JPS60146376A - Device for segmenting character - Google Patents

Device for segmenting character

Info

Publication number
JPS60146376A
JPS60146376A JP59002910A JP291084A JPS60146376A JP S60146376 A JPS60146376 A JP S60146376A JP 59002910 A JP59002910 A JP 59002910A JP 291084 A JP291084 A JP 291084A JP S60146376 A JPS60146376 A JP S60146376A
Authority
JP
Japan
Prior art keywords
character
cutting
pitch
projection
start position
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
JP59002910A
Other languages
Japanese (ja)
Inventor
Hiroyuki Kami
上 博行
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.)
NEC Corp
Original Assignee
NEC Corp
Nippon Electric Co Ltd
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 NEC Corp, Nippon Electric Co Ltd filed Critical NEC Corp
Priority to JP59002910A priority Critical patent/JPS60146376A/en
Publication of JPS60146376A publication Critical patent/JPS60146376A/en
Pending legal-status Critical Current

Links

Landscapes

  • Character Input (AREA)

Abstract

PURPOSE:To obtain a character segmenting device which can prevent erroneous segmenting of characters by detecting an optimum segmenting position around the position, which is determined by a character pitch or estimated character pitch to chop the character. CONSTITUTION:An estimation processing part 15 obtains the right end of a character at the right end or the position in correspondence to the left end of a character at the left side from a projection signal of one character string set in a projection signal memory 13 by detecting the position which first goes to ''0''. By utilizing the character pitch stored in a character pitch memory, binary signals in the projection signal memory 13 are overlapped from the position one character pitch at the right or left end away from a character at the right or left end, that is, folding by width of a character pitch is executed, and frequency distribution is generated. The estimation processing part 15 moves said start position at the place having the most blank lines from the obtained frequency distribution, that is, by length to a characteristic position of the frequency distribution, and transmits it to a segmenting start position memory 16.

Description

【発明の詳細な説明】 不発明は文字切出し装置に関し、特に等ピッチで印字又
は書かれた文字行からの各文字を切出す文字切出し装置
に関する。従来の文字の切出し装置は、二種類の方法が
知られている・ −7一つ方法は、ラインマーク、又は
文字印字あるいは記入の枠から一足長ごとに切出す方法
である。
DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a character cutting device, and more particularly to a character cutting device that cuts out each character from a line of characters printed or written at equal pitches. Two types of conventional character cutting devices are known: -7 One method is to cut out each foot length from a line mark or character printing or writing frame.

この方法は非常に簡単な方法ではあるが、印字がずれた
シ、はみ出して記入されたシすると、切出し誤シを起す
、また、この問題を解決するための方法としては一足長
ごとで決まった切出し位置の周囲で最適な位置を検出す
るように改良した方法もあるが、どちらも文字切出し開
始位置が既知あるいは正確に検出する必要があった。
Although this method is very simple, if the printed characters are misaligned or the characters are written outside, it will result in cutting errors. There are improved methods that detect the optimal position around the cutting position, but both require that the character cutting start position be known or accurately detected.

もう一つの方法は一行の2値の行イメージを投影して得
られる投影信号よシ5文字に対応する連続した@lHの
長さが与えられる文字ピッチと比較して文字切出し?行
う方法である。この方法では、文字切出し開始位置は不
要で、分離文字全切出すことは可能であるが1文字が接
触していると文字に対応する連続したl”の長さは文字
ピッチよシ長くなるため1文字ピッチより文字数を予測
し等しい間隔で切出している。しかしこの方法は谷文字
が咎しい幅ではないので、切出し誤シを起し、また分離
文字と幅の狭い文iが隣合うときに切出し誤りを起し易
かった。この切出し誤シを起す原因は、文字切出し開始
位置が不明なために部分的に得られる1w報で切出し位
置を次足しなければならないことによる。
Another method is to cut out characters by comparing the projection signal obtained by projecting a binary line image of one line with the character pitch given by the length of continuous @lH corresponding to 5 characters? This is the way to do it. With this method, a character extraction start position is not required, and it is possible to extract all separated characters, but if one character is touching, the length of consecutive l'' corresponding to the characters will be longer than the character pitch. The number of characters is predicted from the pitch of one character and the characters are cut out at equal intervals. However, this method causes cutting errors because the width of the valley characters is not bad, and also when a separated character and a narrow sentence i are adjacent to each other, It is easy to make a cutting error.The cause of this cutting error is that the character cutting start position is unknown and the cutting position must be added using the partially obtained 1W information.

しかじ槙々の文字幅の文字が等ピッチで記入又は印字さ
れていると、部分的な情報では困難でめるが、全体から
は文字切出し開始位置を予測することは容易である。た
とえば文字切出し開始位置を仮定して第1図の文字行を
等ピッチ(幻で切シ。
When characters with a wide width are written or printed at equal pitches, it is difficult to predict the starting position of character cutting from the entire information, although it is difficult to predict from partial information. For example, assuming the character cutting start position, the character lines in Figure 1 are cut at equal pitches (phantom cutting).

得られる211Nイメージを重ね合せて投影すると、谷
文字切出し開始位置に対応して第2図に示すような2値
1ば号″′1”の濃度ヒストグラムが得られる0文字切
出し開始位置が正しければピッチの両端位置では文字部
がないので、濃度ヒストグラムの両端は濃度が少なくな
るはずである。従って第2図(b)の切出し開始位置か
ら文字切出しを行えば切出し誤シを防ぐことが出来る。
When the obtained 211N images are superimposed and projected, a density histogram of binary 1-bar ``'1'' as shown in Figure 2 is obtained corresponding to the valley character extraction start position.If the 0 character extraction start position is correct, Since there is no character portion at both ends of the pitch, the density should be low at both ends of the density histogram. Therefore, if characters are cut out from the cutting start position shown in FIG. 2(b), cutting errors can be prevented.

また得られた濃度ヒストグラムは同一の分布でアリ、開
始位置が異っただけである。@3図に例示するような接
触文字や投影で皿なる文字があっても、その数が少なけ
れは他の文字の位置で切出し開始位置が決する自 ここで2愼イメージを重ね合せた投影は2値イメージの
文字行と@父する方間に投影して各位置の投影値全記憶
し1等間隔の位置における記憶された投影値’i 7J
O算することで得られる。また文字の接触や重なシは第
3図に例、ボするように上端。
Furthermore, the obtained density histograms have the same distribution, only the starting position is different. @3 Even if there are touching characters or characters that become plates due to projection, as shown in figure 3, if the number of characters is small, the cutting start position will be determined by the position of other characters.Here, the projection of two images superimposed is 2 Project it between the character line of the value image and the front side, store all the projection values at each position, and store the stored projection values at positions at equal intervals.'i 7J
It can be obtained by calculating O. Also, as shown in Figure 3, if the letters touch or overlap, the top edge should be blurred.

下端部で生じ易いのでめようとする投影値に悪影響を与
える。従って上端、又は下端、又は両端を除いて投影を
めると文字のセリフの部分の影#全受けにくくなる。さ
らに実際には完全に等ピ、チで印字又は記入されること
はまれであるので。
Since it tends to occur at the lower end, it has a negative effect on the projected value. Therefore, if the projection is reduced except for the upper end, the lower end, or both ends, it becomes difficult to receive the entire shadow of the serif portion of the character. Furthermore, in reality, it is rare that the information is printed or filled out completely in equal pitch and square.

