0008]

[Function] Following the character segmentation process of extracting a single character area from the character string extracted from the character image data scanned by the image scanner and stored in the buffer, extracting the character string of the text, the modified line, etc. and labeling the character string. The provided character attribute determination process determines the font based on the measurement result of the size of the cut out character and the thickness of the stroke, and if it cannot be determined, the cut out character is rotated 90 degrees and then the font is determined again. Even if there is a mixture of vertically written English characters after processing, the font is properly recognized, and the character string attribute information and font information acquired in the character string extraction labeling processing and character attribute determination processing are used as the character recognition result. Output by attaching to the character code.

Thus, according to the document reading method, which is provided with a process of blocking the character range in which the fonts are recognized to be the same after the attribute determination process of the cut out characters, within a predetermined character length not belonging to the fonts of the body. The character block of is output as a keyword. In addition, according to the document reading method, after the process of blocking the character range of the same font, the process of detecting and determining the document-specific attribute based on the recognition result obtained by the character string labeling process is provided, -Character string attributes such as medium heading / small heading are added to the character code of the character recognition result and output.

Further, according to the document reading method, after the process of determining the attribute peculiar to the document, the process of selectively extracting only the document portion having the attribute designated through the input means is provided.
Only the document part with the specified attributes is output and displayed.

[0011]

[Example] An image scanner optically scans a document to be read, analyzes character image data stored in a buffer, cuts out an area of one character, and determines the attribute of the cut out character by collating a character pattern dictionary. 1 and FIG. 2 shows the processing flow of one embodiment of the document reading method in which the step of determining the attribute of a character is provided before the step of assigning the character code according to the present invention. 1 shows an example of the processing result when the document reading method of the flow of FIG. 1 is processed, and the method of the present invention will be described with reference to these figures.

In FIG. 1, the steps from the image input S1 to the character cut-out S3 are equivalent to the document image input of A to D and the observation processing in the document reading apparatus according to the prior art described with reference to FIG. The processing up to the character cutting S3 is hierarchically executed through the processing S2 in which the character image data is once regarded as a character string or an image area and labeled.

The processes of S5 to S9 provided during the process of character attribute determination process S4 are the process steps provided according to the present invention, and the character image data cut out as one character is collated with the dictionary. In the process of normalizing and canonicalizing the character size and the line thickness performed to enable the character code attribution determination, the character size is determined in step S5.
In step S6, the thickness of the stroke line of the character is measured and recognized. Then, the process of determining the type classification of the font such as Gothic or Mincho based on the size of the recognized character and the thickness of the stroke is S7, and if a matching font cannot be detected, the process is S8. Rotate the character 90 degrees and press S again
The character type determination and recognition process is repeated at 9 to extract vertically written characters such as English words to determine the typeface and recognize the characters.

In addition, when the document to be emphasized is not only based on the typeface and the character size but also underlines, sidelines, sidepoints, ruby, etc., is to be read, Japanese Patent Laid-Open No. 151396/1989. And JP-A-1-19
The emphasized character string may be extracted by diverting the method of recognizing underlines, sidelines, sidepoints, ruby, etc. disclosed in Japanese Patent No. 6685.

When the character attributes are grasped together with the character code by the above processing, the character attribute is added as shown in FIG. 2 and the character code is output in step S10. Next, a processing flow of one embodiment of a document reading method for extracting and outputting a keyword from a document read based on the second method of the present invention is shown in FIG. 3, and the method of the second invention will be described. .

In FIG. 3, the processes of S1 to S3 and the process of determining the fonts of S5 to S9 in the process S4 are the same as the processes denoted by the same reference numerals in FIG. In the document reading method based on the second invention, a character block forming process S11 for detecting a range in which characters having the same character form continue and grouping is provided subsequent to the character form recognition processing S9 for the cut out characters. A process S12 for outputting, as a keyword, a block that does not belong to the font of the body of the generated character block is provided.

FIG. 4 shows an example of the processing result when the input document is processed by the document reading method of the flow of FIG. 3 based on the second invention. Next, based on the read character information that is divided into blocks according to the font type, a sentence attribute determination process that determines the large heading / medium heading and the small heading is provided, and the large heading / medium heading / small heading in the character is extracted and output. The process flow of one embodiment of the document reading method based on the method of the third invention will be described with reference to FIG.

Also in FIG. 5, the processes from S1 to S11 are the same as the processes with the same reference numerals in FIG. 3, and the headline, the text, and the underline of the document read by the processes up to S11 are processed. Lines and ruby lines are obtained as labeling information, and character codes and font information and a range of characters belonging to the same font are obtained as blocking information.

