JPH1153469A - Character segmentation device, optical character reader and storage medium - Google Patents

Character segmentation device, optical character reader and storage medium

Info

Publication number
JPH1153469A
JPH1153469A JP9210480A JP21048097A JPH1153469A JP H1153469 A JPH1153469 A JP H1153469A JP 9210480 A JP9210480 A JP 9210480A JP 21048097 A JP21048097 A JP 21048097A JP H1153469 A JPH1153469 A JP H1153469A
Authority
JP
Japan
Prior art keywords
character
circumscribed rectangle
circumscribed
line segment
pattern
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
JP9210480A
Other languages
Japanese (ja)
Inventor
Teruki Oikawa
晃樹 及川
Takahiro Oura
貴裕 大浦
Hiromi Kida
博巳 木田
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.)
N T T DATA KK
NTT Data Group Corp
Original Assignee
N T T DATA KK
NTT Data Corp
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 N T T DATA KK, NTT Data Corp filed Critical N T T DATA KK
Priority to JP9210480A priority Critical patent/JPH1153469A/en
Publication of JPH1153469A publication Critical patent/JPH1153469A/en
Pending legal-status Critical Current

Links

Landscapes

  • Character Input (AREA)

Abstract

PROBLEM TO BE SOLVED: To provide an optical character reader capable of recognizing characters highly precisely. SOLUTION: The superimposing of plural character circumscribing rectangles for respectively indicating character components is detected in a superimposing boundary detection part 15. A superimposing boundary correction part 16 sets a superimposing boundary area between the character circumscribing rectangle and the character circumscribing rectangle adjacent to it for which the superimposing is detected, specifies the character circumscribing rectangle to which a line segment circumscribing rectangle present inside the area is to belong and adds the coordinate information of the line segment circumscribing rectangle to the pertinent character circumscribing rectangle. A pattern extraction part 14 extracts (segments) the character pattern of the area indicated by the coordinate information of the circumscribing rectangle and the line segment circumscribing rectangle from an image memory 11 and preserves it in a pattern memory 17. A character recognition part 18 performs the recognition processing of the character pattern inside the pattern memory 17.

Description

【発明の詳細な説明】DETAILED DESCRIPTION OF THE INVENTION

【0001】[0001]

【発明の属する技術分野】本発明は、光学的手法に基づ
く文字認識技術に係り、例えば枠無し罫線枠(フリーピ
ッチ枠)の領域内の文字パタンの切り出し及びその読み
取りを効率的に行う手法に関する。
BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character recognition technique based on an optical technique, and more particularly to a technique for efficiently cutting out and reading a character pattern in an area of a frameless ruled line frame (free pitch frame). .

【0002】[0002]

【従来の技術】帳票等に印刷ないし手書きされた文字を
光学的手段により走査し、当該文字の画像イメージを切
り出してその自動認識を行う光学式文字読取装置(OC
R:Optical Character Reader)が知られている。
2. Description of the Related Art An optical character reader (OC) which scans a character printed or handwritten on a form or the like by optical means, cuts out an image of the character, and automatically recognizes it.
R: Optical Character Reader) is known.

【0003】図5は、この種の従来のOCRの機能ブロ
ック図である。このOCR5は、イメージメモリ11、
外接方形抽出部12、外接方形統合部13A、パタン抽
出部14、パタンメモリ17、及び文字認識部18を具
備し、例えば帳票等の画像イメージ(以下、帳票イメー
ジ)内のフリーピッチ枠内にある文字を抽出して文字認
識を行っている。外接方形抽出部12、外接方形統合部
13A、パタン抽出部14、及びパタンメモリ17によ
り文字切り出し装置が構成される。
FIG. 5 is a functional block diagram of a conventional OCR of this kind. The OCR 5 has an image memory 11,
It includes a circumscribed rectangle extracting unit 12, a circumscribed rectangle integrating unit 13A, a pattern extracting unit 14, a pattern memory 17, and a character recognizing unit 18, and is in a free pitch frame in an image image of a form (hereinafter, form image), for example. Character recognition is performed by extracting characters. The character extracting device is configured by the circumscribed rectangle extracting unit 12, the circumscribed rectangle integrating unit 13A, the pattern extracting unit 14, and the pattern memory 17.

【0004】イメージメモリ11は、図示しないスキャ
ナ等の光学的走査手段より入力される帳票イメージを格
納するものである。外接方形抽出部12は、イメージメ
モリ11内の帳票イメージから黒画素の連結領域を囲む
外接方形(線分外接方形と称する)を抽出する。具体的
には、イメージメモリ11内の帳票イメージから黒画素
が連結する複数の閉領域を各々特定するとともに、各閉
領域において黒画素の連結領域を囲むことが可能な最小
の矩形の始点座標及び終点座標を各々算出して線分外接
方形を決定する。決定した複数の線分外接方形は、それ
ぞれ外接方形統合部13Aに入力される。
[0004] The image memory 11 stores a form image input from optical scanning means such as a scanner (not shown). The circumscribed rectangle extracting unit 12 extracts a circumscribed rectangle (referred to as a line segment circumscribed rectangle) surrounding the connected area of the black pixels from the form image in the image memory 11. More specifically, a plurality of closed regions to which black pixels are connected are specified from the form image in the image memory 11, and the starting point coordinates of the smallest rectangle capable of surrounding the connected region of black pixels in each closed region. The coordinates of the end points are calculated to determine the circumscribed rectangle of the line segment. The determined plurality of line segment circumscribed rectangles are input to the circumscribed rectangle integration unit 13A.

【0005】外接方形統合部13Aは、外接方形抽出部
12より入力された複数の線分外接方形を第1方向(例
えば縦方向)に統合するとともに、第2方向(例えば横
方向)の文字幅の推定を行うことで統合後の線分外接方
形を第2方向に分割し、文字パタンを囲む外接方形を
(文字外接方形と称する)を作成する。作成された文字
外接方形は、先頭文字に対応するものから順次パタン抽
出部14に入力される。
The circumscribed rectangle integrating unit 13A integrates a plurality of line segment circumscribed rectangles input from the circumscribed rectangle extraction unit 12 in a first direction (for example, a vertical direction) and a character width in a second direction (for example, a horizontal direction). Is performed, the line segment circumscribed rectangle after integration is divided in the second direction, and a circumscribed rectangle surrounding the character pattern is created (referred to as a character circumscribed rectangle). The created character circumscribed rectangle is sequentially input to the pattern extraction unit 14 from the one corresponding to the first character.

【0006】パタン抽出部14は、イメージメモリ11
上の帳票イメージから、文字外接方形の座標情報に対応
する領域の文字パタンを抽出してパタンメモリ17に格
納する。文字認識部18は、パタンメモリ17内の文字
パタンの文字認識を行い、認識結果を図示しない出力手
段に対して出力する。
[0006] The pattern extraction unit 14 is provided in the image memory 11.
The character pattern in the area corresponding to the coordinate information of the character circumscribed rectangle is extracted from the above form image and stored in the pattern memory 17. The character recognition unit 18 performs character recognition of the character pattern in the pattern memory 17 and outputs a recognition result to an output unit (not shown).

【0007】次に、上記OCR5の動作を具体的に説明
する。便宜上、図6に示すように、文字列”東京都”を
含む帳票イメージがイメージメモリ11に格納されてい
るものとする。この帳票イメージは、文字枠が設けられ
たフィールドの一部であり、X−Y座標系のうち、X軸
方向を横方向(第2方向)、Y軸方向を縦方向(第1方
向)として設定してある。
Next, the operation of the OCR 5 will be specifically described. For convenience, it is assumed that a form image including the character string “Tokyo” is stored in the image memory 11 as shown in FIG. This form image is a part of a field provided with a character frame. In the XY coordinate system, the X-axis direction is a horizontal direction (second direction) and the Y-axis direction is a vertical direction (first direction). It has been set.

【0008】外接方形統合部13Aでは、線分外接方形
を下記の基準によって統合する。まず、対象となる2つ
の線分外接方形の重なり部分の長さ“a”と、各線分外
接方形のうちで小さい方の長さ“b”とを求める。次に
“a/b”を演算し、演算値が、予め定めた線分外接方
形の統合閾値th1以上かどうかを判定する。演算値が
統合閾値th1以上の場合は統合して一次外接方形とな
し、統合閾値に満たない場合は統合しない。図7(a)
は、この場合の模式図、図7(b)は、図7(a)に示
した2つの外接方形を統合した一次外接方形の例を示す
ものである。外接方形統合部13Aは、また、作成した
一次外接方形から標準文字幅swを推定する。この場合
の標準文字幅swは、例えば、一次外接方形に含まれる
線分外接方形の幅の平均である。外接方形統合部13A
は、この標準文字幅swに基づいて一次外接方形の横方
向の分割を行う。例えば、図7(b)に示した一次外接
方形の実測した幅wiと標準文字幅swとの比“wi/
sw”が予め定めた分割閾値th2以上の場合には、一
次外接方形を分割して複数の文字外接方形とする。
The circumscribed rectangle integrating unit 13A integrates line segment circumscribed rectangles based on the following criteria. First, the length “a” of the overlapping portion of two target line segment circumscribed rectangles and the smaller length “b” of each line segment circumscribed rectangle are determined. Next, “a / b” is calculated, and it is determined whether or not the calculated value is equal to or greater than a predetermined line segment circumscribed rectangle integration threshold th1. If the calculated value is equal to or greater than the integration threshold th1, the result is integrated into a primary circumscribed rectangle, and if the calculated value is less than the integration threshold, the integration is not performed. FIG. 7 (a)
Is a schematic diagram in this case, and FIG. 7 (b) shows an example of a primary circumscribed rectangle obtained by integrating the two circumscribed rectangles shown in FIG. 7 (a). The circumscribed rectangle integrating unit 13A also estimates the standard character width sw from the created primary circumscribed rectangle. The standard character width sw in this case is, for example, the average of the widths of the line circumscribed rectangles included in the primary circumscribed rectangle. Bounding rectangular integration unit 13A
Performs a horizontal division of the primary circumscribed rectangle based on the standard character width sw. For example, the ratio "wi /" of the measured width wi of the primary circumscribed rectangle shown in FIG.
If sw "is equal to or greater than a predetermined division threshold th2, the primary circumscribed rectangle is divided into a plurality of character circumscribed rectangles.

【0009】図8は、“東京都”に対応する3つの文字
外接方形に分割された一次外接方形の例を示す図であ
る。文字外接方形は、個々の文字相当領域を囲む最小矩
形の始点座標及び終点座標によって表すことができる。
図8の例では、先頭文字“東”における文字外接方形
は、始点座標が(X1、Y1)で終点座標が(X2、Y
2)、二番目の文字“京”における文字外接方形は、始
点座標が(X3(<X2)、Y3)で終点座標が(X
4、Y4)、最終文字“都”における文字外接方形は、
始点座標が(X5(<X4)、Y5)で終点座標が(X
6、Y6)でそれぞれ表される。これらの座標情報は、
それぞれ該当する文字外接方形に対応付けられてパタン
抽出部14に渡される。
FIG. 8 is a diagram showing an example of a primary circumscribed rectangle divided into three character circumscribed rectangles corresponding to "Tokyo". The character circumscribed rectangle can be represented by the start point coordinates and end point coordinates of the smallest rectangle surrounding each character equivalent area.
In the example of FIG. 8, the character circumscribed rectangle in the first character "East" has a start point coordinate of (X1, Y1) and an end point coordinate of (X2, Y
2), the character circumscribed rectangle of the second character “K” has a start point coordinate of (X3 (<X2), Y3) and an end point coordinate of (X
4, Y4), the character circumscribed rectangle in the final character “To” is
The start point coordinates are (X5 (<X4), Y5) and the end point coordinates are (X5
6, Y6). These coordinate information are
The corresponding character circumscribed rectangle is passed to the pattern extraction unit 14 in association with the corresponding character.

【0010】パタン抽出部14は、各文字外接方形の座
標情報に基づいて、イメージメモリ11から該当する領
域の文字パタン“東”、“京”、“都”を切り出し、こ
れらを切り出し順にパタンメモリ17に格納する。切り
出された文字パタンの例を図9(a)、(b)、(c)
に示す。図9に示される例では、各文字外接方形につい
て重なりが発生している。このような場合、文字外接方
形の座標情報に対応する領域の文字パタンを単純に切り
出しただけでは、隣接文字による線分の侵入が発生し、
文字認識精度の低下を招く。そこで、従来、パタン抽出
部14において、侵入した線分を除去するための処理が
行われている。すなわち、図10(a)に示すように、
個々の文字外接方形上に文字線境界a1、a2を設定
し、その後、文字線境界a1を開始点、文字線境界a2
を終了点として、侵入した線分の境界を図示のようにト
レースし、パタンメモリ17上の当該トレース範囲内を
除去するように白点化を行う。このようにしてパタンメ
モリ17上の文字パタンを補正した後、文字認識部18
において文字認識を行う。
The pattern extracting section 14 extracts character patterns "East", "K", and "Miya" of the corresponding area from the image memory 11 based on the coordinate information of each character circumscribed rectangle, and extracts these in the order of extraction. 17 is stored. FIGS. 9A, 9B, and 9C show examples of cut-out character patterns.
Shown in In the example shown in FIG. 9, overlapping occurs for each character circumscribed rectangle. In such a case, simply cutting out the character pattern in the area corresponding to the coordinate information of the character circumscribed rectangle may cause intrusion of a line segment by an adjacent character,
This leads to a reduction in character recognition accuracy. Therefore, conventionally, a process for removing the invading line segment is performed in the pattern extraction unit 14. That is, as shown in FIG.
The character line boundaries a1 and a2 are set on the individual character circumscribed rectangles.
Is set as the end point, the boundary of the invading line segment is traced as shown in the figure, and white spotting is performed so as to remove the inside of the trace range on the pattern memory 17. After correcting the character pattern on the pattern memory 17 in this way, the character recognition unit 18
Perform character recognition.

【0011】[0011]

【発明が解決しようとする課題】上述のように、従来の
OCR5において、文字外接方形の重なりが発生する場
合は、文字外接方形に対応する文字パタンを単純に切り
出しただけでは、隣接文字外接方形による線分の侵入が
発生し、文字認識精度の低下を招く。
As described above, in the conventional OCR5, when overlapping of character circumscribed rectangles occurs, simply cutting out a character pattern corresponding to the character circumscribed rectangle will result in an adjacent character circumscribed rectangle. , A line segment is invaded, thereby lowering the character recognition accuracy.

【0012】また、侵入した線分の除去を行う場合は、
少なくとも文字線境界をトレースする分だけ処理時間が
余分にかかる。さらに、従来は、線分の侵入の発生頻度
が少ないような場合においても、トレースを含む線分除
去処理を行っていたために、文字認識を開始するまで時
間がかかり、それがOCR全体の処理速度の低下を招い
ていた。
In the case of removing the penetrated line segment,
Extra processing time is required at least for tracing the character line boundary. Further, conventionally, even in the case where the intrusion frequency of the line segment is low, the line segment including the trace is removed, so that it takes time until the character recognition is started, which is the processing speed of the entire OCR. Had been reduced.

【0013】本発明の課題は、文字認識に用いる文字パ
タンを迅速に切り出すことができる、改良された文字切
り出し装置を提供することにある。本発明の他の課題
は、文字認識精度を一定値以上に維持するとともに、高
速な文字認識処理を可能にする光学式文字読取装置、及
び文字切り出し装置を汎用のコンピュータ装置で実現す
るための記録媒体を提供することにある。
SUMMARY OF THE INVENTION An object of the present invention is to provide an improved character extracting device capable of quickly extracting a character pattern used for character recognition. Another object of the present invention is to provide an optical character reader and a character cutout device that can perform high-speed character recognition processing while maintaining the character recognition accuracy at a certain value or more, and a recording method for realizing the character cutout device with a general-purpose computer device. To provide a medium.

【0014】[0014]

【課題を解決するための手段】上記課題を解決する本発
明の文字切り出し装置は、複数の文字パタンを含む画像
イメージから線分外接方形を抽出し、抽出した線分外接
方形を第1方向に統合するとともに、統合後の線分外接
方形を第1方向と垂直の第2方向に分割して複数の文字
外接方形を作成する文字外接方形作成手段と、前記第2
方向に隣接する一対の文字外接方形の一方の終点座標と
他方の始点座標との差分が所定の重畳検出閾値以上かど
うかを判定し、判定結果に応じたサイズの文字切り出し
領域を指定する重畳検出手段と、前記重畳検出手段の出
力情報に基づいて前記画像イメージから該当する文字パ
タンを切り出すパタン抽出手段と、を備えてなる。
A character segmenting apparatus according to the present invention for solving the above problems extracts a line segment circumscribed rectangle from an image including a plurality of character patterns, and converts the extracted line segment circumscribed rectangle in a first direction. A character circumscribed rectangle creating means for creating a plurality of character circumscribed rectangles by integrating and dividing the combined line segment circumscribed rectangle in a second direction perpendicular to the first direction;
Superimposition detection that determines whether the difference between one end point coordinate and the other start point coordinate of a pair of character circumscribed rectangles adjacent in the direction is equal to or greater than a predetermined superimposition detection threshold, and specifies a character cutout area of a size according to the determination result Means, and pattern extracting means for extracting a corresponding character pattern from the image based on output information of the superimposition detecting means.

【0015】上記文字切り出し装置において、前記重畳
検出手段は、例えば、前記差分が前記重畳検出閾値に満
たない場合は個々の文字外接方形の座標情報を前記文字
切り出し領域として指定し、一方、前記差分が前記重畳
検出閾値以上の場合は、各文字外接方形を他の文字外接
方形との重畳を回避するサイズに変更するとともに前記
差分の領域に存する線分外接方形の座標情報とその線分
外接方形が属すべき文字外接方形の座標情報とを前記文
字切り出し領域として指定するように構成される。な
お、前記差分の領域は、所定のマージン値に基づいて自
動設定されるようにする。
In the above-described character segmentation apparatus, for example, when the difference is less than the overlap detection threshold, the superimposition detecting means designates coordinate information of an individual circumscribed rectangle as the character segmentation area. Is greater than or equal to the superimposition detection threshold, the size of each character circumscribed rectangle is changed to a size that avoids superimposition with other character circumscribed rectangles, and the coordinate information of the line segment circumscribed rectangle existing in the difference area and the line segment circumscribed rectangle And the coordinate information of a character circumscribed rectangle to which the character belongs. The difference area is automatically set based on a predetermined margin value.

【0016】また、上記他の課題を解決する本発明の光
学式文字読取装置は、光学的走査手段に読み取られた複
数の文字成分を含む画像イメージを格納するイメージメ
モリと、前記イメージメモリに格納された画像イメージ
から複数の線分外接方形を抽出し、抽出された複数の線
分外接方形を第1方向に統合するとともに統合後の線分
外接方形を第1方向と垂直の第2方向に分割して複数の
文字外接方形を作成する文字外接方形作成手段と、前記
第2方向に隣接する一対の文字外接方形の一方の終点座
標と他方の始点座標との差分が所定の重畳検出閾値以上
かどうかを判定し、判定結果に応じたサイズの文字切り
出し領域を指定する重畳検出手段と、前記重畳検出手段
の出力情報に基づいて前記画像イメージから該当する文
字パタンを切り出すパタン抽出手段と、切り出された文
字パタンに基づいて文字認識を行う文字認識手段と、を
備えてなる。
According to another aspect of the present invention, there is provided an optical character reading apparatus for storing an image including a plurality of character components read by an optical scanning means, and storing the image in the image memory. A plurality of line segment circumscribed rectangles are extracted from the extracted image, the extracted line segment circumscribed rectangles are integrated in a first direction, and the combined line segment circumscribed rectangles are integrated in a second direction perpendicular to the first direction. A character circumscribed rectangle creating means for creating a plurality of character circumscribed rectangles by dividing, and a difference between one end point coordinate and the other start point coordinate of the pair of character circumscribed rectangles adjacent in the second direction is equal to or more than a predetermined superimposition detection threshold value Determining whether a character cutout area of a size corresponding to the determination result is provided, and extracting a corresponding character pattern from the image based on output information of the superimposition detection means. Consisting includes a pattern extracting unit, a character recognition means for performing character recognition based on the extracted character patterns, the.

【0017】さらに、上記他の課題を解決する本発明の
記録媒体は、下記の処理をコンピュータに実行させるプ
ログラムを記録してなるコンピュータ読み取り可能な記
録媒体である。 (1)複数の文字パタンを含む画像イメージから線分外
接方形を抽出する処理、(2)抽出した線分外接方形を
第1方向に統合するとともに、統合後の線分外接方形を
第1方向と垂直の第2方向に分割して複数の文字外接方
形を作成する処理、(3)前記第2方向に隣接する一対
の文字外接方形の一方の終点座標と他方の始点座標との
差分が所定の重畳検出閾値以上かどうかを判定する処
理、(4)前記差分が前記重畳検出閾値に満たない場合
は個々の文字外接方形の座標情報を文字切り出し領域と
して指定し、一方、前記差分が前記重畳検出閾値以上の
場合は、各文字外接方形を他の文字外接方形との重畳を
回避するサイズに変更するとともに前記差分の領域に存
する線分外接方形の座標情報とその線分外接方形が属す
べき文字外接方形の座標情報とを文字切り出し領域とし
て指定する処理、(5)前記指定された座標情報に基づ
いて前記画像イメージから該当する文字パタンを切り出
す処理。
Further, a recording medium according to the present invention for solving the above-mentioned other problems is a computer-readable recording medium in which a program for causing a computer to execute the following processing is recorded. (1) a process of extracting a line segment circumscribed rectangle from an image including a plurality of character patterns; (2) integrating the extracted line segment circumscribed rectangle in a first direction, and extracting the line segment circumscribed rectangle after integration in the first direction (3) a process of creating a plurality of character circumscribed rectangles by dividing in a second direction perpendicular to the first direction, and (3) determining a difference between one end point coordinate and the other start point coordinate of a pair of character circumscribed rectangles adjacent in the second direction. (4) When the difference is less than the superimposition detection threshold, the coordinate information of each circumscribed rectangle is designated as a character cutout area. If the detection threshold value or more, each character circumscribed rectangle should be changed to a size that avoids overlapping with another character circumscribed rectangle, and the coordinate information of the line circumscribed rectangle existing in the difference area and the line circumscribed rectangle should belong Character circumscribed square Processing for specifying the index information as a character cutout region, (5) on the basis of the specified coordinates information cutting out a character pattern corresponding from the picture image processing.

【0018】[0018]

【発明の実施の形態】以下、図面を参照して本発明の実
施の形態を詳細に説明する。図1は、本発明の文字切り
出し装置を具備した光学式文字読取装置(OCR)の実
施の形態を表す機能ブロック図である。本実施形態のO
CR1は、コンピュータ装置が所定のプログラムを読み
込んで実行することにより形成される、イメージメモリ
11、外接方形抽出部12、外接方形統合部13、パタ
ーン抽出部14、重畳境界検出部15、重畳境界補正部
16、パタンメモリ17、及び文字認識部18、を備
え、外接方形抽出部12、外接方形統合部13、パタン
抽出部14、重畳境界検出部15、重畳境界補正部1
6、及びパタンメモリ17によって文字切り出し装置を
構成している。
Embodiments of the present invention will be described below in detail with reference to the drawings. FIG. 1 is a functional block diagram showing an embodiment of an optical character reading device (OCR) provided with a character cutout device of the present invention. O of this embodiment
The CR1 is formed by a computer device reading and executing a predetermined program, and is formed by an image memory 11, a circumscribed rectangle extraction unit 12, a circumscribed rectangle integration unit 13, a pattern extraction unit 14, a superimposition boundary detection unit 15, a superimposition boundary correction. A circumscribed rectangle extraction unit 12, a circumscribed rectangle integration unit 13, a pattern extraction unit 14, a superimposed boundary detection unit 15, and a superimposed boundary correction unit 1 including a unit 16, a pattern memory 17, and a character recognition unit 18.
6 and the pattern memory 17 constitute a character cutout device.

【0019】なお、図5で示した従来型OCR5と同一
の機能ブロックについては同一符号を付してある。ま
た、上記プログラムは、通常、コンピュータ装置の内部
記憶装置あるいは外部記憶装置に格納され、随時読み取
られて実行されるようになっているが、コンピュータ装
置とは分離可能な記録媒体、例えばCD−ROMやFD
等に格納され、使用時にコンピュータ装置に読み取ら
れ、上記内部記憶装置または外部記憶装置にインストー
ルされて、随時実行に供されるものであってもよい。
The same functional blocks as those of the conventional OCR 5 shown in FIG. 5 are denoted by the same reference numerals. The program is usually stored in an internal storage device or an external storage device of the computer device, and is read and executed as needed. However, a recording medium separable from the computer device, for example, a CD-ROM And FD
Or the like, read by a computer device at the time of use, installed in the internal storage device or the external storage device, and provided for execution at any time.

【0020】外接方形統合部13は、外接方形抽出部1
2からの複数の文字外接方形について、それぞれの始点
座標及び終点座標を付加するとともに、各文字外接方形
及びその座標情報を重畳境界検出部15に入力する。こ
の点、各文字外接方形をパタン抽出部14に入力してい
る従来型OCR5の外接方形統合部13Aと異なる。
The circumscribed rectangle integrating unit 13 is a circumscribed rectangle extracting unit 1
With respect to a plurality of character circumscribed rectangles from No. 2, respective start point coordinates and end point coordinates are added, and each character circumscribed rectangle and its coordinate information are input to the superimposed boundary detection unit 15. This is different from the circumscribed rectangle integration unit 13A of the conventional OCR 5 in which each character circumscribed rectangle is input to the pattern extraction unit 14.

【0021】重畳境界検出部15は、入力された複数の
文字外接方形間の重畳の有無を検出し、個々の文字外接
方形にそれぞれ重畳有無フラグを付加する。重畳が検出
された場合は、当該重畳を回避するように文字外接方形
のサイズ、具体的には始点座標または終点座標を変更す
るとともに重畳有無フラグをオンにする。一方、重畳が
検出されない場合は、重畳有無フラグをオフにする。す
べての文字外接方形について上記各情報を付加した後、
重畳有無フラグがオンになっている文字外接方形の座標
情報及び重畳有無フラグを重畳境界補正部16に入力す
る。重畳有無フラグがすべてオフである場合は、各文字
外接方形の座標情報を、先頭のものから順次、パタン抽
出部14に入力する。
The superimposition boundary detection unit 15 detects the presence or absence of superimposition between a plurality of input character circumscribed rectangles, and adds a superimposition flag to each of the character circumscribed rectangles. When the superimposition is detected, the size of the circumscribed rectangle outside the character, specifically, the start point coordinates or the end point coordinates is changed so as to avoid the superposition, and the superimposition flag is turned on. On the other hand, if no superimposition is detected, the superimposition flag is turned off. After adding the above information for all circumscribed rectangles,
The coordinate information of the circumscribed rectangle with the superimposition flag turned on and the superimposition flag are input to the superimposition boundary correction unit 16. When all the superimposition presence / absence flags are off, the coordinate information of each character circumscribed rectangle is input to the pattern extraction unit 14 sequentially from the top.

【0022】重畳境界補正部16は、重畳境界検出部1
5から入力された文字外接方形の座標情報の補正を行
う。具体的には、重畳有無フラグがオンの文字外接方形
及び隣接する文字外接方形の座標情報に基づいて重畳境
界を設定する。また、隣接の重畳境界の間の領域に存在
するすべての線分外接方形を外接方形抽出部12より取
得し、これらの線分外接方形が各々属すべき文字外接方
形を特定する。文字外接方形が特定できた場合は、該当
する文字外接方形の座標情報に各線分外接方形の座標情
報をさらに付加し、イメージメモリ11内の線分外接方
形の分類を行う。
The superimposition boundary correction section 16 includes a superposition boundary detection section 1
Correction of the coordinate information of the character circumscribed rectangle input from 5 is performed. Specifically, the superimposition boundary is set based on the coordinate information of the character circumscribed rectangle whose superimposition presence / absence flag is on and the adjacent character circumscribed rectangle. In addition, all the line segment circumscribed rectangles existing in the region between the adjacent superimposed boundaries are acquired from the circumscribed rectangle extraction unit 12, and the character circumscribed rectangle to which each of these line segment circumscribed rectangles belongs is specified. If the character circumscribed rectangle can be specified, the coordinate information of each line circumscribed rectangle is further added to the coordinate information of the corresponding character circumscribed rectangle, and the line segment circumscribed rectangle in the image memory 11 is classified.

【0023】ここで、重畳境界検出部15及び重畳境界
補正部16の処理内容をより具体的に説明する。便宜
上、図8に示したものと同様、“東”、“京”、“都”
がこの順に並ぶ3つの文字外接方形を例に挙げて説明す
る。
Here, the processing contents of the superimposed boundary detector 15 and the superimposed boundary corrector 16 will be described more specifically. For convenience, “East”, “Kyoto”, “Tokyo” as in FIG.
Will be described with an example of three character circumscribed rectangles arranged in this order.

【0024】重畳境界検出部15は、まず、先頭の
“東”の文字外接方形と2番目の“京”の文字外接方形
の座標情報に基づいて両者の重畳の有無を検出する。す
なわち、図8に示した“東”の終点座標“X2”と、
“京”の始点座標“X3”とを比較し、両者の差分が予
め設定された重畳検出閾値P1よりも大きい場合は、重
畳があると判定し、“東”及び“京”についての重畳有
無フラグをオンにする。同様に、2番目の“京”の文字
外接方形と3番目の“都”の文字外接方形との座標情報
“X4”、“X5”に基づいて両者の重畳の有無を検出
する。そして、両者の差分が予め設定された重畳検出閾
値P1よりも大きければ“都”についての重畳有無フラ
グをオンにする。
First, the superimposition boundary detecting section 15 detects the presence or absence of superimposition of the two based on the coordinate information of the character circumscribed rectangle of the first "east" and the second character circumscribed rectangle of "Kyo". That is, the end point coordinates “X2” of “East” shown in FIG.
The start point coordinate “X3” of “K” is compared. If the difference between the two is greater than a preset overlap detection threshold value P1, it is determined that there is a overlap, and whether “East” and “K” are overlapped Turn on the flag. Similarly, based on the coordinate information "X4" and "X5" of the second character circumscribed rectangle of "K" and the third character circumscribed rectangle of "To", the presence or absence of superposition of both is detected. If the difference between the two is greater than a preset superimposition detection threshold value P1, the superimposition presence / absence flag for "city" is turned on.

【0025】重畳境界補正部16は、以下のようにして
各文字外接方形の座標情報を補正する。まず、各文字外
接方形に付加された重畳有無フラグがオンになっている
文字外接方形のサイズを隣接の文字外接方形との重畳を
回避し得るサイズに補正する。すなわち、本例では、す
べての文字外接方形の重畳有無フラグがオンになってい
るので、“東”の終点座標“X2”を“京”との重畳を
回避する座標“X21(<X3)”、“京”の始点座標
“X3”を“東”との重畳を回避する座標“X31(>
X2)”、“京”の終点座標“X4”を“都”との重畳
を回避する座標“X41(<X4)”、“都”の始点座
標“X5”を“京”との重畳を回避する座標“X51
(>X5)”にそれぞれ補正する。
The superimposition boundary correction section 16 corrects the coordinate information of each character circumscribed rectangle as follows. First, the size of the character circumscribed rectangle in which the superimposition flag added to each character circumscribed rectangle is on is corrected to a size that can avoid superimposition with an adjacent character circumscribed rectangle. That is, in this example, since the superimposition presence / absence flags of all the character circumscribed rectangles are turned on, the coordinates of the end point “X2” of “East” are set to coordinates “X21 (<X3)” for avoiding superimposition with “K”. , The starting point coordinate “X3” of “K” to the coordinate “X31 (>
X2)), the coordinates "X41 (<X4)" for avoiding the superimposition of "K4" with the end point coordinate "X4" of "Kyo" and the superimposition of the coordinates "X5" of the start point of "Kyoto" with "Kyo" are avoided. Coordinates “X51
(> X5) ".

【0026】これらの座標情報の補正は、予め定めた選
択マージン値を用いて行う。図3は、この補正原理を示
す模式図であり、2つの文字外接方形“東”及び“京”
の場合の例を示すものである。 図3において、“X
L”及び“XR”は、それぞれ“東”及び“京”の文字
外接方形の重畳境界の座標である。いま、選択マージン
値を切り出し対象となる文書の種類に応じて定まる値P
2とすると、“東”についての重畳境界の座標XLは
「X2−P2」、“京”についての重畳境界の座標XR
は「X3+P2」で求めることができる。この処理を
“京”及び“都”についても同様に行うことで、各座標
情報を補正する。
The correction of the coordinate information is performed using a predetermined selection margin value. FIG. 3 is a schematic diagram showing this correction principle, in which two character circumscribed rectangles “East” and “K”
FIG. In FIG. 3, "X
“L” and “XR” are the coordinates of the superimposed boundary of the circumscribed rectangles of the characters “East” and “K.” The selection margin value is a value P determined according to the type of the document to be cut out.
2, the coordinate XL of the superimposed boundary for “East” is “X2-P2”, and the coordinate XR of the superimposed boundary for “K” is
Can be obtained by “X3 + P2”. This process is similarly performed for “K” and “To” to correct each coordinate information.

【0027】重畳境界補正部16は、また、重畳境界間
の領域(重複境界領域)に存在するすべての線分外接方
形を外接方形抽出部12より取得し、これらの線分外接
方形が各々属すべき文字外接方形を特定する。図3の例
では、“東”と“京”の間の重複境界領域に2つの線分
外接方形が存在するが、図示上方の線分外接方形の始点
座標は重畳境界XLと一致するので、これを“東”の文
字外接方形に属するものとして分類する。そして、この
線分外接方形の始点座標(Xa,Ya)と終点座標(X
b,Yb)を“東”の文字外接方形の座標情報に付加す
る。また、図示下方の線分外接方形は、始点座標が重畳
境界XLには一致しないが、終点座標が他方の重畳境界
XRと一致するので、“京”の文字外接方形に属するも
のとして分類する。そして、この線分外接方形の始点座
標(Xc,Yc)と終点座標(Xd,Yd)を“京”の
文字外接方形の座標情報に付加する。この処理を“京”
と“都”の間の重複境界領域についても同様に行う。
The superimposed boundary correction section 16 also obtains from the circumscribed rectangle extracting section 12 all the line segment circumscribed rectangles existing in the area between the superimposed boundaries (overlap boundary area), and each of these line segment circumscribed rectangles belongs. Specifies the character circumscribed rectangle. In the example of FIG. 3, two line segment circumscribed rectangles exist in the overlapping boundary region between “East” and “K”, but since the starting point coordinates of the line segment circumscribed rectangle in the upper part of the drawing coincide with the superimposed boundary XL, This is classified as belonging to the character circumscribed rectangle of "East". Then, the start point coordinates (Xa, Ya) and the end point coordinates (X
b, Yb) is added to the coordinate information of the circumscribed rectangle of the character “East”. Further, the line segment circumscribed rectangle shown in the lower part of the figure is classified as belonging to the character circumscribed rectangle of "K" because the start point coordinates do not coincide with the superimposed boundary XL, but the end point coordinates coincide with the other superimposed boundary XR. Then, the start point coordinates (Xc, Yc) and the end point coordinates (Xd, Yd) of the line segment circumscribed rectangle are added to the coordinate information of the character circumscribed rectangle of “K”. This process is called “Kyo”
The same applies to the overlapping boundary area between and "city".

【0028】このようにして各文字外接方形に付加され
る各種情報の例を図4に示す。図4中、“f1”、“f
2”、“f3”は重畳有無フラグである。また、重畳境
界領域に存在する線分外接方形(1)、(2)、
(3)、(4)に対応する座標情報が、それぞれ(X
a,Ya,Xb,Yb)、(Xc,Yc,Xd,Y
d)、(Xe,Ye,Xf,Yf)、(Xg,Yg,X
h,Yh)である。パタン抽出部14は、イメージメモ
リ11上の帳票イメージから、各座標情報に対応する領
域の文字パタンを抽出してパタンメモリ17に格納す
る。文字認識部18は、パタンメモリ17内の文字パタ
ンの文字認識を行い、認識結果を図示しない出力手段に
対して出力する。
FIG. 4 shows an example of various types of information added to each character circumscribed rectangle. In FIG. 4, "f1", "f
2 ”and“ f3 ”are superimposition presence / absence flags, and the line segment circumscribed rectangles (1), (2), and
The coordinate information corresponding to (3) and (4) is (X
a, Ya, Xb, Yb), (Xc, Yc, Xd, Y
d), (Xe, Ye, Xf, Yf), (Xg, Yg, X
h, Yh). The pattern extraction unit 14 extracts a character pattern in an area corresponding to each piece of coordinate information from the form image on the image memory 11 and stores the character pattern in the pattern memory 17. The character recognition unit 18 performs character recognition of the character pattern in the pattern memory 17 and outputs a recognition result to an output unit (not shown).

【0029】このように、本実施形態のOCR1では、
文字外接方形相互間において文字外接方形の重畳が検出
された場合に、該当する文字外接方形に対して線分外接
方形の座標情報を付加してパタン抽出部14に出力する
ので、パタン抽出部14では、侵入した線分を除外した
最適な文字パタンの切り出し領域を直ちに特定できるよ
うになる。従って、従来型OCR5のような文字線境界
のトレース及び白点化の処理が不要になり、文字間の重
畳が発生しやすい帳票等であっても、迅速な文字パタン
の切り出しが可能になる。
As described above, in the OCR 1 of the present embodiment,
When the superposition of the character circumscribed rectangle is detected between the character circumscribed rectangles, the coordinate information of the line segment circumscribed rectangle is added to the corresponding character circumscribed rectangle and output to the pattern extraction unit 14, so that the pattern extraction unit 14 Thus, an optimal character pattern cut-out area excluding the invading line segment can be immediately specified. Therefore, the process of tracing and whitening the character line boundaries as in the conventional OCR 5 is not required, and even a form or the like in which superimposition between characters is likely to occur can quickly extract a character pattern.

【0030】また、文字外接方形の重畳が検出されない
場合には、上記補正処理を行うことなく即座に文字認識
に移行できるため、読み取り処理全体の高速化が可能と
なり、OCRにおける実用性が大幅に向上する。
Further, when the superposition of the character circumscribed rectangle is not detected, the character recognition can be immediately started without performing the above-mentioned correction processing, so that the entire reading processing can be speeded up and the practicality in OCR is greatly improved. improves.

【0031】なお、本実施形態の説明では、帳票イメー
ジが横書きされた文字であることを想定しているが、本
発明は、上記実施形態に限定されることなく、例えば、
縦書きされた文字に対しても同様に適合可能である。こ
の場合には、X座標軸に着目した上述の重畳境界等を、
Y座標軸に関して適宜設定することにより容易に実現す
ることができる。
In the description of the present embodiment, it is assumed that the form image is a horizontally written character. However, the present invention is not limited to the above-described embodiment.
The same applies to vertically written characters. In this case, the above-described superimposition boundary or the like focusing on the X coordinate axis is expressed by:
It can be easily realized by appropriately setting the Y coordinate axis.

【0032】[0032]

【発明の効果】以上の説明から明らかなように、本発明
の文字切り出し装置によれば、文字認識に用いる文字パ
タンを正確且つ迅速に特定できるようになる。また、本
発明の光学式文字読取装置によれば、文字認識精度を一
定値以上に維持しながら認識処理に要する時間を格段に
短縮できるようになる。
As is clear from the above description, according to the character extracting apparatus of the present invention, the character pattern used for character recognition can be specified accurately and quickly. Further, according to the optical character reading device of the present invention, the time required for the recognition process can be significantly reduced while maintaining the character recognition accuracy at a certain value or more.

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

【図1】本発明の一実施形態に係る光学式文字読取装置
の機能ブロック図。
FIG. 1 is a functional block diagram of an optical character reading device according to an embodiment of the present invention.

【図2】本実施形態により個々の文字外接方形に付加さ
れる座標情報の説明図。
FIG. 2 is an explanatory diagram of coordinate information added to each character circumscribed rectangle according to the embodiment;

【図3】重畳境界と、重畳境界領域に存在する余分な線
分外接方形を表す図。
FIG. 3 is a diagram illustrating a superimposed boundary and an extra line segment circumscribed rectangle existing in the superimposed boundary region.

【図4】本実施形態の処理によって各文字外接方形に付
加される座標情報の説明図。
FIG. 4 is an explanatory diagram of coordinate information added to each character circumscribed rectangle by the processing of the embodiment.

【図5】従来型の光学式文字読取装置の機能ブロック
図。
FIG. 5 is a functional block diagram of a conventional optical character reader.

【図6】帳票イメージの例を示す説明図。FIG. 6 is an explanatory diagram showing an example of a form image.

【図7】(a)は線分外接方形の統合の基準、(b)は
統合後の一次外接方形を表す図。
FIGS. 7A and 7B are diagrams illustrating a criterion for integrating line circumscribed rectangles, and FIG. 7B illustrates a primary circumscribed rectangle after integration.

【図8】文字外接方形に付加される座標情報の説明図。FIG. 8 is an explanatory diagram of coordinate information added to a character circumscribed rectangle;

【図9】(a)、(b)、(c)は、それぞれ文字外接
方形により切り出された文字パタンの例。
FIGS. 9A, 9B, and 9C are examples of character patterns cut out by circumscribed rectangles;

【図10】(a)は従来の侵入線分の除去処理過程を示
す説明図、(b)は侵入線分が除去された文字外接方形
の説明図。
FIG. 10A is an explanatory diagram showing a conventional process of removing an intruding line, and FIG. 10B is an explanatory diagram of a character circumscribed rectangle from which an intruding line is removed.

【符号の説明】[Explanation of symbols]

1,5 光学式文字読取装置(OCR) 11 イメージメモリ 12 外接方形抽出部 13,13A 外接方形統合部 14 パタン抽出部 15 重畳境界検出部 16 重畳境界補正部 17 パタンメモリ 18 文字認識部 1,5 Optical character reader (OCR) 11 Image memory 12 Bounding rectangle extracting unit 13,13A Bounding rectangle integrating unit 14 Pattern extracting unit 15 Superimposed boundary detecting unit 16 Superimposing boundary correcting unit 17 Pattern memory 18 Character recognizing unit

Claims (5)

