JPH0210472B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a character reading system, and more particularly to a character reading system that can read characters in a document whose character pitch is not constant with high precision and at high speed.

In the conventional character reading device, as shown in FIG. The cutting unit 3 cuts out one character at a predetermined interval from the beginning of the character string, transfers the character pattern to the identification unit 4 to determine what character it is, and outputs the identification result to the output terminal. It was designed to output from 5. For this reason, there was a drawback that unless the character pitch on the form was constant, it could not be read accurately.

Further, an auxiliary identification section 6 is provided in the character cutting section 3 of the character reading device, and the auxiliary identification section 6 determines whether only a narrow pattern is one character or a part of a character, and the pattern of one character is If it is determined that the narrow pattern is one character, the character is extracted, and if it is determined to be part of a character, the character is extracted in addition to the next pattern, so that the character pitch is not constant. Devices that allow characters to be read from forms have already been proposed.

However, in the above device, the narrow pattern forms a pattern similar to other character patterns (for example, in a horizontally written form, the kanji character ``ka'', whose partial ``〓'' forms the same character shape as the katakana ``i'' ), when a character is cut into many patterns, or when two or more characters written consecutively are judged as one character, it is difficult to obtain sufficient extraction accuracy. .
Further, when the auxiliary identification section 6 operates, the operation of the character cutting section 3 is stopped until the determination result is output, so that there is a drawback that the reading speed becomes slow.
Moreover, this reduction in reading speed becomes more noticeable as the number of patterns that the auxiliary identification section 6 has to deal with increases.

In order to eliminate the above-mentioned conventional drawbacks, the present invention, when a plurality of patterns exist within a predetermined interval in a character line on a form, performs character extraction by sequentially combining the patterns, and identifies all of them. Based on the results, we selected those that looked more like characters, and the purpose was to develop a character reader that can read characters in documents such as forms with variable character pitch with high precision and at high speed. The goal is to provide a method. The drawings will be explained in detail below.

3 to 8 show an embodiment of the present invention, in which 11 is an input terminal, 12 is a pattern memory, 13 is a character cutting section, 14 is an identification section, 15
1 is a character determining section, and 16 is an output terminal.

To operate this, first, characters on a form are converted into pattern data by a photoelectric conversion device (not shown), and this data is temporarily stored in a pattern memory 12 via an input terminal 11. The character cutting unit 13 cuts out a line pattern 20 including one line of characters as shown in FIG. 4 from the pattern memory 12, scans it in the row direction (direction of arrow Data (hereinafter referred to as "data") where the part where the pattern exists (in the direction of arrow Y) is shown in black ("1" in binary code), and the part where the pattern does not exist is shown in white ("0" in binary code).
This is called black column data. ) Take out 30. Furthermore, the character cutting unit 12 executes a process to be described later based on the black column data 30 to cut out individual patterns 21 from the row pattern 20.
and information regarding the extraction are sequentially sent to the identification unit 14 as a pair of individual data. The identification unit 14 sequentially identifies the characters of only the individual patterns 21 of the individual data, and sequentially sends the identification result (for example, character code) and the information regarding the cutout as a pair of data to the character determination unit 15. The character determination unit 15 performs processing to be described later on the data and sequentially outputs only the correct character identification results from the output terminal 16.

The individual pattern 21 is cut out by the character cutting unit 13 by starting from the beginning of the black row data 30 (hereinafter referred to as the reference position) and starting from a set of black parts (hereinafter referred to as "reference position") existing within a certain interval α set in advance. This is called a cluster of black rows.) If it is one, consider that section as an individual pattern of one character, and if there are multiple, consecutive black row clusters are added one by one and combined. The combined plurality of patterns are each regarded as an individual pattern of one character, and the position excluding the first block among the plurality of blocks of black lines is set as the reference position of the next fixed interval α. Next, a detailed explanation will be given according to the flowchart shown in Fig. 5. In the figure, DNO is a certain interval α
PNO is an operation number that indicates how many times the operation to detect the number N of clusters in the black row has occurred, and PNO is an individual pattern (hereinafter referred to as a combination pattern) based on a combination of clusters in the black row. ) is a pattern number indicating the order of patterns when creating the pattern, and the operation number DNO and pattern number PNO constitute information regarding cutting out.

First, count the number N of black column clusters that exist within a certain interval α from the beginning of the black column data 30, that is, the starting position of character extraction, and when N=0, if the section is a space, DNO=1 , PNO=1 is assigned and the space pattern is sent to the identification unit 14, and if the section is a block of all black columns, the end position of the next block of black columns is detected, and that section is designated as a contact character. Then, force separation is performed on each individual pattern.
It is assigned DNO=1 and PNO=1 and sent to the identification unit 14. When N=1, the section is considered to be an individual pattern of one character, and DNO=1 and PNO=1 are assigned to the individual pattern and sent to the identification section 14.

Furthermore, when N>1, the blocks in the black column that appear from the beginning are sequentially combined without changing the order of appearance of the blocks in the black column, N combination patterns are created, and the operation number DNO and the patterns from 1 to N are combined. It is assigned a number PNO and sent to the identification unit 14. For example, when N=3, if the black rows are a, b, c, in the first process, pattern "a" with DNO=1 and PNO=1,
Pattern “ab” with DNO=1 and PNO=2, DNO=
1, the identification unit 1 identifies a combination pattern with information regarding three cutouts of the pattern “abc” with PNO=3.
Send to 4. Next, remove the block “a” in the first black column and repeat the process with DNO=2.
Pattern “b” with PNO=1, PNO= with DNO=2
The pattern "bc" of 2 is sent to the identification unit 14, and the processing is repeated with DNO=3 except for the block "b" in the black column at the beginning, and the pattern "c" of PNO=1 is sent to the identification unit with DNO=3. 14.

In the character determination unit 15, if a pattern within a certain interval α is considered to be an individual pattern of one character based on the information regarding cutting out, the identification result is output as is, and if it is considered to be a plurality of patterns, then the plurality of patterns are output. Among the identification results of each of the combination patterns, the one with the longest pattern width is outputted as the correct identification result by excluding the ``JECT'' combination pattern, and the block of black columns in the section is output as the pattern. It is designed to exclude subsequent identification results contained within. FIG. 6 is a flowchart showing the details of the processing in the character determination unit 15, in which information about the cutting pattern, that is, whether or not it is a combination pattern of target sections to be output as a reading result from the operation number DNO and pattern number PNO. is determined, the identification result is stored in a buffer, and the one that can be identified with the longest combination pattern from the stored buffer is output as a reading result.

Next, the process of character extraction and character determination will be explained using the line pattern 20 of FIG. 4 as an example. For the patterns "be", "ku", and "to" in the row pattern 20, the number of clusters of black columns within the fixed interval α in the black column data 30 is one, so each individual pattern for each character is 21, and the identification result is sent to the output terminal 16 as is. Since there are two clusters of black columns in a certain interval α (referred to as the target interval here) that includes the next pattern "ru", the character cutting unit 13 sequentially combines the two patterns The information regarding the patterns "NO" and "RU" and their extraction is sent to the identification unit 14, and the position of the black column in the target section excluding the first block "NO" is determined as the next target section. Set as the reference position. Here, two clusters of black rows are detected in the target section as well, and information regarding the combination pattern and its extraction is transmitted in the same manner as described above, and the same applies to the target section below. The identification unit 14 outputs characters such as "NO" and "1" for the pattern "NO" in the target section as identification results, and outputs the character "RU" for the pattern "RU". The character determining unit 15 outputs the character with the largest pattern width and the most accurate identification result, ``RU'', as the reading result in the target section, and at the same time eliminates the identification results of the target section including the pattern ``LE''. The next character is determined based on the identification result. From the identification result of the target section, the character "ka" is output as a reading result, and the next section is excluded. The following pattern "wo",
"Shin" etc. are read as one character in the same way as above. FIG. 7 shows how the line pattern cutout, identification, and character determination processing of FIG. 4 described above is executed, and FIG. 8 shows the flow of the processing.

