JPH0863545A - Method for deciding direction of character and line in character recognition processor - Google Patents

Abstract

PURPOSE: To accurately decide the direction of a character and a line, which forms the most natural signal word string in language by generating a character lattice and judging the directions of the character and the line based on a constituent word rate. CONSTITUTION: As the result of recognition processing at a character recognition processing part 4, the character lattice being the matrix of plural recognition candidate characters is generated for each character constituting circumscribed lines. A language processing part 6 switches processing in accordance with the presence of information of the character direction and line direction concerning an object area in reference to an information storing part 5. Then, a character direction and line direction judgement part 7 judges the character and line directions which seems to be the correctest in language based on language information extracted by a language processing part 6 with respect to the character string of each circumscribed line. Character recognition processing is executed by each character direction of all the circumscribed lines of a specified area and language processing is executed to the generated character lattice to decide the directions of the character and the line in the specified area based on the constituent word rate of the character lattice.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character recognition processing device (OCR) for inputting a document original as image data and performing character recognition processing, in which the character direction and the line direction are set in units of character areas to be recognized. It is about how to make a decision.

[0002]

2. Description of the Related Art Conventionally, when character recognition processing is performed on a document such as newspapers and magazines in which vertical characters and horizontal characters are mixed, an area to be processed on the character recognition processing device side. It was difficult to accurately specify the character direction and line direction for each. Therefore, before the character recognition process,
In general, an input image is displayed on a display device, and an operator preliminarily specifies the character direction and line direction of the target area, and corrects the result of erroneous recognition to perform character recognition processing. .

In order to reduce the burden of the operator's designation work and correction work, a method of estimating the character direction and the line direction from the area information such as the size and aspect ratio of each area in the document may be adopted. . However, in the determination of the character direction based only on the area information, there are many exceptions and sufficient accuracy cannot be obtained.

Further, in Japanese Laid-Open Patent Publication No. 4-321162, character strings of vertical writing and horizontal writing are created for each area separated by ruled lines, morphological analysis is performed on the character string, and the division is performed. A technique for determining a character string having a small number of unknown words in the character line as a correct character string from the result is described. As described above, the method of estimating the line direction from the linguistic information is presumed to have higher accuracy than the method of determining only the area information.

However, the method described in the above publication is premised on the processing for the area separated by the ruled line in which the vertical direction of the image is correctly input, and there is a problem in determining the line direction of the document image not separated by the ruled line. May occur. Further, the determination of the character direction is not considered.

Even when a character string is generated in the wrong character direction and line direction, the character string determined as the unknown word character string is often reduced as a result of the morphological analysis, and thus an erroneous determination occurs. There are cases.

[0007]

The present invention has been made in view of such circumstances.
In the character recognition processing device, the burden of the operator's designation work and correction work is reduced, and the character direction and line direction of the specific area are determined with high accuracy based on the language information obtained from the Japanese character area specified as the processing target. It is an object of the present invention to provide a method for determining a character direction and a line direction to be performed.

[0008]

In order to achieve the above object, the invention according to claim 1 specifies a character area in an input image, cuts out a character line in the specific area, and A character recognition processing device that performs character recognition processing on character image data of a specific area, performs character recognition processing for each character direction of all circumscribing lines of the specific area, and performs language processing on the generated character lattice. Then, the character direction and the line direction of the specific region are determined based on the constituent word ratio of the character lattice.

According to the second aspect of the invention, a character area in the input image is specified, a character line in the specific area is cut out, and character recognition processing is performed on the character image data in the specific area. In the character recognition processing device to perform, character recognition processing is performed for each character direction of all circumscribing lines of the specific area, language processing is performed on the generated character lattice, and based on the content rate of the independent word of the character lattice. And the character direction and line direction of the specific area are determined.

[0010]

According to the present invention, the circumscribed line of the character area specified as the processing target is extracted, and the character direction is generated for each circumscribed line. The recognition process is performed for each circumscribed line and each character direction to generate the character lattice of the recognition candidate character, and the language process is performed for each character lattice to determine the character string and the language information is extracted. Based on the constituent word rate or the independent word content rate of each character string, the linguistically most appropriate character direction / line direction is determined as the character direction / line direction of the processing area.

[0011]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described in detail below with reference to the drawings. FIG. 1 is a block diagram in which the present invention is applied to a character recognition device. A document to be processed is input as an image image, and in the area identification unit 1, the input image is a character area, a photographic area, and other areas such as a table area. And identify in a rectangular shape.

This area identification processing is described in, for example, Japanese Patent Application Laid-Open No. 5-81475 by the present applicant.
A black run (black pixel connected component) is extracted from the continuity of black pixels in the input document image, a character line is extracted by a compression process that integrates adjacent black runs, and the extracted character lines are further integrated. This is done by a method of distinguishing the character area from other areas. Incidentally, it is possible to specify the area to be processed by the area specifying means such as a mouse in place of the above-mentioned processing means.