文字切出し開始位置と等しい文字ピッチとで決まる位置
の周囲で最適な位置を検出し切出しを行うことにより正
確に切出せる。たとえば第4図り1の位置が文字切出し
開始位置と等しい文字ピッチとでまった位置であるとす
ると、1の位置の周囲の2の位置の方が文字パターンを
切断しないので切出し位置とじては望ましい、また前述
の文字ピッチはタイプライタ−のようにあらかじめわか
る場合と等ピッチであるという情報音もとに2値行イメ
ージで投影1百号よp文字ピッチを推定出来る場合とが
ある。
Accurate cutting can be performed by detecting the optimum position around the position determined by the character cutting start position and the same character pitch and performing cutting. For example, if the position of the fourth pattern 1 is a position where the character pitch is equal to the character cutting start position, position 2 around position 1 is preferable as the cutting position because it does not cut the character pattern. In addition, there are cases in which the above-mentioned character pitch is known in advance, such as with a typewriter, and cases in which the pitch of the p character can be estimated from the projection number 100 using a binary line image based on the information sound that the pitch is equal.

不発明の目的は従来のかかる欠点全除去すると共に、2
値の行イメージを投影して得られる投影1ぎ号全体又は
投影の際に抽出される特徴信号全体から文字切出し開始
位置を検出し請求まった開始位置よシ与えられた文字ピ
ッチで、又は推定文字ピッチで順次切出す、又はまった
開始位置と与えられた文字ピッチ又は推定文字ピッチで
決まる位置の周囲で最適な切出し位置の検出を行い切出
すことによp文字の切出しfJ4Dを防ぐ文字切出し装
置を提供することにある。
The purpose of the invention is to eliminate all such drawbacks of the conventional technology, and to
The character extraction start position is detected from the entire projection obtained by projecting the row image of the value or the entire feature signal extracted during projection, and the character extraction start position is determined from the requested start position at a given character pitch or estimated. Character cutting that prevents the p character from being cut out fJ4D by sequentially cutting out the character pitch, or by detecting and cutting the optimal cutting position around the position determined by the full starting position and the given character pitch or estimated character pitch. The goal is to provide equipment.

不発明によれは、2値の行イメージを投影して得られる
投影信号を用いて文字切出し位置ヲ求める文字切出し装
置において、前記投影16号を文字列の端から離れた位
置よシ文字ピッチの幅でたたみこみ全行い得られる頻度
分布の特徴位置と、前記文字列の端位置とから切出し開
始位置を永める手段と前記切出し開始位置と文字ピッチ
を用いて個々の文字切出し位i’t−決定する手段とを
具備することを%倣とする文字切出し装置が得られる。
According to another aspect of the present invention, in a character cutting device that determines a character cutting position using a projection signal obtained by projecting a binary line image, the projection No. 16 is moved to a position away from the end of a character string and the character pitch is A means for lengthening the cutting start position from the characteristic position of the frequency distribution obtained by convolving with the width and the end position of the character string, and the cutting start position and character pitch are used to determine the individual character cutting position i't- A character cutting device is obtained, which is characterized in that it is equipped with a determining means.

更に不発明によれば、2値り行イメージを投影して得ら
れる投影信号を用いて文字切出し位置C求める文字切出
し表置において、前記投影信号を文字列の端から陥れた
位置より文字ピッチの幅でたたみこみ全行い侍られるM
[分布の特徴位置と、前記文字列の端からの位置とから
切出し開始位置をめる手段と、前記切出し開始位置の周
囲の谷位置より前記文字ピッチ長を増減してする谷幅で
前記投影信号をたたみこみ、得られる頻度分布のうちで
特徴装置の頻度が最小の、頻度分布の特徴位置から修正
切出し開始位置記憶部する手段と、対応する幅を分字ピ
ッチ長とし、4@られた前記切出し開始装置と前記文字
ピッチを用いて個々の文字切出し位置を決定する手段と
t具備したことを特徴とする文字切出し装置が得られ′
る。
Furthermore, according to the invention, in a character cutting table in which a character cutting position C is determined using a projection signal obtained by projecting a binary line image, the character pitch is determined from a position where the projection signal is sunk from the end of the character string. M that is folded in width and served all the way
[Means for determining the cutout start position from the characteristic position of the distribution and the position from the end of the character string; and the projection with the valley width by increasing or decreasing the character pitch length from the valley position around the cutout start position; Means for convolving the signal and storing a corrected cutting start position from the feature position of the frequency distribution where the frequency of the feature device is the minimum among the frequency distributions obtained, There is obtained a character cutting device characterized in that it is equipped with a cutting start device and means for determining individual character cutting positions using the character pitch.
Ru.

次に本発明の実施例について図面全参照して詳訓に説明
する。
Next, embodiments of the present invention will be explained in detail with reference to all the drawings.

第5図は不発明の第1の実施例をボ丁・第5図(a)に
おいて、第1の実施例は帳祭上の一部分の文字列に対応
する2値信号を記憶する行イメージ記憶部1工と、該行
イメージ記f、!郡11に接続され、一部分の文字列の
2値信号を走置してブランクラインおよび文字ラインを
検出し、投影1g号を得る投影処11.bμm2と、該
投影処理部12に接続され、前記投影信号を記憶する投
影信号記憶部13と、予め文字ピッチ葡記憶する文字ピ
ッチ記憶部14と、前記文字ピッチのもとにたたみこみ
全行ない。
FIG. 5 shows a first embodiment of the present invention. In FIG. 5(a), the first embodiment is a line image memory that stores a binary signal corresponding to a part of a character string on a bookmark. Part 1 engineering and the image of the line f,! A projection processor 11. connected to the group 11, detects blank lines and character lines by scanning a binary signal of a partial character string, and obtains a projection number 1g. bμm2, a projection signal storage section 13 which is connected to the projection processing section 12 and stores the projection signal, and a character pitch storage section 14 which stores the character pitch information in advance, and is convolved under the character pitch.

前記投影信号を評価し、切出し位置をめる計測処理部1
5と、この評価にもとすき文字列の切出し開始位置を記
憶する切出し開始位置記憶部16と、前記行イメージ記
憶部11、前記投影処理部12および前記切出し開始位
置記憶部16に接続され、これらの1d号にもとずき文
字母の切出し位置を決定する切出し処理s17とを含む
一゛ まず、第1の実施例においては帳票上の文字列し
第6図(a))t−走査して得られた一部分の文字列の
2値信号金行イメージ記憶部11にセットする。
Measurement processing unit 1 that evaluates the projection signal and determines the cutting position
5, a cutout start position storage unit 16 that stores the cutout start position of the favorite character string for this evaluation, and is connected to the line image storage unit 11, the projection processing unit 12, and the cutout start position storage unit 16, The first embodiment includes a cutting process s17 for determining the cutting position of the character base based on these No. A binary signal of a part of the character string obtained by doing so is set in the gold row image storage section 11.

投影処理部12は行イメージ記憶部11にセットされて
いる2値信号金行方同と直交する方向に走査して、走査
したラインがブランクラインであれは′1”、文字ライ
ンであれば′0”とする2値信号である投影信号(第6
図(b))を出力し、投影信号記憶s13に送る。投影
ii1号記憶都13にはたとえば第6図(a)に示すご
とき行イメージとすると、第6図(blに示すように2
恒信号の投影信号がセットされる0文字ピッチ記憶部1
4にはあらかじめ決った文字ピッチたとえば1インチ当
シlO文字のときは0.1インチとスキャナの分解能と
でまる文字ピッチ長に相当するビクセル数が記憶されて
いる。評価処理部15では投影信号記憶部13にセット
されている一行の文字列の投影信号から右端の文字の右
端又は左端の文字のん端位置に対応する位置を最初にO
″になる位置を検出することによりめ、次に文字ピッチ
記憶部14に記憶されている文字ピッチを用いて、前述
の文字の右端又は左端から右側又は左側に1文字ピッチ
離れた開始位置から、投影信号記憶部13の2値・1ぽ
号を息ね合せて、すなわち、文字ピッチの幅でたfc−
’y−こみを行ない頻度分布の作成全行う。更にこの評
価処理部15は得られた頻度分布よシブランクラインの
最も多い位置丁なわち、頻度分布の特徴位O1までの長
さDで前述の開始位置を移動させて切出し開始位置とし
、切出し開始位置記憶部16に送る。第1(1)実庭例
においては第6図(b)の投影信号のへ〇位Ifを開始
位置とし、第7図のごとき頻度分布とすると、前述の説
明によシ%A位置からDv)長さシフトしたB位置全切
出し開始位置とし、切出し開始位置記憶部16に記憶す
る。
The projection processing unit 12 scans in a direction perpendicular to the binary signal gold direction set in the line image storage unit 11, and if the scanned line is a blank line, it is '1', and if it is a character line, it is '0'. The projection signal (6th
(b)) is output and sent to the projection signal storage s13. For example, if the projection II 1 memory capital 13 has a line image as shown in FIG. 6(a), 2 as shown in FIG.
0 character pitch storage unit 1 in which the projection signal of the constant signal is set
4 stores the number of pixels corresponding to a predetermined character pitch, for example, 0.1 inch in the case of a 1-inch character and the character pitch length calculated by the resolution of the scanner. The evaluation processing unit 15 first calculates the position corresponding to the right end of the rightmost character or the leftmost character from the projection signal of one line of character string set in the projection signal storage unit 13.
'', and then using the character pitch stored in the character pitch storage unit 14, start from a starting position one character pitch away from the right or left end of the character to the right or left. By matching the binary/1po numbers of the projection signal storage unit 13, that is, fc-
'y-comput and complete the creation of the frequency distribution. Furthermore, this evaluation processing unit 15 moves the above-mentioned starting position by a distance D from the obtained frequency distribution to the position where the blank line is most numerous, that is, the length D to the feature position O1 of the frequency distribution, and sets it as the cutting start position, and performs cutting. The data is sent to the start position storage section 16. In the 1st (1) practical example, if the starting position is position If of the projection signal in Fig. 6(b) and the frequency distribution is as shown in Fig. 7, then Dv ) The length-shifted position B is set as the full cutting start position and is stored in the cutting start position storage section 16.

この切出し開始位置が決まると、切出し処理部17は文
字ピッチ記憶部14の文字ピッチと切出し開始位置記憶
s16の切出し開始位置とで行イメージ記憶部11の2
値の行イメージを等分割し、−文字に対応する切出し位
置をめる・ 第5図fb)は本発明の第1の実施例の変形で、切出し
処理部171を有し、この切出し処理部17′は文字ピ
ッチ記憶部140文字ピッチで切出し開始位置記憶部1
6の切出し開始位置から等間隔で決まる位置に対応する
投影1ざ号記憶部13′の投影信号全チェックし%2値
1d号の1″であるブランクラインであれば切出し位置
とし、文字ラインすなわちブランクラインでなけれはそ
の位置の周囲で最も近いブランクライン位置を、ブラン
クラインがなければ文字ピッチで決まる位it切出し位
置として、−文字に対応する21庫イメージを。
Once this cutting start position is determined, the cutting processing unit 17 uses the character pitch in the character pitch storage unit 14 and the cutting start position in the cutting start position storage s16 to
Divide the line image of the value into equal parts and find the cutting position corresponding to the character. FIG. 17' is a character pitch storage unit 140 character pitch cutting start position storage unit 1
All the projection signals in the projection 1 mark storage unit 13' corresponding to positions determined at regular intervals from the cutout start position of 6 are checked, and if the blank line is 1" of the % binary value 1d, it is set as the cutout position, and the character line, i.e. If it is not a blank line, use the closest blank line position around that position, and if there is no blank line, use it as the cutting position determined by the character pitch, and then create the 21-box image corresponding to the - character.

行イメージ記憶部11の2値のイメージを分離し切出し
位置をめる。
The binary image in the row image storage unit 11 is separated and the cutting position is determined.

即ち、不発明の第lv実施例はあらかじめ与えられる文
字ピッチと、得られた文字行に直交する方間の投影信号
よ請求まる文字の存在範囲の右端又は左端とから位置を
決め、前記位@に開始位置として前記投影信号を、前述
の文字ピッチによる等間隔位置で重ね合せし、すなわち
加算しまる7Jl]算甑が歳小となる位置をめ、前記開
始位置とで切出し開始位置をめ、得られた切出し開始位
置よシ順次前記文字ピッチの間隔ごとに2値の行イメー
ジを切出す文字切出し装置である。
That is, in the uninvented 1vth embodiment, the position is determined from the right or left end of the range of the requested character based on the projection signal between the character pitch given in advance and the direction perpendicular to the obtained character line, and the position is determined by the position @ As a starting position, superimpose the projection signals at equal intervals according to the character pitch described above, that is, add them.7Jl] Find the position where the calculation number becomes the yearly and short, and find the cutting start position with the starting position, This character cutting device sequentially cuts out binary line images at intervals of the character pitch from the obtained cutting start position.

第8図は不発明の第2の実施例を下す、第8図(a)に
おいて%第2の実施例は第1の実施例と同様に帳票上の
一部分の文字列を2値1d号として記憶する行イメージ
記憶部21と、該行イメージ記憶tkis21に接続さ
れ、−行方の文字列の211消号を走査してブランクラ
インおよび文字ラインを検出し、投影信号を得る投影処
理部22と、該投影処理部22に接続され、ブランクラ
インおよび文字ラインとt示す投影・16号を記憶する
投影信号Hピ僧都23と1行イメージ記憶都21に接続
され、投影して得られる21直1g号の頻度分布によシ
文字高さをめる文字高さ検出部28と、投影信号記憶部
23と文字高さ検出部28とに接続され、ブランクライ
ンと文字等會示す文字ラインと金ホす2値11号と文字
高さをボす信号とにより文字ピッチを推定する文字ピッ
チ推足部29と、投影信号記憶部23に接続され、前H
ピ文字ピッチのもとに前記2値信号である投影信号を評
価し、切出し開始位置金求める評価処理部25と、前記
評価処理部25に接続され、投影信号の評価のもとに文
字の切出し開始位置r記憶する切出し開始位置記憶部2
6と、前記行イメージ記1意部211前記切出し開始位
置記憶部26および文字ピッチ抽足部29に接続され、
これらの信号にもとすき文字切出し位置全決定する切出
し処理部27とヲ貧む・なお、行イメージ記憶部21、
投影処理部22、投影信号記憶部23.評価処理部25
および切出し処理@27は第5図で説明し7’C@1の
実施例の機能と同一であるので、これらの4表能につい
ての説明全省略する。
Figure 8 shows the second example of non-invention. In Figure 8 (a), the second example uses a part of the character string on the form as a binary number 1d, similar to the first example. a line image storage unit 21 for storing, a projection processing unit 22 connected to the line image storage tkis21, which scans the 211 erasure of the character string in the -direction to detect blank lines and character lines, and obtains projection signals; It is connected to the projection processing unit 22 and is connected to the projection signal Hpi-Souzu 23 which stores the blank line and the character line and the projection No. 16 which is indicated by t. A character height detection unit 28 that calculates the character height according to the frequency distribution of A character pitch estimator 29 is connected to the projection signal storage unit 23 and a character pitch estimation unit 29 estimates the character pitch based on the binary number 11 and a signal indicating the character height.
An evaluation processing unit 25 that evaluates the projection signal, which is the binary signal, based on the character pitch and calculates the cutting start position; Cutting start position storage unit 2 that stores the start position r
6, the line image memory unit 211 is connected to the cutting start position storage unit 26 and the character pitch extraction unit 29,
These signals also include a cutout processing unit 27 that determines all the character cutout positions, and a line image storage unit 21.
Projection processing section 22, projection signal storage section 23. Evaluation processing section 25
Since the functions of 7'C and 27 are the same as those of the embodiment 7'C@1, a complete explanation of these four functions will be omitted.

文字高さ検出部18は行イメージ記憶部21の2値イメ
ージを行方向に投影して得られる2値信号の頻度分布よ
υ文字高さをめるもので、前述の頻度分布が第9図に示
すとお夛であるとすると。
The character height detection unit 18 calculates the character height from the frequency distribution of the binary signal obtained by projecting the binary image in the line image storage unit 21 in the line direction, and the frequency distribution described above is shown in FIG. Suppose that it is Otani as shown in .

文字の下端はベースラインと言われる頻度が多いa位置
、文字の上端はa位置から連続した頻度の端のb位置で
あり、a位置とb位置の差が文字高さとなり、あるいは
行イメージ記憶部21の2値イメージの文字列の左端と
右端とIJず検出し。
The lower edge of the character is at position a, which is often referred to as the baseline, and the upper edge of the character is at position b, which is the end of the continuous frequency from position a.The difference between position a and position b is the character height, or line image memory. The left and right ends of the character string in the binary image of part 21 are detected.

得られた前記左端と右端の間をいくつかに分割し、分割
された範囲ごとにめた行方向に投影して得られる2直1
11侶号の頻度分布よシ文字筒さ才求め、得られた文字
高さの平均化によりめられた文字向さを文字ピッチ推足
部29に出力する。文字ピッチ推足部29は前記投影1
ぎ@記1意部23にセットされている21直・1d号よ
りブランクライン以外の連続したツインすなわち連結し
た″θ″位置の中心位置をめ、隣合う中心位置間距離の
うらで、文字向さ検出部28から送られてきた文字1%
さから決筐るl1j41111円例えば文字高さの70
−から170チまでにある中心位置間距Pi1の平均値
又は頻度の多い中心位置間距離を文字ピッチと推定する
。このようにして得られた推定文字ピッチのもとに切出
し処理部27は切出し開始位置記憶部26からの切出し
開始位置から文字の切出し位置τ等出する。
The area between the obtained left end and right end is divided into several parts, and each divided range is projected in the row direction.
The character length is determined from the frequency distribution of the number 11, and the character orientation determined by averaging the obtained character heights is output to the character pitch estimator 29. The character pitch foot part 29 is the projection 1
From No. 21 Straight/1d set in the 1st part 23 of the GI@Ki, locate the center position of continuous twin or connected "θ" positions other than the blank line, and align the character direction with the back of the distance between the adjacent center positions. 1% of characters sent from the detection unit 28
For example, the character height is 70
The average value of the distance between center positions Pi1 from - to 170 inches or the frequently occurring distance between center positions is estimated as the character pitch. Based on the estimated character pitch obtained in this manner, the cutout processing unit 27 outputs a character cutout position τ, etc. from the cutout start position from the cutout start position storage unit 26.

第8図(b)は不発明の第2の実施例の変形で、切出し
処理部27′ケ有しておシ、この切出し処理部27′は
前記行イメージ記憶部21、前記切出し開始位置記憶部
26および文字ピッチ推定品29に接続されると共に投
影jd号記憶部23’に接続され、これらの1g号のも
とに同様に文字切出し位置を決定する。
FIG. 8(b) shows a modification of the uninvented second embodiment, which includes a cutout processing section 27', which includes the row image storage section 21, the cutout start position storage section 27', and the cutout processing section 27'. It is connected to the projection jd number storage unit 23' as well as to the character pitch estimation unit 26 and the character pitch estimation product 29, and similarly determines character cutting positions based on these 1g numbers.

以上のように第2の実施例は帳票上の文字列によシ得ら
れる文字ピッチを推足し1文字の切出しを行なうもので
ある。即ち、不発明の第20笑施例は、2値の行イメー
ジを投影して得られた文字行方向と文字行と直交する方
間の投影信号からめられる推定文字ピッチと、得られた
投影信号よ請求まる文字の存在範囲の右端又は左端とか
ら位置を決め、前記位1を全開始位置と、前記投影16
号および文字高さから決まる範囲内のiμ分投影信号を
、推定文字ピッチによる等間隔位置で重ね合せ(又はf
cfcみ込み)すなわち1等間隔位置でカロ算しまる加
算値が最小となる位置をめ、前記開始位置とで切出し、
開始位置をめ、得られた切出し開始位置より推定文字ピ
ッチを用いた等間隔ごとの位置と七の位mの周辺位置に
おける文字行と直交する方向の投影信号にもとづき2値
の行イメージを切出す文字切出し装置である。
As described above, the second embodiment extracts one character by adding up the character pitch obtained from the character string on the form. That is, the uninvented 20th embodiment calculates the estimated character pitch obtained from the character line direction obtained by projecting a binary line image and the projection signal in the direction orthogonal to the character line, and the obtained projection signal. Determine the position from the right or left end of the existence range of the character called ``Yo'', and set the position 1 as the entire starting position and the projection 16.
The iμ-minute projection signals within the range determined by the symbol and character height are superimposed (or f
cfc inclusion), that is, find the position where the added value calculated by Calo at one equal interval position is the minimum, and cut out with the starting position,
After determining the starting position, a binary line image is cut based on the projected signal in the direction orthogonal to the character line at positions at equal intervals using the estimated character pitch and the peripheral position of the seventh digit m from the obtained cutting start position. This is a character cutting device.

第10図は不発明の第3の実施例τ示す、 gi。FIG. 10 shows a third embodiment of the invention τ, gi.

図(a)において、第3の実施例は第1の実施例と同様
に帳票上の一竹分の文字列t2値1d号として自と憶す
る行イメージ記憶郡31と、該行イメージ記憶部31に
接続され、前記2値1−号炉ら文字の上端、下端および
文字向さ請求める文字向は検出部38と、前記何イメー
ジ記憶部31および前記文字向さ検出部38に接続され
、前記21直16号と文字向さから1文字位置範囲を走
立し、ブランクラインおよび文字ラインをめる投影処理
部32と。
In Figure (a), the third embodiment, like the first embodiment, has a row image storage group 31 that stores the character string t2 value 1d for one character string on a form, and a row image storage section 31, the upper end, lower end, and character orientation of the binary 1-character are connected to the character orientation detection unit 38, and connected to the image storage unit 31 and the character orientation detection unit 38; and a projection processing unit 32 that runs a one character position range from the 21 straight 16 and the character direction and inserts a blank line and a character line.

該投影処理部32に接続され、前記2値1ば号のブラン
クラインか文字ラインか全力(す投影16号を目ピ憶す
る投影信号記1急郡33と、前記文字高さ検出部38に
接続され1文字高さ・16号から文字ピッチをめ、これ
全記憶する文字ピッチ記憶部34と。
A projection signal register 33 is connected to the projection processing section 32 and memorizes the blank line or character line of the binary 1 bar (projection number 16) and the character height detection section 38. It is connected to a character pitch storage section 34 which determines the character pitch from the height of one character and number 16, and stores all of this.

投影信号a記憶部33および文字ピッチ記憶部34に接
続され、投影信号を評価し、切出し開始位置をめる評価
処理部35と、該評価処理部35からの切出し開始位置
を記憶する切出し開始位置記憶部36と、前記行イメー
ジ記憶部31.文字ピ、チ記憶部34および切出し開始
位置記憶部36に接続され、文字の切出し位置を決定す
る切出し処理部37とを含む。
an evaluation processing section 35 that is connected to the projection signal a storage section 33 and the character pitch storage section 34 and evaluates the projection signal and determines a cutting start position; and a cutting start position that stores the cutting start position from the evaluation processing section 35. a storage section 36; and the row image storage section 31. It includes a cutout processing section 37 that is connected to the character pitch/chi storage section 34 and the cutout start position storage section 36 and determines the cutout position of the character.

まず第3の実施例は帳票上を走査して侍られた一部分の
2伝信号を行イメージ記憶部31にセットする。次に文
字高さ検出部38により前記一台分の2値信号から文字
の上端、下端および文字間さをめる1文字高さがめられ
ると、投影処理部32は行イメージ記憶部31にセット
されている2伝信号を行方向と直交する方間に文字高さ
検出部38で得られた文字の上端、下端及び文字高さか
らまった位置範囲内で走査して、走査し/こラインがブ
ランクラインであれば′1”文字ラインであれば0”に
する2伝信号を出力し投影信号記憶部33に送る。
First, in the third embodiment, a document is scanned and two transmission signals of the scanned part are set in the line image storage section 31. Next, when the character height detection unit 38 determines the height of one character by calculating the upper end, lower end, and character spacing from the binary signal for one unit, the projection processing unit 32 sets the height in the line image storage unit 31. 2 transmission signals are scanned in a direction orthogonal to the line direction within a position range defined by the upper edge, lower edge, and character height of the character obtained by the character height detection unit 38, and the line is scanned. If it is a blank line, it outputs a second transmission signal that changes it to '1' and if it is a character line, it changes to '0' and sends it to the projection signal storage section 33.

一方文字ピッチ記憶部34では文字高さ検出部38から
の文字高さにもとすき、予め定まった文字ピッチが記憶
される。評価処理部35は2値の投影1ざ号と文字ピッ
チから切出し開始位置をめ、切出し開始位置記憶部36
に供給する。 9J出し処理部37では行イメージの2
値1δ号1文字ピッチおよび切出し開始位置にもとすい
て又字毎の切出し位置をめる・ 第1θ図(blは本発明の第3の実施例の変形で、切出
し処理品37′を有し、この切出し処理部37′は投影
・1百号を加味して文字毎の切出し位iifをめる。
On the other hand, in the character pitch storage section 34, a predetermined character pitch is stored in addition to the character height from the character height detection section 38. The evaluation processing unit 35 determines the cutting start position from the binary projection number 1 and the character pitch, and stores the cutting start position in the cutting start position storage unit 36.
supply to. In the 9J output processing unit 37, 2 of the row image
The value 1δ is based on the 1-character pitch and the cutting start position, and the cutting position for each character is also set. However, this cutout processing section 37' calculates the cutout position iif for each character by taking into account the projection and 100 numbers.

以上のように不発明の第3の実施例は文字高さから限定
される文字行と直交する範囲内のみの定歪で投影信号を
侍て2文字の切出し位置をめるものである。
As described above, the uninvented third embodiment determines the cutout positions of two characters by receiving the projection signal with constant distortion only within the range orthogonal to the character line, which is limited by the character height.

即ち5不発明の第3の実施例は2値の何イメージを投影
して倚らる文字高さ信号葡求め、この文字高さ範囲内で
、2値の何イメージを投影して得られる投影11号と、
文字高さ信号からめられる推定文字ピッチとにより、前
記投影信号よ請求まる文字の存在範囲の右端又は左端と
から位置を決め、前記位置を開始位置として前記行イメ
ージの2伝信号を推定文字ピッチによる等間隔位置で息
ね合せ(およびたたみ込み)、すなわち力g其し、求ま
る加算値が最小となる位置をめ、前記開始位置とで切出
し開始位置をめ、得られた切出し開始位置よプ順次推定
文子ピッチの間隔ごとに2値の行イメージを切出す文字
切出し装置である。
In other words, the third embodiment of the invention is to obtain a character height signal obtained by projecting a binary image, and to obtain a projection obtained by projecting a binary image within this character height range. No. 11 and
Based on the estimated character pitch derived from the character height signal, determine the position from the right or left end of the existence range of the character requested by the projection signal, and use the position as the starting position and generate the second transmission signal of the line image based on the estimated character pitch. Balance (and convolution) at equally spaced positions, that is, find the position where the force g and the obtained added value are the minimum, find the cutting start position with the above starting position, and sequentially from the obtained cutting start position This is a character cutting device that cuts out binary line images at intervals of estimated sentence pitch.

第11図は不発明の第4の実施例をボす。第11図にお
いて、第4の実施例は第1G、l來施例と同様に帳票上
の一部分の文字列全21直16号として記憶する行イメ
ージ記憶部41と、該行イメージ記憶部41に接続され
、前記2伝信号を走をしてブランクライン會検出し、投
影匿号を侍、かつ文字りりの特徴を検出し、その特徴値
を検出し、特徴11号を得る投影処理部42と、該投影
処理部42に接続され、前記投影信号全記憶する投影1
J号記憶部43と、前記投影処理部42に接続され、前
記特徴信号を記憶する特徴信号記憶部49と、予め文字
ピッチを記憶する文字ピッチ記憶部44と、前記特徴値
号、投影侶号および文字ピッチにもとついて文字の切出
し位置をめる評価処理部45と、評価処理部45に接続
され、文字の切出し位置信号にもとすいて文字の切出し
開始位置全記憶する切出し開始位置記憶部46と、前記
何イメージ記憶部41.投影16号記憶部43%文字ピ
ッチ記憶部44および切出し開始位置記憶部46に接続
され、これらの信号にもと3′き、文字列の切出し位#
、を決定する切出し処理′$47とを含む。
FIG. 11 shows a fourth embodiment of the invention. In FIG. 11, the fourth embodiment has a line image storage unit 41 that stores character strings of a part of a form as all 21 straight numbers and 16 numbers, as in the first embodiment. and a projection processing unit 42 which is connected, runs the two transmission signals to detect a blank line, detects the feature of the projection code being Samurai and the character Riri, detects the feature value, and obtains feature No. 11. , a projection 1 connected to the projection processing section 42 and storing all the projection signals.
J number storage section 43, a feature signal storage section 49 connected to the projection processing section 42 and storing the feature signal, a character pitch storage section 44 that stores the character pitch in advance, and the feature value number and the projection number number. and an evaluation processing unit 45 that determines the character extraction position based on the character pitch, and a extraction start position memory that is connected to the evaluation processing unit 45 and stores all character extraction start positions based on the character extraction position signal. section 46, and the image storage section 41. The projection number 16 storage unit 43% is connected to the character pitch storage unit 44 and the cutting start position storage unit 46, and based on these signals, the cutting position # of the character string is determined by 3'.
, and an extraction process '$47 to determine .

まず、第4の実施例は帳票上を走査して得られた一部分
の2稙狼号が行イメージ記憶部41にセ、トされる。投
影処理部42は行イメージ記憶部41にセットされてい
る2伝信号を行方向と直交する方向に走査して、走査ラ
インがブランクラインであれは”1”1文字ラインであ
れば0”とする21fflの投影1言号を記帳する投影
1g+3’記憶部43に出力する。また投影処理部42
は行イメージ記憶都41にセットされている21直信号
を行方向に直交する方向に走査する際、たとえば、第1
2図に示すマスクで特徴を抽出し、検出された特徴値を
特徴信号記憶部49に出力する。第13図(a)に示す
る文字例イメージは第12図のマスクをあてて走査して
、マスクに一攻するかどうかを調べ2値の特徴値がめら
れ、第13図tblに示すような特徴値分布が得られる
。同図かられかるように文字の両端では2伝信号の′0
”となる。文字ピッチ記憶部44にはあらかじめわかっ
ている文字ピッチ長に相轟するビクセル長が記憶されて
おう、評価処理部45では投影信号記憶部43にセット
されている一行の文字の投影信号から右端の文字の右端
又は左端の文字の左端位置に対応する位置を、最初に′
θ″になる位*に検出することによ請求める0次に文字
ピッチ記憶部44に記憶されている文字ピッチを用いて
、前述の文字の右端又は左端から右側又は左側に1文字
ピッチ離れた位置7求め、得られた位置ケ開始位置とし
て特徴信号記憶部49の2伝信号を重ね合せて頻度分布
を作成し、得られた頻度分布よシwO”が最も多い位置
を検出し、その位置までの長さで前記開始位置rシフト
して得られる位置全切出し開始位置とし、切出し開始位
置記憶部46に送る。切出し開始位置が決凍ると、切出
し処理部47は文字ピッチ記憶部440文字ピッチと切
出し開始位l1tsピ憶fis46の切出し開始位置と
で行イメージ記憶部41の2値イメージを等分割し、−
文字に対応する2値イメージとする。又は切出し処理部
47は文字ピッチ記憶部44の文字ピッチで切出し開始
位置記憶部46の切出し開始位置から等間隔で決まる位
置に対応する投影1δ号記憶部430投影侶号をチェッ
クし、2伝信号の1″であるブランクラインであれは切
出し位置とし5文字ラインでブランクラインでなけれは
その位置の周囲ぞ最も近いブランクライン位置管、葦た
ブランクラインがなけれは文字ピッチで決まる位置を切
出し位置として、−文字に対応する2値イメージを2行
イメージ記憶部41の2値イメージを分離し、切出し位
置でめる。
First, in the fourth embodiment, a portion of the 2nd line number obtained by scanning the form is set in the line image storage section 41. The projection processing unit 42 scans the two-transmission signal set in the line image storage unit 41 in a direction perpendicular to the line direction, and if the scanning line is a blank line, it is “1” and if it is a single character line, it is “0”. The projection 1g+3' storage unit 43 records the projection 1 word of 21ffl.
For example, when scanning the 21 orthogonal signal set in the row image storage capital 41 in a direction orthogonal to the row direction,
Features are extracted using the mask shown in FIG. 2, and the detected feature values are output to the feature signal storage section 49. The character example image shown in Fig. 13 (a) is scanned by applying the mask shown in Fig. 12, checking whether it hits the mask or not, and finding a binary feature value, as shown in Fig. 13 tbl. A feature value distribution is obtained. As can be seen from the figure, at both ends of the character, there are two transmission signals '0
”. The character pitch storage unit 44 stores the pixel length that corresponds to the character pitch length known in advance, and the evaluation processing unit 45 calculates the projection of one line of characters set in the projection signal storage unit 43. First, select the position corresponding to the rightmost position of the rightmost character or the leftmost position of the leftmost character from the signal.
0, which can be requested by detecting * at the position where θ'' is reached. Next, using the character pitch stored in the character pitch storage unit 44, the character pitch that is one character pitch away from the right or left end of the aforementioned character to the right or left side. Position 7 is determined, the obtained position is used as the starting position, and the two transmission signals in the characteristic signal storage unit 49 are superimposed to create a frequency distribution. Based on the obtained frequency distribution, the position with the most ``wO'' is detected, and the position is The position obtained by shifting the starting position r by the length up to r is set as the entire cutting start position and is sent to the cutting start position storage section 46. When the extraction start position is fixed, the extraction processing unit 47 equally divides the binary image in the line image storage unit 41 by the character pitch of the character pitch storage unit 440 and the extraction start position of the extraction start position l1ts file fis 46, and -
It is a binary image corresponding to a character. Alternatively, the cutout processing unit 47 checks the projection 1δ number storage unit 430 projection number corresponding to the position determined at equal intervals from the cutout start position in the cutout start position storage unit 46 based on the character pitch in the character pitch storage unit 44, and calculates the 2nd transmission signal. If it is a 1" blank line, the cutting position will be used. If it is a 5 character line and it is not a blank line, the closest blank line position will be around that position. If there is no reed blank line, the cutting position will be the position determined by the character pitch. , - Separates the binary image corresponding to the character from the two-line image storage section 41 and stores it at the cutting position.

このように不発明の第4の実施例は切出し開始位置を決
める評価に、2値の行イメージから抽出される特徴値を
使用し、文字の切出し位1ft−決定するものである。
As described above, the uninvented fourth embodiment uses feature values extracted from a binary line image for evaluation to determine the cutting start position, and determines the character cutting position 1 ft.

即ち、不発明の第4の実施例は2値の行イメージを投影
する際に特徴τ抽出し投影特徴信号とし。
That is, in the uninvented fourth embodiment, when projecting a binary row image, the feature τ is extracted and used as a projected feature signal.

あらかじめ与えられる文字ピッチと1文字行と直交する
方向の投影信号よりまる文字の存仕範囲の右端又は左端
とから位置を決め、前記位置を開始位置としてMil記
文字行と直交する方向の投影特徴信号全前記文字ピッチ
による号間隔位置で重ね合せ(又はたたみこみ)%すな
わち、加昇してまる加算値が最小となる位置をめ、前記
開始位置とで切出し開始位t+fe求め、得られた切出
し開始位置よシ前述の文字ビッチスは推定文字ピッチを
用いた等間隔ごとの位置とその位置の周辺位置における
文字行と直交する方間の投影・11号にもとづき2il
l!の行イメージを切出す文字切出し装置である。
The position is determined from the right or left end of the existing range of characters based on the character pitch given in advance and the projection signal in the direction perpendicular to one character line, and the projection feature in the direction perpendicular to the Mil character line is determined from the position as the starting position. Find the position where the overlapping (or convolution) percentage, that is, the sum value of the whole signal is the minimum, at the symbol interval position according to the character pitch, calculate the cutting start position t + fe from the above starting position, and calculate the obtained cutting start position. The above-mentioned character bits are based on 2il based on No. 11, a projection between positions at equal intervals using the estimated character pitch and a direction perpendicular to the character line at surrounding positions of the position.
l! This is a character cutting device that cuts out line images of .

第14図は不発明の第5の実施例を示す。第14図にお
いて、第5の実施例は第1の実施例と同様に帳票上の一
部分の文字列を2値1m号として記憶する行イメージ記
憶部51と、該行イメージ記憶部51に接続され、前記
2領置号を走査してブランクラインおよび文字ラインを
検出し、投影信号を得る投影処理部52と、該投影処理
部52に接続され、前記投影信号を記憶する投影1a@
記憶部53と、該投影信号記憶部53に接続され、投影
信号により、文字の切出し位置をめる評価処理部55と
、該評価処理部55に微枕され5文字の切出し開始位置
全記憶する切出し開始位置記憶部56と、前記行イメー
ジd己憶部51に接角洗され。
FIG. 14 shows a fifth embodiment of the invention. In FIG. 14, the fifth embodiment has a line image storage unit 51 that stores a part of a character string on a form as a binary number 1m, and a line image storage unit 51 that is connected to the line image storage unit 51, as in the first embodiment. , a projection processing unit 52 that scans the two markings to detect blank lines and character lines and obtains a projection signal, and a projection 1a@ connected to the projection processing unit 52 and storing the projection signal.
a storage section 53; an evaluation processing section 55 connected to the projection signal storage section 53 and determining the cutting position of characters based on the projection signal; The tangent angle is stored in the cutting start position storage section 56 and the row image data storage section 51.

前記2値1百号から文字高さをめる文字高さ検出s58
と、前記文字高さ1a号、投影信号おまひ文字の切出し
位置をボす1g号によシ推定文竿ピッチをめる文字ピッ
チ推定部59と、前記2値偏号、切出し論始位置消号お
よび推定文字ピッチの信号によシ文字列の切出し位置を
める切出し処理部57とを宮む。
Character height detection s58 that calculates the character height from the binary value 100
and a character pitch estimating unit 59 which calculates the estimated text pitch based on the character height 1a and the projection signal 1g which marks the cutting position of the character, and the binary polarization code and the cutout starting position. and a cutout processing section 57 which determines the cutout position of the character string based on signals of the character string and the estimated character pitch.

この第5の実施例は、最初にまった推定ピッチによシ決
貰った切出し開始位置の周囲で推定文字ピッチ*を増減
して再評価によシ切出し開始位置を得るが、まず1文字
ピッチ抽足部59からの推定文字ピッチにより最初の切
出し開始位置をめる0次に評価処理部55は文字ピッチ
推足部59から出力された文字ピッチ長を増減して得ら
れる複数の文字ピッチ長に対して最初の切出し開始位置
の周囲の位置で前述と同じく投影信号記憶部53の2伝
信号を重ね合せて頻度分布を作成し、得られた頻度分布
よシ切出し開始位置の再決定を推定文字ピッチの修正を
行う、評価処理部55は出力される切出し開始位置を切
出しb(」始位置記憶部56にセットし、また修正した
推定文字ピッチを文字ピッチ推定郡59に送る。切出し
処理部57は文字ピッチ推足部59からの文子ピッチと
切出し開始位置記憶部56の切出し開始位置とで行1゛
メージ記憶品51の2値イメージを等分割し、−文字に
対応する2値イメージを作p%また。i&初に得られた
文字ピッチが正確に推定出来ないときに再評価によp文
字ピッチを修正するようにする。
In this fifth embodiment, the estimated character pitch * is increased or decreased around the cutout start position determined based on the initially estimated pitch, and the cutout start position is obtained through re-evaluation. The first cutting start position is determined based on the estimated character pitch from the extraction section 59.The evaluation processing section 55 calculates a plurality of character pitch lengths obtained by increasing or decreasing the character pitch length output from the character pitch estimating section 59. , a frequency distribution is created by superimposing the two transmission signals of the projection signal storage unit 53 at positions around the first extraction start position as described above, and the re-determination of the extraction start position is estimated based on the obtained frequency distribution. The evaluation processing unit 55, which corrects the character pitch, sets the output cutting start position in the cutting start position storage unit 56 and sends the corrected estimated character pitch to the character pitch estimation group 59. 57 equally divides the binary image of the row 1 image memory 51 using the sentence pitch from the character pitch extension section 59 and the cutting start position of the cutting start position storage section 56, and divides the binary image corresponding to the - character into equal parts. Also, when the initially obtained character pitch cannot be accurately estimated, the p character pitch is corrected by re-evaluation.

このように不発明の第5の実施例は一度求まった文字ピ
ッチを修正して2文字の切出し位置をもとめるものであ
る。
As described above, the uninvented fifth embodiment corrects the character pitch once determined to determine the cutting positions of two characters.

なお、本発明は前記実施例に限足避れるものではなく、
文字ピッチの推定手段として隣合う文字の端間距離よ請
求める方法等を採用しても良い。
Note that the present invention is not limited to the above embodiments,
As a means for estimating the character pitch, a method that calculates the distance between the edges of adjacent characters may be adopted.

また走査の際に特徴の検出に用いるマスクは1文字の端
と中心部とでは異なる出力を出すマスクであっても良く
、特徴値としてマスクに一致する頻度のような多値であ
っても良く、さらに又1文字ピッチで決まる位置の周囲
で最適な切出し位置を検出する際、評価対象の位置範囲
を文字ピッチ推だのときに得られる統計量から決定して
も良い。
Furthermore, the mask used to detect features during scanning may be a mask that produces different outputs for the edges and the center of one character, or may be a multi-valued feature value such as the frequency of matching the mask. Furthermore, when detecting the optimal cutting position around a position determined by one character pitch, the position range to be evaluated may be determined from the statistical amount obtained when character pitch estimation is performed.

不発明は以上説明したように帳票上に文字又は印字の開
始位置が与えられなくとも誤シなく切出すことができ、
さらに文字ピッチが与えられない際には文字ピッチの推
定分合うので、文字ピッチが既知であるタイプ文字の他
に枠金基準に記入した文字列でも切出すことができる効
果がある。
As explained above, non-invention can be cut out without any error even if the starting position of characters or printing is not given on the form,
Furthermore, when the character pitch is not given, the estimated character pitch is matched, so that in addition to typed characters whose character pitch is known, it is also possible to cut out character strings written on frame standards.