Subsequent to the character blocking process S11, a process S14 for determining the division between the large heading / medium heading and the small heading based on the labeling information and the typeface / blocking information.
And a text attribute determination process S13 including a line heading determination process S15, and a process S16 for outputting the processing result of the text attribute determination process S13 along with the character attribute information to a character code.

FIG. 6 shows an example of the result output when the document input by the method of the flow of FIG. 5 is processed.
Finally, FIG. 7 shows an example of the processing flow of one embodiment of the document reading method based on the method of the fourth invention, in which only the text part having the attribute designated through the input procedure is output and displayed. The present invention will be described.

In FIG. 7, the image input process S1 to the sentence attribute determination process S13 are the same processes as those designated by the same reference numerals in the process of FIG. 5 according to the third invention described above. By the processes up to S13, the attribute information of the sentence related to the headline and the information related to the font and the size of the character can be obtained. Specified attribute reading processing S17 that reads the attribute specifying information related to the sentence or font to be extracted that is input from the keyboard, and the attribute specifying information that is read in the processing flow is compared with the information obtained by the processing up to processing S13. Attribute selection processing S17 for selecting the data of the attribute to be matched and output processing S18
Causes only the characters in the selected part based on the attribute specification to be output.

FIG. 8 shows an example of the output result when the document input by the method of the flow of FIG. 7 is processed.

[0023]

According to the document reading apparatus for reading and processing a document based on the document reading method of the present invention, not only the character code of the cut-out character but also the character attributes such as the character size and font are determined. Since the information is held as information, the character attribute information and the area extraction result can be combined and the structure of the document to be read and input can be reproduced as it is without the need for additional auxiliary input.

Further, since it becomes possible to selectively extract a specified portion such as an emphasized portion of a Gothic font or a heading portion and output it, an editing operation such as reading and reprinting a document including creation of a table of contents and an index is possible. It is possible to obtain the effect that can be executed accurately and efficiently without manpower. Further, even when a keyword is extracted from a plurality of documents and a database is constructed, the work can be quickly and efficiently performed by using the document reading device to which the document reading method of the present invention is applied.

[Brief description of drawings]

FIG. 1 is a flow chart of a main part of processing in a character recognition method according to the present invention.

FIG. 2 is a diagram illustrating an example of a result of processing of the flow of FIG.

FIG. 3 is a flow chart of processing for extracting and outputting a keyword.

FIG. 4 is a diagram illustrating an example of a result of processing of the flow of FIG.

FIG. 5 is a flow chart of processing for extracting and outputting a text attribute.

FIG. 6 is a diagram illustrating an example of a result of processing of the flow of FIG.

FIG. 7 is a flow chart of processing for extracting and outputting a sentence having a specified attribute.

FIG. 8 is a diagram showing an example of a result of processing of the flow of FIG.

FIG. 9 is a basic configuration diagram of a character reading device.

FIG. 10 is a basic flowchart of character recognition processing in the character reading device.

[Explanation of symbols]

 1 Character reading device 2 Character recognition device 21 Image scanner 22 Character recognition processor 3 Host computer 4 Input document 5 Processing result

Claims (4)

[Claims] 1. An image scanner optically scans a document to be read, analyzes character image data stored in a buffer, extracts a character string and labels it, then cuts out a single character area, and extracts the attribute of the cut out character. Before the step of deciding by assigning a character code by collating the character pattern dictionary, the process of measuring and recognizing the size of the character and the process of measuring the thickness of the character line are performed. A process for recognizing and a process for determining and recognizing the font type based on the recognition results of the above two processes are provided, and when the above-described process cannot determine and recognize the font type, the cut-out character is After rotating once, the process for determining and recognizing the font type is repeated again, and information relating to the font of the character recognized by the above font type determination recognition process and the character string labeling process And information relating to the recognized character string attributes, a document reading method for outputting character attribute information, characterized in that to output subjected to the character recognition result character codes. 2. The process of extracting and labeling a character string includes a process of recognizing characters with underlines, sidelines, sidepoints and ruby as different typefaces, and the same typeface after the attribute determination process of the cut out characters. 2. The document reading method according to claim 1, further comprising: providing a process of blocking the character range recognized as "," and outputting a character block within a predetermined character length that does not belong to the font of the text as a keyword. 3. A large heading / medium heading / small heading is detected based on the recognition result obtained in the processing of dividing the character range of the same font into blocks and the processing of extracting and labeling the character string. A document-specific attribute consisting of a process, a line-heading detection process, and a document-specific attribute detection process is provided. Based on the document attribute determination result, the document attribute information and the character attribute information are attached to the character code. 3. The document reading method according to claim 2, wherein the document is read out and output. 4. A process for selectively extracting only a document part having a specified attribute through the input means after the process for determining a document-specific attribute, and outputting and displaying only the document part having the specified attribute. The document reading method according to claim 3, which is characterized in that.