In the circumscribing line extraction unit 2, with respect to the region to be processed, which has been rectangularly identified as a character region by the region identification unit 1, the uppermost line, the lowermost line, and the rightmost line located on the outermost side of this character region. , The leftmost row, a total of 4 circumscribed rows are extracted as shown in FIG. This circumscribing line extraction processing is performed by using the extraction result of the character line by the extraction and integration of the black run in the area identification unit 1.

In the character direction generation unit 3, the top line, the bottom line, the rightmost line, and the leftmost line extracted by the circumscribed line extraction unit 2 are four.
For each circumscribed line, as shown in FIG. 3, a total of four character directions of upper / lower / right / left are generated for each circumscribed line.

In the character recognition processing unit 4, first, the information storage unit 5 is referred to in order to confirm the initial values regarding the character direction and line direction of the region to be processed, and the information on the character direction and line direction regarding the region to be processed is stored. Check for the presence. If the information storage unit 5 does not have information about the character direction and line direction of the target area, the upper, lower, and right sides generated by the character direction generation unit 3 for each circumscribed line extracted by the circumscribed line extraction unit 2. -Perform character recognition processing by matching with the dictionary for character recognition in the left four directions. In the matching process of the present invention,
A plurality of candidate characters are output for each character included in each circumscribed line.

As a result of the recognition processing in the character recognition processing unit 4, a character lattice, which is a matrix of a plurality of recognition candidate characters, is generated for each character forming the circumscribed line.
A character lattice is a matrix of candidate characters in which a plurality of recognition candidate characters are held in a grid pattern when a certain character string is recognized, and are separated by the concept of words. It means no state.

The character recognition processing unit 4 performs recognition processing on the characters of the circumscribing line of the specific area in all directions of up, down, left and right for each circumscribing line to generate a total of 16 character lattices. Therefore, the discrimination accuracy is particularly improved when determining the character direction and the line direction.

The language processing section 6 refers to the information storage section 5 to switch the processing depending on the presence / absence of character direction / line direction information regarding the target area. If the information storage unit 5 does not have information about the character direction and line direction of the target area,
The 16 types of character lattices generated by the character recognition processing unit are subjected to language processing by matching with a predetermined language dictionary containing word information and part-of-speech information. This is a process for selecting recognition candidate characters that make up a linguistically correct character string from the character lattice that is a sequence of recognition candidate characters.As a result of the language processing, the character string length and the words that make up the character string are selected. The number, the breakdown of the part of speech of each word, and the notation length of each word are extracted as the language information of the character lattice.

The character direction / line direction determining unit 7 determines the linguistically most probable character direction and line direction based on the linguistic information extracted by the linguistic processing unit 6 for each circumscribed line character string. Is applied. FIG. 4 shows a flowchart of character direction / line direction determination processing.

<Process 1> First, for all the character strings of the circumscribed line, the constituent word rate is calculated from the language information obtained from the language processing unit. The constituent word rate is obtained as a quotient obtained by dividing the number of words constituting a character string by the character string length. The character direction / line direction of the character lattice having the smallest value is determined to be the correct character direction / line direction in the target area. That is, it is determined that the character direction and the line direction in which the most words are formed in the character string of the circumscribing line are the linguistically valid character direction and the line direction.

<Process 2> When there are a plurality of character strings having the smallest constituent word rate, the independent word content rate having a notation length of 1 is calculated for the character string having the smallest constituent word rate. To do. The independent word content rate is obtained as a quotient obtained by dividing the number of independent words having a notation length of 1 by the character string length. The character direction / line direction of the character lattice having the smallest value is determined to be the correct character direction / line direction in the target area. That is, it is determined that the character direction and the line direction in which the number of independent words having a notation length of 1 which is linguistically isolated is the linguistically valid character direction and the line direction.

The process of extracting an independent word having a notation length of 1 is executed by searching the independent word part-of-speech table 8 with the part-of-speech information of each word obtained by subjecting the character lattice to language processing. The independent word part-of-speech table 8 describes the part-of-speech names such as Chinese nouns, Japanese nouns, proper nouns, etc. that have a high probability of becoming independent words that make sense independently, and the part-of-speech information that is the result of language processing for each word. An independent word with a notation length of 1 is extracted by the word length.