【特許請求の範囲】[Claims] 【請求項1】 複数の文字パタンを含む画像イメージか
ら線分外接方形を抽出し、抽出した線分外接方形を第1
方向に統合するとともに、統合後の線分外接方形を第1
方向と垂直の第2方向に分割して複数の文字外接方形を
作成する文字外接方形作成手段と、 前記第2方向に隣接する一対の文字外接方形の一方の終
点座標と他方の始点座標との差分が所定の重畳検出閾値
以上かどうかを判定し、判定結果に応じたサイズの文字
切り出し領域を指定する重畳検出手段と、 前記重畳検出手段の出力情報に基づいて前記画像イメー
ジから該当する文字パタンを切り出すパタン抽出手段
と、 を備えてなる文字切り出し装置。
1. A line segment circumscribed rectangle is extracted from an image including a plurality of character patterns, and the extracted line segment circumscribed rectangle is defined as a first line segment circumscribed rectangle.
Direction, and the circumscribed rectangle of the line
A character circumscribed rectangle creating means for dividing a plurality of character circumscribed rectangles in a second direction perpendicular to the direction, and a coordinate of one end point and a start point coordinate of the other of the pair of character circumscribed rectangles adjacent in the second direction. Superimposition detection means for determining whether the difference is equal to or greater than a predetermined superimposition detection threshold, and designating a character cut-out area of a size corresponding to the determination result; and a corresponding character pattern from the image image based on output information of the superimposition detection means. And a character extracting device for extracting a character.
【請求項2】 前記重畳検出手段は、前記差分が前記重
畳検出閾値に満たない場合は個々の文字外接方形の座標
情報を前記文字切り出し領域として指定し、一方、前記
差分が前記重畳検出閾値以上の場合は、各文字外接方形
を他の文字外接方形との重畳を回避するサイズに変更す
るとともに前記差分の領域に存する線分外接方形の座標
情報とその線分外接方形が属すべき文字外接方形の座標
情報とを前記文字切り出し領域として指定することを特
徴とする請求項1記載の文字切り出し装置。
2. The superimposition detecting means, when the difference is less than the superimposition detection threshold, designates coordinate information of an individual character circumscribed rectangle as the character cutout area, while the difference is greater than or equal to the superimposition detection threshold. In the case of, each character circumscribed rectangle is changed to a size that avoids overlapping with another character circumscribed rectangle, and coordinate information of the line circumscribed rectangle existing in the difference area and the character circumscribed rectangle to which the line circumscribed rectangle belongs 2. The character segmenting device according to claim 1, wherein the coordinate information is designated as the character segmenting region.
【請求項3】 前記差分の領域が、所定のマージン値に
基づいて自動設定されるように構成されることを特徴と
する請求項2記載の文字切り出し装置。
3. The character clipping device according to claim 2, wherein the difference area is automatically set based on a predetermined margin value.
【請求項4】 光学的走査手段に読み取られた複数の文
字成分を含む画像イメージを格納するイメージメモリ
と、 前記イメージメモリに格納された画像イメージから複数
の線分外接方形を抽出し、抽出された複数の線分外接方
形を第1方向に統合するとともに統合後の線分外接方形
を第1方向と垂直の第2方向に分割して複数の文字外接
方形を作成する文字外接方形作成手段と、 前記第2方向に隣接する一対の文字外接方形の一方の終
点座標と他方の始点座標との差分が所定の重畳検出閾値
以上かどうかを判定し、判定結果に応じたサイズの文字
切り出し領域を指定する重畳検出手段と、 前記重畳検出手段の出力情報に基づいて前記画像イメー
ジから該当する文字パタンを切り出すパタン抽出手段
と、 切り出された文字パタンに基づいて文字認識を行う文字
認識手段と、を備えてなる光学式文字読取装置。
4. An image memory for storing an image including a plurality of character components read by an optical scanning means, and extracting a plurality of line segment circumscribed rectangles from the image stored in the image memory. A character circumscribed rectangle creating means for integrating the plurality of line segment circumscribed rectangles in a first direction and dividing the combined line segment circumscribed rectangle in a second direction perpendicular to the first direction to create a plurality of character circumscribed rectangles; It is determined whether a difference between one end point coordinate of the pair of character circumscribed rectangles adjacent in the second direction and the other start point coordinate is equal to or greater than a predetermined superimposition detection threshold value, and a character cutout area having a size according to the determination result is determined. Superimposition detection means to be designated; pattern extraction means for extracting a corresponding character pattern from the image based on output information of the superimposition detection means; and a sentence based on the extracted character pattern. Optical character reading apparatus comprising and a character recognition means for recognizing.
【請求項5】 複数の文字パタンを含む画像イメージか
ら線分外接方形を抽出する処理、 抽出した線分外接方形を第1方向に統合するとともに、
統合後の線分外接方形を第1方向と垂直の第2方向に分
割して複数の文字外接方形を作成する処理、 前記第2方向に隣接する一対の文字外接方形の一方の終
点座標と他方の始点座標との差分が所定の重畳検出閾値
以上かどうかを判定する処理、 前記差分が前記重畳検出閾値に満たない場合は個々の文
字外接方形の座標情報を文字切り出し領域として指定
し、一方、前記差分が前記重畳検出閾値以上の場合は、
各文字外接方形を他の文字外接方形との重畳を回避する
サイズに変更するとともに前記差分の領域に存する線分
外接方形の座標情報とその線分外接方形が属すべき文字
外接方形の座標情報とを文字切り出し領域として指定す
る処理、 前記指定された座標情報に基づいて前記画像イメージか
ら該当する文字パタンを切り出す処理、 をコンピュータに実行させるプログラムを記録してなる
コンピュータ読み取り可能な記録媒体。
5. A process for extracting a line segment circumscribed rectangle from an image including a plurality of character patterns, integrating the extracted line segment circumscribed rectangle in a first direction,
A process of dividing the integrated line segment circumscribed rectangle in a second direction perpendicular to the first direction to create a plurality of character circumscribed rectangles; one end coordinate of a pair of character circumscribed rectangles adjacent in the second direction and the other The process of determining whether the difference from the start point coordinates is equal to or greater than a predetermined superimposition detection threshold.If the difference is less than the superimposition detection threshold, the coordinate information of each character circumscribed rectangle is specified as a character cutout area. If the difference is equal to or greater than the superimposition detection threshold,
The size of each character circumscribed rectangle is changed to a size that avoids overlapping with other character circumscribed rectangles, and the coordinate information of the line circumscribed rectangle existing in the difference area and the coordinate information of the character circumscribed rectangle to which the line circumscribed rectangle should belong A computer-readable recording medium storing a program for causing a computer to execute a process of designating a character extraction region as a character extraction region, and a process of extracting a corresponding character pattern from the image image based on the designated coordinate information.
JP9210480A 1997-08-05 1997-08-05 Character segmentation device, optical character reader and storage medium Pending JPH1153469A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP9210480A JPH1153469A (en) 1997-08-05 1997-08-05 Character segmentation device, optical character reader and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
JP9210480A JPH1153469A (en) 1997-08-05 1997-08-05 Character segmentation device, optical character reader and storage medium

Publications (1)

Publication Number Publication Date
JPH1153469A true JPH1153469A (en) 1999-02-26

Family

ID=16590051

Family Applications (1)

Application Number Title Priority Date Filing Date
JP9210480A Pending JPH1153469A (en) 1997-08-05 1997-08-05 Character segmentation device, optical character reader and storage medium

Country Status (1)

Country Link
JP (1) JPH1153469A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2021005315A (en) * 2019-06-27 2021-01-14 キヤノン株式会社 Information processing device, program, and control method

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2021005315A (en) * 2019-06-27 2021-01-14 キヤノン株式会社 Information processing device, program, and control method

Similar Documents

Publication Publication Date Title
JPH05233873A (en) Area dividing method
JP3574584B2 (en) Front image processing apparatus and its program storage medium
JP2006338578A (en) Character recognition apparatus
JPH1153469A (en) Character segmentation device, optical character reader and storage medium
JPH06208625A (en) Method and device for processing image
JP3348224B2 (en) Table frame line intersection correction device, table recognition device, and optical character reading device
JPH07230525A (en) Method for recognizing ruled line and method for processing table
JP3019897B2 (en) Line segmentation method
JPH08263588A (en) Character recognition device
JPH07160810A (en) Character recognizing device
JP4810995B2 (en) Image processing apparatus, method, and program
JP3052438B2 (en) Table recognition device
JPH117493A (en) Character recognition processor
JP2003016385A (en) Image processor, method, program and storage medium
JP3133797B2 (en) Character recognition method and apparatus
JP4040231B2 (en) Character extraction method and apparatus, and storage medium
JP2004158041A (en) Surface image processor and its program storage medium
JP2004152048A (en) Vehicle number reading device
JP3517077B2 (en) Pattern extraction device and method for extracting pattern area
JPH11242716A (en) Image processing method and storage medium
JP2001250084A (en) Method and device for processing image and computer- readable recording medium with program for realizing the method recorded thereon
JPH06274690A (en) Character recognizing device and optical character reader
JPH05128305A (en) Area dividing method
JP3566738B2 (en) Shaded area processing method and shaded area processing apparatus
JPH07168911A (en) Document recognition device