In this way, according to the above embodiment, a single character pattern or
Since it is possible to distinguish between the sections that are to be extracted as one character and the sections that should constitute multiple combination patterns, it is possible to reliably distinguish between
In addition, when multiple black column clusters exist within a certain interval α, the position excluding the first cluster is used as the reference position for the next interval, so all possible combination patterns can be extracted. , reading accuracy can be improved. In addition, since the character cutting section only needs to mechanically cut out the pattern according to the number of blocks in the black row, there is no need to wait for the identification result of the auxiliary identification section as in the conventional case, and this entire process is configured as a pipeline. You can also speed up the processing.

As explained above, according to the present invention, in a character reading method in which character recognition is performed by cutting out each character from a character line pattern obtained by scanning and photoelectrically converting text on a document, a predetermined The number of clusters of black columns existing in an interval is checked, and if there is one, the interval is treated as a single character pattern and cut out, and if there are multiple clusters, the clusters of black columns are sequentially combined appropriately. A cutting step in which each of the combination patterns is regarded as a pattern of one character and is cut out, and the cut out pattern and information regarding the cutting out are outputted; If it is considered as multiple patterns, output the identification result as is, and if it is considered as multiple patterns, output the identification result corresponding to the combination pattern with the longest pattern width from among the identification results of each of the multiple combination patterns. Because it has a character determination process to output, it is possible to unambiguously cut out characters from a document that includes separated characters or cut characters and whose character pitch is not constant, without requiring complicated identification or judgment. In addition to speeding up the reading speed, even when some characters are similar to other characters, they can be read correctly. The patterns in which the clusters of columns are sequentially increased one by one and combined are treated as one character pattern and cut out, and the position of the clusters of the plurality of black columns excluding the first cluster is calculated in the next certain interval. The reference position has the advantage that all possible combination patterns can be extracted, the reading accuracy can be improved, and the range of objects to be read can be expanded.

[Brief explanation of drawings]

The drawings serve to explain the present invention; FIG. 1 is a block diagram showing a conventional character reading device, FIG. 2 is a block diagram showing another conventional character reading device, and FIG. 3 is a block diagram showing another conventional character reading device.
The figure is a block diagram showing one embodiment of a character reading device to which the method of the present invention is applied, FIG. 4 is an explanatory diagram showing an example of a row pattern and its black column data, and FIG. , FIG. 6 is a flowchart of the character determination unit 15, FIG. 7 is an explanatory diagram showing the execution of cutting, identification, and character determination processing for the line pattern of FIG. 4, and FIG. 8 is a flowchart of the processing of FIG. 7. It is an explanatory diagram showing a flow. 11...Input terminal, 12...Pattern memory, 13
...Character cutting section, 14...Identification section, 15...Character determining section, 16...Output terminal.

Claims (1)

[Scope of Claims] 1. In a character reading method that performs character recognition by cutting out each character from a pattern of character lines obtained by scanning and photoelectrically converting text on a document, a character existing within a predetermined interval on a character line Find out the number of blocks in the black column, and if there is one, consider that section as a pattern of one character and cut it out; if there are multiple blocks, create multiple combination patterns by suitably combining the blocks in the black column sequentially. A cutting process in which each pattern is regarded as a single character pattern and is output, and the cutout pattern and information regarding the cutout are output.If the cutout pattern is recognized as a single character pattern, the cutout process is performed. Character determination that outputs the identification result as is, and if it is considered as multiple patterns, outputs the identification result corresponding to the combination pattern with the longest pattern width from among the identification results of each of the multiple combination patterns. A character reading method characterized by having a process. 2 In a character reading method that performs character recognition by cutting out characters one by one from a character line pattern obtained by scanning and photoelectrically converting text on a form, a cluster of black columns that exists within a predetermined interval on a character line is used. If there is one, consider the interval as a pattern of one character and cut it out. If there are multiple, consecutive black rows are sequentially increased by one and the combined patterns are regarded as patterns of one character. A cutting step of cutting out the plurality of black row blocks, and setting a position obtained by removing the first block among the blocks of the plurality of black rows as a reference position for the next certain section, and outputting the cut out pattern and information regarding the cutting out, and the cutting out. If the pattern is considered to be a single character pattern, the identification result is output as is, and if it is considered to be multiple patterns, each of the multiple combination patterns is output. 1. A character reading method comprising: a character determining step of outputting an identification result corresponding to a combination pattern having the longest pattern width from among the identification results.