When the document to be recognized is a Japanese character string in which Chinese characters and Hiragana are mixed, among the character codes obtained as a result of the language processing, most of the Chinese character character codes have a notation length of 1 character. Recognized by Japanese nouns and proper nouns, most hiragana are often recognized as postpositional particles. At this time, if a character string is formed by mistakenly character direction, the number of independent words with a written length of 1 character is larger than that of the correct answer string, and the number of constituent words is larger than the number of words forming the correct answer string. There is a tendency. In the present invention, by utilizing this feature, the character string most likely to be linguistically selected is selected.

The character direction / line direction determined by the character direction / line direction determination unit 7 is the information storage unit 5 and the previous area information storage unit 9.
Are held by the line cutout unit 10 and the character cutout unit 11
Is input to Line cutout unit 10, character cutout unit 11
Then, in accordance with the line direction and the character direction stored in the information storage unit 5, the character line cutout process and the character cutout process of the entire character area to be processed are performed, and the result is output to the character recognition processing unit 4. . The character recognition processing unit 4 performs a character recognition process for matching the entire character region to be processed with a dictionary for character recognition to generate a character lattice, and a language processing unit 6 verbalizes the character lattice. After selecting the most probable character string, the character recognition result of the processing target area is output from an output device such as a CRT or a printer.

When the information storage unit 5 stores information about the character direction and line direction of the target area, the character recognition processing unit 4 and the language processing unit 6 store the stored character direction,
Character recognition processing and linguistic processing are performed on the character lattice of the entire target area based on the line direction, and output as a character string of the recognition result. Information storage unit 5 after outputting the recognition result of the target area
Is initialized and the same processing is repeated for the next processing target area.

Further, when the character area to be processed includes characters or fonts having a size out of the specification, there may be a plurality of candidates even after the above-mentioned <Process 2> is executed. . In this case, the character direction / line direction determination unit 7 refers to the previous region information storage unit 9 and determines that the region currently being processed has the same character direction / line direction as the region processed immediately before. Then, the information is stored in the character direction / line direction information storage unit 5, and the process proceeds to the line cutout unit / character cutout unit.
It should be noted that it is desirable to set an initial value in the previous area information storage unit 9 according to the usage situation.

Further, in this embodiment, the independent word part-of-speech table 8 is used.
Is independently provided, but the language dictionary of the language processing unit 6 can be configured to include information on the independent word part of speech. In this case, the output of the language processing unit 6 includes the independent word part of speech determination information.

[0028]

According to the first aspect of the present invention, for the circumscribed line of the processing target area, character lattices in all the character directions and line directions that are correct candidates are generated and language processing is performed, and based on the constituent word rate. It is possible to determine with high accuracy the character direction and line direction forming the linguistically most natural word sequence by determining the character direction and line direction.

According to the second aspect of the present invention, for the circumscribed line of the processing target area, character lattices in all the character directions and the line directions that are correct candidates are generated and language processing is performed.
By determining the character direction and the line direction based on the content rate of the independent word, and by further strictly defining the linguistic information, it is possible to highly accurately determine the character direction and the line direction.

Further, according to the present invention, since the character direction and line direction determination processing is executed with only the circumscribed line of the processing target area as the target of the language processing, the character direction and line direction of the processing target area can be accurately adjusted from a small processing area. It is possible to reduce the processing time, which is particularly suitable for the preprocessing of the character recognition processing.

[Brief description of drawings]

FIG. 1 is a block diagram in which the present invention is applied to a character recognition device.

FIG. 2 is a diagram illustrating a circumscribed line extraction process according to the present invention.

FIG. 3 is a diagram illustrating character direction generation according to the present invention.

FIG. 4 is a flowchart of a character direction / line direction determination process according to the present invention.

FIG. 5 is a diagram showing an example of an independent word part-of-speech table.

[Explanation of symbols]

 1 area identification unit 2 circumscribed line extraction unit 3 character direction generation unit 4 character recognition processing unit 5 information storage unit 6 language processing unit 7 character direction / line direction determination unit 8 independent word part-of-speech table 9 previous region information storage unit 10 line cutout unit 11 character cutout

Claims (2)

[Claims] 1. A character recognition processing device which specifies a character area in an input image, cuts out a character line in the specific area, and performs character recognition processing on the character image data of the specific area. Character recognition processing is performed for each character direction of the circumscribing line, linguistic processing is performed on the generated character lattice, and the character direction and line direction of the specific area are determined based on the constituent word ratio of the character lattice. A method for determining a character direction and a line direction in a character recognition processing device characterized by the above. 2. A character recognition processing device which specifies a character area in an input image, cuts out a character line in the specific area, and performs character recognition processing on the character image data in the specific area. Character recognition processing is performed for each character direction of the circumscribed line, and language processing is performed on the generated character lattice, and the character direction and line direction of the specific area are determined based on the content rate of the independent word of the character lattice. A method for determining a character direction and a line direction in a character recognition processing device, characterized by: