JPH0352089A - Character information deciding system - Google Patents

Abstract

PURPOSE:To improve the decision rate of character information by comparing the previous and the next characters, when there is a discrepancy as a result of comparison of a character- string read by an OCR and a character-string of a word of a dictionary. CONSTITUTION:When a character-string of 'mo shi hi ka . ku shi yo ri' is read from an OCR 4 and stored in a storage device 3, and compared with a word of a dictionary 5, first characters are the read character 'mo' and the dictionary word 'mo' and comparison (1) coincides, therefore, both of them are brought to comparison (2) by the next character. Since the comparison (2) coincides, when comparison (3) is executed, the read character 'hi' and the dictionary character '''' do not coincide. In this state, the read character 'hi' is fixed, comparison (4) with the next character of the dictionary character is executed, and by the comparison (4), the characters coincide, and also, coincide by comparison (5), as well. When comparison (6) is executed, the characters do not coincide, therefore, comparison (7) is execut ed, but they do not coincide by this comparison either, therefore, this time the dictionary character 'ku' is fixed, brought to comparison (8) with the next character of the read charac ter and they coincide. In such a way, comparison characters ( ) all coincide, and 'mo ji hi ka ku sho ri' (character comparison processing) which corrects an omission of '''' of the read character and a surplus character can be obtained.

Description

【発明の詳細な説明】 〔産業上の利用分野〕 本発明は文字情報判定方式、特にＯＣＲ（光学式文字読
取装置）によって読取った文字列から文字情報を判定す
る文字情報判定方式に関する。DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a character information determination method, and particularly to a character information determination method for determining character information from a character string read by an OCR (optical character reader).

〔従来の技術〕[Conventional technology]

従来、この種の文字情報判定方式は、ＯＣＲによって読
取った文字列と辞書にある単語の文字列との文字の１対
１の比較をとる方式となっている． 〔発明が解決しようとする課題〕 上述した従来の文字情報判定方式は、ＯＣＲによって読
取った文字列と辞書にある単語の文字列との文字の１対
１の比較をとる方式となっているため、ＯＣＲで読込ま
れた文字列に余分な文字があったり、文字抜けがあった
場合には照合がとれないとして、読取り率が低下すると
いう欠点がある． 〔課題を解決するための手段〕 本発明の文字情報判定方式は、ＯＣＲによって読取った
文字列と辞書内の単語の文字列との比較を、まず文字列
の頭から１対１で行い、一致した場合は次に辞書内の単
語の文字列（以下辞書文字列という〉とＯＣＲから読込
んだ文字列（以下読取り文字列という）とのそれぞれの
次の文字を比較していくが、文字の比較が不一致だった
場合は辞書文字列の文字と読取り文字列の文字の両方の
次の文字を比較するのではなく、辞書文字列のある文字
の順番をｍ，読取り文字列のある文字の順番をｎとする
として、ｒｎ＝ｎにおいて不一致がおきた場合 ■まず、読取り文字列のｎ番目の文字を固定し、このｎ
番目の文字と辞書文字列のｍ＋１番目の文字を比較する
。Conventionally, this type of character information determination method involves a one-to-one character comparison between a character string read by OCR and a word character string in a dictionary. [Problem to be solved by the invention] The conventional character information determination method described above is a method that performs a one-to-one character comparison between a character string read by OCR and a word character string in a dictionary. , if there are extra characters or missing characters in the character string read by OCR, the reading rate decreases because matching cannot be performed. [Means for Solving the Problems] The character information determination method of the present invention first compares a character string read by OCR with a word string in a dictionary on a one-to-one basis starting from the beginning of the character string, and then In this case, the character string of the word in the dictionary (hereinafter referred to as the dictionary character string) is compared with the character string read from the OCR (hereinafter referred to as the read character string), and each next character is compared. If the comparison does not match, instead of comparing the next character of both the character in the dictionary string and the character in the read string, the order of the characters in the dictionary string is m, and the order of the characters in the read string is is n, and if a mismatch occurs at rn=n ■First, fix the nth character of the read string, and
Compare the th character with the m+1th character of the dictionary string.

■■の比較で一致したならば、続いてｎ＋１番目とｍ＋
２番目との比較・・・・・・を続ける。If they match in the comparison of ■■, then the n+1st and m+
Continue comparing with the second one.

■また■の比較において不一致だった場合は、今度は辞
書文字列のｍ番目の文字を固定し、ｍ番目の文字と読取
り文字列のｎ＋１番目の文字とを比較する． ■ここで文字の比較が一致したならば続いて、ｎ千２番
目とｍ＋１番目との比較・・・・・・を続ける．■■の
比較において不一致だった場合は、次にｎ＋１番目εｍ
＋１番目との比較・・・・・・を続ける．上記のような
比較をすることにより、読取り文字列に含まれる余分な
１文字および読取り文字列の１文字抜けに対しても、文
字情報としての判定を可能としている． 〔実施例〕 次に、本発明について図面を参照して説明する． 第１図は本発明の一実施例のブロック図で、第２図は第
１図における文字の比較順序を説明するための図である
． 第１図において、比較処理装置１にはバス２を介して記
憶装Ｗ３と辞書５とが接続されている．またＯＣＲ４か
ら読込まれた文字情報は一度、記憶装置３に格納され、
辞書５との文字列比較を比較処理装置１で行い、一定の
規準以上で一致したときに単語に置きかえを行う。■If there is a mismatch in the comparison of ■, then fix the mth character of the dictionary string and compare the mth character with the n+1th character of the read string. ■If the comparison of characters matches, then the comparison between the n1002nd and m+1st characters is continued. If there is a mismatch in the comparison of ■■, then the n+1th εm
+Continue comparison with 1st. By performing the above comparison, it is possible to determine that even one extra character included in the read string and one character missing from the read string can be determined as character information. [Example] Next, the present invention will be explained with reference to the drawings. FIG. 1 is a block diagram of an embodiment of the present invention, and FIG. 2 is a diagram for explaining the comparison order of characters in FIG. 1. In FIG. 1, a comparison processing device 1 is connected to a storage device W3 and a dictionary 5 via a bus 2. In addition, the character information read from the OCR 4 is once stored in the storage device 3,
The character string is compared with the dictionary 5 by the comparison processing device 1, and when the character string matches more than a certain criterion, it is replaced with a word.

次に、ＯＣＲ４から第２図に示すように“モシヒカ・ク
ショリ”という文字列を読取り記憶装置３に格納し、辞
書５の単語と比較を行い、最終的に“モジヒカクショリ
”を得るまでを例にして説明する。まず第２図のように
文字列の先頭の文字から比較を行っていく。最初の文字
は読取った文字“モ”と辞書文字“モ”で比較■は一致
するから両方とも次の文字で比較■を行う．比較■は比
較■と同様に一致するから、比較■を行うと読取った文
字“ヒ”と辞書文字“ｙ”で一致しない。Next, as shown in Figure 2, the character string "Moshihika Kushori" is read from the OCR 4 and stored in the storage device 3, compared with the words in the dictionary 5, and finally "Mojihika Kushori" is obtained. I will explain. First, as shown in Figure 2, comparison is performed starting from the first character of the string. For the first character, compare the read character "mo" with the dictionary character "mo" (■). Since they match, compare them both with the next character (■). Comparison ■ matches the same as comparison ■, so when comparison ■ is performed, the read character "hi" and the dictionary character "y" do not match.

ここで実施例では、読取った文字“ヒ”を固定し、この
文字゛′ヒ”と辞書文字の次の文字との比較■を行う．
比較■では文字が一致し比較■でも一致する．次に比較
■を行うと一致しないため、比較■と同様に比較■を行
うが、ここでも゛読取った文字“・”と辞書文字“シ”
で一致しないため、今度は辞書文字“ク″を固定し、こ
の文字“ク”と読取った文字の次の文字と比較■を行い
一致する．もし、ここで不一致であれば、両方とも次の
文字の比較に移る．比較■，［株］，■については全て
一致し、最終的に読取った文字の“κ”の抜けと“”の
余分な文字を補正した“モジヒカクショリ”を得ること
ができる。In this example, the read character ``hi'' is fixed, and this character ``hi'' is compared with the next character in the dictionary.
The characters match in the comparison ■, and they also match in the comparison ■. Next, when comparison ■ is performed, there is no match, so comparison ■ is performed in the same way as comparison ■, but here too, the read character “・” and the dictionary character “shi”
Since there is no match, this time fix the dictionary character "ku" and compare this character "ku" with the next character after the read character and find a match. If there is a mismatch here, both move on to comparing the next character. Comparisons ■, [stock], and ■ all match, and it is possible to obtain "Mojihikakushori" in which the omission of "κ" and the extra character of "" are corrected in the final read characters.

〔発明の効果〕〔Effect of the invention〕

以上説明したように本発明は、ＯＣＲで読取った文字列
と辞書の単語の文字列との比較で不一致があったとき、
前後の文字比較をすることにより、従来の１対１の文字
比較に比べて文字情報の判定率を飛躍的に向上させるこ
とができる効果がある．As explained above, in the present invention, when there is a mismatch between a character string read by OCR and a character string of a word in a dictionary,
Comparing the characters before and after each other has the effect of dramatically improving the determination rate of character information compared to the conventional one-to-one comparison of characters.

【図面の簡単な説明】[Brief explanation of drawings]

第１図は本発明の一実施例のブロック図、第２図は本発
明の文字の比較順序を説明するための図である． １・・・比較処理装置、２・・・パス、３・・・記憶装
置、４・・・ＯＣＲ、５・・・辞書。FIG. 1 is a block diagram of an embodiment of the present invention, and FIG. 2 is a diagram for explaining the character comparison order of the present invention. 1... Comparison processing device, 2... Path, 3... Storage device, 4... OCR, 5... Dictionary.

Claims (1)

【特許請求の範囲】[Claims] 文字を読取るＯＣＲと、読取った文字列を格納する記憶
装置と、読取った文字列の比較対象となる辞書と、辞書
内の単語の文字列と読取った文字列とを比較して読取っ
た文字列を辞書内の単語の文字情報と判定する文字情報
判定方式において、読取った文字列の文字と辞書内の単
語の文字列との２文字以降に不一致を生じたとき、読取
った文字を固定して辞書内の文字列の次の文字とを比較
する第１の比較手段と、この第１の比較手段で不一致が
生じたとき、辞書内の単語の文字を固定して読取った文
字の次の文字とを比較する第２の比較手段と、この第１
または第２の比較手段で一致が生じたときは一致の生じ
た文字を起点としてそれぞれ次の文字を比較する第３の
比較手段とを有することを特徴とする文字情報判定方式
。An OCR device that reads characters, a storage device that stores the read character strings, a dictionary to which the read character strings are compared, and a character string that is read by comparing the word strings in the dictionary with the read character strings. In the character information determination method that determines character information of a word in a dictionary, when a mismatch occurs after the first two characters between the read character string and the word character string in the dictionary, the read character is fixed. A first comparing means for comparing the next character of the character string in the dictionary, and when a mismatch occurs in this first comparing means, the character next to the character read while fixing the character of the word in the dictionary. a second comparison means for comparing the
Alternatively, when a match occurs in the second comparing means, a third comparing means compares each next character using the character with which the match occurs as a starting point.