JPH01191992A - Character recognizing device - Google Patents

Character recognizing device

Info

Publication number
JPH01191992A
JPH01191992A JP63015729A JP1572988A JPH01191992A JP H01191992 A JPH01191992 A JP H01191992A JP 63015729 A JP63015729 A JP 63015729A JP 1572988 A JP1572988 A JP 1572988A JP H01191992 A JPH01191992 A JP H01191992A
Authority
JP
Japan
Prior art keywords
character
candidate
characters
groups
group
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
JP63015729A
Other languages
Japanese (ja)
Inventor
Toru Futaki
徹 二木
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Canon Inc
Original Assignee
Canon Inc
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Canon Inc filed Critical Canon Inc
Priority to JP63015729A priority Critical patent/JPH01191992A/en
Publication of JPH01191992A publication Critical patent/JPH01191992A/en
Pending legal-status Critical Current

Links

Abstract

PURPOSE:To attain the character recognition with a high correct reading ratio with a small quantity of a memory capacity in a short time by classifying into plural groups of the type of characters which are a recognition object and adding the coupling degree between groups to decision. CONSTITUTION:An input character picture from an input part 1 is inputted through a pre-treatment part 2 to a characteristic extracting part 3 and the extracted character is inputted to a deciding part 5. The candidate character of an attention character and the preceding character obtained by the deciding part 5 are classified into prescribed groups with a group classifying part 6. A re-deciding part 7 re-decides the input character based on the coupling degree of the classification result by a group classifying part 6.

Description

【発明の詳細な説明】 [産業上の利用分野] 本発明は文字認識装置、特に連続した文字列を認識する
文字認識装置に関するものである。
DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a character recognition device, and particularly to a character recognition device that recognizes continuous character strings.

[従来の技術] 、文字認識装置においては、入力された文字画像に対し
て2値化、1文字の切り出し、ノイズ除去等の前処理を
行い、予め定められた手順に従い文字の特徴を抽出し、
予め登録しておいた辞書の標準パターンとの比較を行い
候補の文字コードを出力する。しかし、照合の結果が唯
一に決定されればそのまま最終出力となるが、複数の文
字が候補となる場合には何らかの後処理によって候補を
一つに絞る必要がある。
[Prior art] A character recognition device performs preprocessing such as binarization, cutting out one character, and removing noise on an input character image, and extracts character features according to a predetermined procedure. ,
A comparison is made with standard patterns in a dictionary registered in advance, and candidate character codes are output. However, if the result of matching is determined to be unique, it becomes the final output as it is, but if there are multiple candidates, it is necessary to narrow down the candidates to one through some kind of post-processing.

そこで、従来から複数の候補から1つの文字を選び出す
方法の一つとして、隣接する文字との組み合わせに注目
し、ある一定の長さの文字列の生起頻度を予め調べてお
き、その文字列の生起頻度に基づいて候補文字を特定す
る方法が知られている。
Therefore, one of the conventional methods for selecting one character from multiple candidates is to focus on combinations with adjacent characters, check in advance the frequency of occurrence of character strings of a certain length, and then A method of identifying candidate characters based on frequency of occurrence is known.

[発明が解決しようとする課題] ところが、上記従来例では予め生起頻度を登録しておく
文字列の数が少ないと後処理の効果が表われないため、
十分な数の文字列の生起頻度を登録しておく必要がある
。しかし、対象とする文字の種類が増えるに従って生起
可能な文字列の数は爆発的に増大し、これらの文字列を
すべて記憶しておくには大規模な記憶素子を必要とし、
また文字列との照合にも時間がかかるという欠点があっ
た。
[Problems to be Solved by the Invention] However, in the conventional example described above, if the number of character strings whose occurrence frequencies are registered in advance is small, the effect of post-processing will not be apparent.
It is necessary to register the frequency of occurrence of a sufficient number of character strings. However, as the number of target character types increases, the number of character strings that can be generated increases explosively, and large-scale storage elements are required to store all of these character strings.
Another disadvantage is that it takes time to match character strings.

本発明は、前記従来例の欠点を除去し、少ない記憶容量
で短時間に正読率の高い文字認識を行う文字認識装置を
提供する。
The present invention eliminates the drawbacks of the conventional example and provides a character recognition device that performs character recognition with a high correct reading rate in a short time with a small storage capacity.

[課題を解決するための手段] この課題を解決するために、本発明の文字認識装置は、
入力文字の特徴を辞書と比較して文字の認識を行う文字
認識装置であって、 各文字を複数のグループに分類する分類手段と、注目文
字に対する候補文字のグループと少なくとも1つの近傍
文字のグループとの結合度に基づいて、前記注目文字を
特定する特定手段とを備える。
[Means for Solving the Problem] In order to solve this problem, the character recognition device of the present invention has the following features:
A character recognition device that recognizes characters by comparing the characteristics of an input character with a dictionary, the device comprising a classification means for classifying each character into a plurality of groups, a group of candidate characters for a character of interest, and a group of at least one neighboring character. and specifying means for specifying the character of interest based on the degree of connection with the character.

又、第1候補の文字の確かさを判定する判定手段を更に
備える。
The apparatus further includes a determination means for determining the certainty of the first candidate character.

[作用] かかる構成において、認識対象とする文字の種類を複数
のグループに分類し、グループ間の結合度を予め登録し
ておき、認識しようとする文字とその前後複数個の文字
のそれぞれ属するグループ間の結合度を判定に加味する
[Operation] In such a configuration, the types of characters to be recognized are classified into a plurality of groups, the degree of connectivity between the groups is registered in advance, and the groups to which the character to be recognized and the plurality of characters before and after it belong are determined. The degree of connectivity between the two is taken into account in the judgment.

[実施例] 以下添付図面に従って、実施例の文字認識装置を説明す
る まず、第5図の従来の文字認識装置の構成を示すブロッ
ク図に従って従来例を説明する。51は入力部でOCR
(光学的文字認識装置)の場合はオンライン手書き文字
認識の場合はペンとタブレットによって構成される。5
2は入力された文字画像に対して2値化、1文字の切り
出し。
[Embodiment] A character recognition device according to an embodiment will be described below with reference to the accompanying drawings. First, a conventional example will be described with reference to a block diagram showing the configuration of a conventional character recognition device shown in FIG. 51 is OCR at the input section
(Optical Character Recognition Device) In the case of online handwritten character recognition, it consists of a pen and a tablet. 5
2 binarizes the input character image and cuts out one character.

ノイズ除去等の前処理を行う前処理部で53は予め定め
られた手順に従い文字の特徴を抽出する特徴抽出部、5
4は認識対象とする文字の特徴を標準パターンとして予
め登録しておく辞書部、55は文字の特徴と辞書の標準
パターンとの比較を行い候補の文字コードを出力する判
定部、56は判定部55の判定結果に応じて最終的な出
力を行うための後処理を行う後処理回路である。
53 is a preprocessing unit that performs preprocessing such as noise removal, and a feature extraction unit that extracts character features according to a predetermined procedure;
Reference numeral 4 denotes a dictionary section in which characteristics of characters to be recognized are registered in advance as standard patterns; 55 a determination section that compares character characteristics with standard patterns in the dictionary and outputs candidate character codes; 56 a determination section This is a post-processing circuit that performs post-processing for final output according to the determination result of step 55.

次に第5図の従来の文字認識装置の動作について説明す
る。
Next, the operation of the conventional character recognition device shown in FIG. 5 will be explained.

まず入力部51によって取り込まれた入力文字は、前処
理回路52によって2値化され、図示しない画像バッフ
ァメモリに書き込まれる。さらに前処理回路52ではノ
イズの除去、1文字の切り出しを行い、特徴抽出部53
に送出する。
First, input characters taken in by the input unit 51 are binarized by the preprocessing circuit 52 and written to an image buffer memory (not shown). Furthermore, the preprocessing circuit 52 removes noise and cuts out one character, and the feature extraction unit 53
Send to.

特徴抽出部53では、予め定められた手順に従って文字
線の形状あるいは背景部の特徴等を抽出する認識アルゴ
リズムを実行する。そして判定部55は特徴抽出部53
で抽出された特徴を、辞書54に予め登録されている標
準パターンと照合し候補の文字コードを後処理回路56
に出力する。
The feature extraction unit 53 executes a recognition algorithm that extracts the shape of character lines, features of background parts, etc. according to a predetermined procedure. The determining unit 55 is the feature extracting unit 53.
The extracted features are compared with standard patterns registered in advance in the dictionary 54, and candidate character codes are sent to the post-processing circuit 56.
Output to.

判定部55での照合の結果が唯一に決定されればそのま
ま最終出力となるが、複数の文字が候補となる場合には
、後処理回路56で隣接する文字との組み合わせに注目
し、ある一定の長さの文字列の生起頻度を予め調べてお
き、その文字列の生起頻度に基づいて候補文字を特定す
る。
If the matching result in the determining unit 55 is determined to be unique, it will be the final output as it is, but if there are multiple candidates, the post-processing circuit 56 will focus on the combination with adjacent characters and The frequency of occurrence of a character string of length is checked in advance, and candidate characters are identified based on the frequency of occurrence of the character string.

第1図は本実施例の文字認識装置の構成を示すブロック
図である。
FIG. 1 is a block diagram showing the configuration of a character recognition device according to this embodiment.

lは文字入力を行う入力部、2は入力文字画像に対して
前処理を行う前処理部、3は予め定められたアルゴリズ
ムに従って文字の特徴を抽出する特徴抽出部、4は認識
対象とする候補文字の特徴を予め登録しておく辞書部、
5は入力文字の特徴と辞書部の標準パターンを比較し候
補文字を出力する判定部、6は判定部5での比較によっ
て得られた注目文字の候補文字及び1つ前の文字を所定
のグループに分類するグループ分類部、7はグループ分
類部6による分類結果の結合度に基づいて、入力文字の
再判定を行う再判定部である。
1 is an input unit that inputs characters; 2 is a preprocessing unit that performs preprocessing on input character images; 3 is a feature extraction unit that extracts character features according to a predetermined algorithm; 4 is a candidate for recognition. Dictionary section that registers the characteristics of characters in advance,
5 is a determination unit that compares the characteristics of the input character with the standard pattern of the dictionary unit and outputs candidate characters; 6 is a determination unit that divides the target character candidate character obtained by the comparison in determination unit 5 and the previous character into a predetermined group; A group classification section 7 is a re-judgment section that re-judges input characters based on the degree of connection of the classification results by the group classification section 6.

尚、本実施例では、1つ前の文字の第1候補文字の類似
度が所定値以上の場合のみに、グループの結合度による
再判定を行っている。
In this embodiment, re-determination based on the degree of group connectivity is performed only when the degree of similarity of the first candidate character to the previous character is equal to or greater than a predetermined value.

本実施例では認識対象文字を第3図に示すように英字、
数字、カタカナとし、これら3 fffi類のグループ
をそれぞれG1.G、、G、と呼ぶことにする。もちろ
んこれらのグループはひらがな。
In this embodiment, the characters to be recognized are English letters,
Numbers and katakana, and these three fffi groups are respectively G1. We will call it G,,G,. Of course, these groups are written in hiragana.

漢字等に拡張可能であり、またグループ分けの方法も本
例に限らず任意である。
It can be expanded to include kanji, etc., and the method of grouping is not limited to this example but can be arbitrary.

そして、各グループ間の結合度γ(G I*GJ ) 
 (t、j−1,2,3)を予め定めてオく、γ(Gs
 、 GJ )の例を第4図に示す。
Then, the degree of connectivity between each group γ (GI*GJ)
(t, j-1, 2, 3) is determined in advance, γ(Gs
, GJ) is shown in FIG.

これは数字が大きいほど結合しゃすいことを表していて
、例えば、英字・英字の組み合わせは起こりやすいが英
字・カタカナの組み合わせは起こりにくいことを表して
いる。これらの結合度は認識対象となる文書・伝票等を
解析することによって決定することができる。
This indicates that the larger the number, the easier the combination. For example, the combination of alphabetic characters is more likely to occur, but the combination of alphabetic characters and katakana is less likely to occur. These degrees of connection can be determined by analyzing documents, slips, etc. to be recognized.

次に第2図のフローチャートに従って本実施例の動作に
ついて説明する。
Next, the operation of this embodiment will be explained according to the flowchart shown in FIG.

ステップ5101で入力されたデータに対して、ステッ
プ5102で1文字の切り出し等の前処理が行われ、ス
テップ5103で予め定められたアルゴリズムに従って
特徴抽出が行われる。
The data input in step 5101 is subjected to preprocessing such as cutting out one character in step 5102, and feature extraction is performed in step 5103 according to a predetermined algorithm.

次にステップ5104で入力文字の特徴を予め辞書に1
!録しである候補文字の特徴と比較し、入力文字と各候
補文字との距M(あるいは類似度)を計算する。そして
最も距離の小さい(類似度が大きい)候補文字が第1候
補となり、さらにその距離に従って第1候補が“決定”
であるか“保留“であるかが判定される。“決定“か“
保留“かの判定基準に関しては、第1候補との距離を予
め定められたしきい値と比較することによって判定して
もよいし、第1候補との距離と第2候補との距離の比較
によって判定してもよい。
Next, in step 5104, the characteristics of the input characters are stored in a dictionary in advance.
! The distance M (or degree of similarity) between the input character and each candidate character is calculated by comparing the characteristics of the candidate characters in the record. Then, the candidate character with the smallest distance (highest similarity) becomes the first candidate, and the first candidate is "determined" according to that distance.
It is determined whether it is "on hold" or "on hold". “Decision”?
As for the criteria for determining whether to hold or not, it may be determined by comparing the distance to the first candidate with a predetermined threshold, or by comparing the distance to the first candidate and the distance to the second candidate. It may be determined by

ステップ5105では第1候補が“決定”であるか“保
留”であるかが記憶され、そして“決定“であったとき
にはステップ5107に進み、第1候補が何であるかを
記憶した後、ステップ5ttoでその第1候補を最終候
補として出力する。
In step 5105, it is stored whether the first candidate is "determined" or "pending", and if it is "determined", the process proceeds to step 5107, and after storing what the first candidate is, step 5tto The first candidate is output as the final candidate.

一方、第1候補が“保留”である場合、ステップ510
8に進み、前の文字が“保留”であった場合には、前の
文字の情報量が低いので前の文字の情報は使わずに、ス
テップ5110で第1候補をそのまま最終候補として出
力する。
On the other hand, if the first candidate is “pending”, step 510
Proceed to step 8, and if the previous character is "reserved", the amount of information of the previous character is low, so the information of the previous character is not used, and the first candidate is output as is as the final candidate in step 5110. .

前の文字が“決定”であったときには、各候補文字の属
するグループと前の文字の属するグループとの結合度を
考慮して、ステップ5109で再判定を行う。
When the previous character is "determined", re-determination is performed in step 5109, taking into account the degree of connection between the group to which each candidate character belongs and the group to which the previous character belongs.

以下に再判定のための評価式の例を示す。An example of an evaluation formula for rejudgment is shown below.

d’ (k)−d (k)+f (γ(am、Gb))
k冨1.2.…、N G、:にの属するグループ Gb :前の文字の属するグループ N個の候補が選び出されたとし、候補文字k(k−1,
2,−・・、N)との距離をd (k)で表すことにす
る。そして、kの属するグループを01、前の文字の属
するグループをG、とするとγ(Ga * Gb )に
よって定まる補正量f(γ(ca 、Gb ) )をb
 (k)に加えることによって再判定のための距11f
td’(k)が得られる。
d' (k) - d (k) + f (γ(am, Gb))
k-tomi 1.2. ..., N G, : Group Gb to which the previous character belongs: Suppose that N candidates are selected, and candidate character k(k-1,
2, -..., N) is expressed as d (k). Then, if the group to which k belongs is 01 and the group to which the previous character belongs is G, then the correction amount f(γ(ca, Gb)) determined by γ(Ga * Gb) is b
Distance 11f for re-determination by adding to (k)
td'(k) is obtained.

f(γ)は距離の定義によって異なる適当な関数で、例
えば距離にマハラノビス距離を用いるならばf(γ)は
γの1次式になる。
f(γ) is an appropriate function that varies depending on the definition of distance. For example, if Mahalanobis distance is used for distance, f(γ) becomes a linear expression of γ.

ステップ5109でd’(k)を用いた再判定の結果、
最終的な第1候補が決定されるとステップ5110で出
力される。
As a result of re-determination using d'(k) in step 5109,
Once the final first candidate is determined, it is output in step 5110.

以上のように、最初に認識を行った結果複数の候補が選
び出された場合でも、一つ前の文字の情報を用いて再判
定を行うことによってより高い確率で正しい認識を行う
ことができる。
As described above, even if multiple candidates are selected as a result of initial recognition, correct recognition can be performed with a higher probability by re-judging using information from the previous character. .

尚、本実施例では認識しようとする文字に対して一つ前
の文字との結合度を考慮して再判定を行っていたが、結
合度を考えるのはその文字と一つ前の2文字間だけに限
らない、前2つの文字を含めた3文字のそれぞれの属す
るグループ間の結合度から再判定を行えばより確実な再
判定が可能である。また前の文字だけでなく、認識しよ
うとする文字とその前後の文字を取り出して、やはりそ
れぞれの属するグループ間の結合度を再判定に利用する
ことができる。
Note that in this embodiment, the character to be recognized was redetermined by taking into consideration the degree of combination with the previous character, but the degree of combination is considered based on the character and the previous two characters. A more reliable re-determination is possible if the re-determination is performed based on the degree of connectivity between the groups to which each of the three characters including the previous two characters belong, rather than just the spaces between them. In addition, it is possible to extract not only the previous character but also the character to be recognized and the characters before and after it, and use the degree of connectivity between the groups to which they belong for re-judgment.

以上説明したように、認識対象とする文字の種類を複数
のグループに分類し、グループ間の結合度に基づいて再
判定を行うことによって、文字列の連続性を考慮した正
統率の高い認識を可能にする効果がある。
As explained above, by classifying the types of characters to be recognized into multiple groups and performing re-judgment based on the degree of connectivity between the groups, recognition with a high probability of correctness takes into account the continuity of character strings. It has the effect of making it possible.

又、登録しておく結合度はグループ間のものなので少数
の組み合わせで済み、小さな記憶素子で実現できると共
に、結合度を参照し再判定のための補正値を計算するた
めの時間も極めて短い時間で可能である。
In addition, since the degree of connectivity to be registered is between groups, only a small number of combinations are required, which can be realized with a small memory element, and the time required to refer to the degree of connectivity and calculate the correction value for re-judgment is extremely short. It is possible.

[発明の効果] 本発明により、少ない記憶容量で短時間に正読率の高い
文字認識を行う文字認識装置を提供できる。
[Effects of the Invention] According to the present invention, it is possible to provide a character recognition device that performs character recognition with a high correct reading rate in a short time with a small storage capacity.

更に、正読率をより高くする文字認識装置をも提供した
Furthermore, we also provided a character recognition device that increases the rate of correct reading.

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

第1図は本実施例の文字認識装置の構成を示すブロック
図、 第2図は本実施例の文字認識装置の動作を示すフローチ
ャート、 第3図はグループ分けの例を示す図、 第4図はグループ間結合度の例を示す図、第5図は従来
の技術による文字認識装置の構成を示すブロック図であ
る。 図中、1・・・入力部、2・・・前処理部、3・・・特
徴抽出部、4・・・辞書、5・・・判定部、6・・・グ
ループ分類部、7・・・再判定部である。
Fig. 1 is a block diagram showing the configuration of the character recognition device of this embodiment, Fig. 2 is a flowchart showing the operation of the character recognition device of this embodiment, Fig. 3 is a diagram showing an example of grouping, and Fig. 4 5 is a diagram showing an example of the degree of connectivity between groups, and FIG. 5 is a block diagram showing the configuration of a conventional character recognition device. In the figure, 1... input unit, 2... preprocessing unit, 3... feature extraction unit, 4... dictionary, 5... determination unit, 6... group classification unit, 7...・This is the re-judgment department.

Claims (2)

【特許請求の範囲】[Claims] (1)入力文字の特徴を辞書と比較して文字の認識を行
う文字認識装置において、 各文字を複数のグループに分類する分類手段と、 注目文字に対する候補文字のグループと少なくとも1つ
の近傍文字のグループとの結合度に基づいて、前記注目
文字を特定する特定手段とを備えることを特徴とする文
字認識装置。
(1) In a character recognition device that performs character recognition by comparing the characteristics of an input character with a dictionary, there is a classification means for classifying each character into a plurality of groups, and a group of candidate characters for the character of interest and at least one neighboring character. A character recognition device comprising: specifying means for specifying the character of interest based on a degree of connection with a group.
(2)第1候補の文字の確かさを判定する判定手段を更
に備え、 特定手段は、前記確かさが所定値より低い場合に、注目
文字に対する候補文字のグループと少なくとも1つの近
傍文字のグループとの結合度に基づいて、前記注目文字
を特定することを特徴とする請求項第1項記載の文字認
識装置。
(2) Further comprising determining means for determining the certainty of the first candidate character, and the specifying means, when the certainty is lower than a predetermined value, determines a group of candidate characters and at least one group of neighboring characters for the character of interest. 2. The character recognition device according to claim 1, wherein the character of interest is specified based on the degree of connection with the character.
JP63015729A 1988-01-28 1988-01-28 Character recognizing device Pending JPH01191992A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP63015729A JPH01191992A (en) 1988-01-28 1988-01-28 Character recognizing device

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
JP63015729A JPH01191992A (en) 1988-01-28 1988-01-28 Character recognizing device

Publications (1)

Publication Number Publication Date
JPH01191992A true JPH01191992A (en) 1989-08-02

Family

ID=11896853

Family Applications (1)

Application Number Title Priority Date Filing Date
JP63015729A Pending JPH01191992A (en) 1988-01-28 1988-01-28 Character recognizing device

Country Status (1)

Country Link
JP (1) JPH01191992A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6551152B2 (en) 2000-06-09 2003-04-22 Kawasaki Jukogyo Kabushiki Kaisha Jet-propulsive watercraft
JP4613397B2 (en) * 2000-06-28 2011-01-19 コニカミノルタビジネステクノロジーズ株式会社 Image recognition apparatus, image recognition method, and computer-readable recording medium on which image recognition program is recorded

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6551152B2 (en) 2000-06-09 2003-04-22 Kawasaki Jukogyo Kabushiki Kaisha Jet-propulsive watercraft
JP4613397B2 (en) * 2000-06-28 2011-01-19 コニカミノルタビジネステクノロジーズ株式会社 Image recognition apparatus, image recognition method, and computer-readable recording medium on which image recognition program is recorded

Similar Documents

Publication Publication Date Title
US6950555B2 (en) Holistic-analytical recognition of handwritten text
US5917941A (en) Character segmentation technique with integrated word search for handwriting recognition
EP0564827B1 (en) A post-processing error correction scheme using a dictionary for on-line handwriting recognition
US6834121B2 (en) Apparatus for rough classification of words, method for rough classification of words, and record medium recording a control program thereof
JP2000353215A (en) Character recognition device and recording medium where character recognizing program is recorded
US20210264189A1 (en) Text Recognition Method and Apparatus, Electronic Device, and Storage Medium
US5621818A (en) Document recognition apparatus
JP2009129253A (en) Pattern recognition apparatus, pattern recognition program and pattern recognition method
CN112949523A (en) Method and system for extracting key information from identity card image picture
Song et al. Recognition of merged characters based on forepart prediction, necessity-sufficiency matching, and character-adaptive masking
JP2001337993A (en) Retrieval device and method for retrieving information by use of character recognition result
Madhavaraj et al. Improved recognition of aged Kannada documents by effective segmentation of merged characters
JPH01191992A (en) Character recognizing device
Amrouch et al. A novel feature set for recognition of printed amazigh text using maximum deviation and hmm
US20030123730A1 (en) Document recognition system and method using vertical line adjacency graphs
Lay et al. On-line Chinese character recognition with effective candidate radical and candidate character selections
JPH0520794B2 (en)
Sarkar Word spotting in cursive handwritten documents using modified character shape codes
JP3157530B2 (en) Character extraction method
JPS60138689A (en) Character recognizing method
KR19990010213A (en) Character Recognition Method with Improved Matching Speed
JPH09114926A (en) Method and device for rough classifying input characters for on-line character recognition
Zhang Construction of handwriting databases using transcript-based mapping
Mondal et al. Improved shape code based word matching for multi-script documents
JP2973898B2 (en) Character recognition method and device