即ち、不発明の第5の実施例は文字行方向と文字行に直
交する方向の投影信号からめられた推定文字ピッチ長と
前記文字行に直交する方向の投影信号よ請求まる文字の
存在範囲の右端又は左端とから位置を決め、前記位置を
開始位置として前記文字行に直交する方向の投影信号を
、前記推定文字ピッチ長による等間隔位置でたたみこみ
、すなわち加算しまる加算値が最小となる位置をめ、前
記開始位置とで切出し開始位置7求め1次に前記切出し
開始位置の周囲の谷位置において前記推定文字ピッチを
増減して得られる長さごとに同方法で加算値分布をめ、
#られfc加算値分布のうちで最小加算値が最小となる
分布の最小頻度位置と前記切出し位置とで切出し開始位
置を修正し、対応する長さを推定文字ピッチ長として修
正し、得られた切出し開始位置よp文字ピッチを用いた
等間隔ごとの位置で2値の行イメージを切出す文字切出
し装置である。
That is, the uninvented fifth embodiment uses the estimated character pitch length obtained from the character line direction and the projection signal in the direction orthogonal to the character line, and the existing range of the character claimed by the projection signal in the direction orthogonal to the character line. A position is determined from the right end or the left end, and the projection signal in the direction orthogonal to the character line is convolved at positions at equal intervals according to the estimated character pitch length, using the starting position as the starting position, that is, the position where the added value is the minimum. , find a cutting start position 7 with the starting position, firstly, calculate the added value distribution using the same method for each length obtained by increasing or decreasing the estimated character pitch at valley positions around the cutting start position,
The cutout start position is corrected using the cutout position and the minimum frequency position of the distribution where the minimum added value is the smallest among the fc added value distributions, and the corresponding length is corrected as the estimated character pitch length. This is a character cutting device that cuts out a binary line image at positions at regular intervals using p character pitches from the cutting start position.

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

第1図は帳票上の文字行の一例を示す図、第2図は第1
図の文字行の2値イメージを文字ピッチ内で重ね合せた
flkKヒストグラムを示す図、第3図は接触した文字
と垂直方間で重なった文字とt示す図、@4図は切出し
位置を移動する方が望ましい文字ペアを不す図%第5図
は本発明の@1の実施例を示すブロック図、第6図ta
+は2値の行イメージを示す図、第6図tb+は第6図
(a)の2イ直イメージからまる2値の投影信号を示す
図、第7図−は切出し開始ii?=(ll−変えて得ら
れる2値の投影信号の頻度分布忘示す図、第8図は不発
明の第2の実施例を示すブロック図、第9図は2埴イメ
一ジ金文字行方向に投影して得られる州度分布f y3
<す図、第10図は不発明の第3の実施例を示すブロッ
ク図%@11図は不発明の第4の来JM丙金がすブロッ
ク図、第12図は特徴抽出マスクを示す図、第13図は
文字例イメージおよび第12図のマスクとから得られた
特徴値分布を示す図、第14図は不発明の第5の実施例
を示すブロック図である。 11.21,31,41.51・・・・・行イメージ記
憶部%12,22,32,42.52・・・・・・投影
処理部、13.13’ 、 23.23’ 、 33.
33’。 43.53・・・・・・投影信号記憶部、14,34,
44゜・・・・・・文字ピッチ記憶部、15.25,3
5.55・・・・・・評価処理部、16,26,36.
46.56・・・・・・切出し開始位置配憶部、17,
27,27’。 37.37’ 、47.57・・・・・・切出し処理部
、28,38,58・・・・・・文字尚さ検出部、29
゜59・・・・・文字推定チ推足部、49・・・・・・
特徴信号配憶部。 h1国 <a) (bン 、(C2 把2国 乙 力4閃 (b) 沁5閃 (61) A B (bン 力6閃 乃8圓 乃7圓 力17閃 (α〕 (b) 力70図 力lI閃 力13閃 力14閃°7
Figure 1 is a diagram showing an example of character lines on a form, and Figure 2 is a diagram showing an example of character lines on a form.
Figure 3 shows a flkK histogram in which the binary images of the character lines in the figure are superimposed within the character pitch, Figure 3 shows the characters that overlap vertically with the characters that touched them, and Figure @4 shows the cutting position moved. Figure 5 is a block diagram showing an embodiment of @1 of the present invention;
+ shows a binary row image, FIG. 6 tb+ shows a complete binary projection signal from the 2-direct image of FIG. 6(a), and FIG. 7- shows the start of cutting ii? =(ll- Figure 8 is a block diagram showing the second embodiment of the present invention, and Figure 9 is a two-panel image with a gold character line direction. The state degree distribution f y3 obtained by projecting on
Figure 10 is a block diagram showing the third embodiment of the invention. Figure 11 is a block diagram of the fourth embodiment of the invention. Figure 12 is a diagram showing the feature extraction mask. , FIG. 13 is a diagram showing the feature value distribution obtained from the character example image and the mask of FIG. 12, and FIG. 14 is a block diagram showing the fifth embodiment of the present invention. 11.21, 31, 41.51... line image storage section %12, 22, 32, 42.52... projection processing section, 13.13', 23.23', 33.
33'. 43.53... Projection signal storage section, 14, 34,
44°...Character pitch storage section, 15.25,3
5.55...Evaluation processing section, 16, 26, 36.
46.56... Cutting start position storage section, 17,
27, 27'. 37.37', 47.57... Cutting processing unit, 28, 38, 58... Character straightness detection unit, 29
゜59...Character estimation chi push part, 49...
Feature signal storage section. h1 country<a) (bn, (C2 2 countries otriki 4sen (b) 沁5sen (61) A B (bnriki 6senno 8enno 7en power 17sen (α)) (b) Power 70 Power lI Shinry 13 Shinry 14 Shin° 7

Claims (1)

【特許請求の範囲】 (1121mの行イメージを投影して得られる投影信号
を用いて文字切出し位置をめる文字切出し装置において
、前記投影信号を文字列の端から離れた位置より文字ピ
ッチの幅でたたみこみ會行い、得られる頻度分布の特徴
位置と、前記文字列の端位置とから切出し開始位置をめ
る千成と前記切出し開始位置と文字ピッチを用いて個々
の文字切出し位置を決定する手段とを具備することt−
%徴とする文字切出し装置。 (2)文字ピッチは既知の長さである特許請求の範囲第
fl)項6己載の文字切出し装置。 (3)文字ピッチは投影信号よシ推足される長さである
特許請求の範囲第1項記載の文字切出し装置・ (4)投影11号は文字行と直交する方向に文字の高さ
で決まる範囲内を走査して検出される文字ラインかブラ
ンクラインかの2値信号であシ、頻度分布の特徴位置は
頻度が最小となる位置とする特許請求の範囲第(2)項
及び第(3)項記載の文字切出し装置。 (5)投影信号は文字行と直交する方向にマスクを当て
て走査し抽出されるブランクライン又は文字の端かそれ
以外かの2値信号である特許請求の範囲第(2)項及び
第(3)項記載の文字切出し装置。 (6)切出し開始位置から文字ピッチ長ごとに決まる位
ft’e個々の文字切出し位置とする特許請求の範囲第
(4)項又は第(5)項記載の文字切出し装置。 (7) 切出し開始位置から文字ピッチ長で決まる位置
の周囲で投影信号を調べ検出されるブランクラインを個
々の文字切出し位置とする特許請求の範囲第(4)項又
は第(5)項記載の文字切出し装置。 (8)2値の行イメージを投影して得られる投影1g号
を用いて文字切出し位置?!−求める文字切出し。 装置において、前記投影信号を文字列の端から離れた位
置よル文字ピッチの幅でたたみごみを行い得られる頻度
分布の特徴位置と、前記文字列の端からの位置とから切
出し開始位置をめる手段と、前記切出し開始位置の周囲
の谷位置よシ前記文字ピッチ長を増減してまる各幅で前
記投影信号音たたみこみ、得られる頻度分布のうちで特
徴位置の頻度が最小の頻度分布の%微位置から修正切出
し開始位置を決足する手段と。 対応する幅全文字ビ、チ長とし、得られた前記切出し開
始位置と前記文字ピッチを用いて1同々の文字切出し位
fを決足する手段と全具備したことを特徴とする文字切
出し装置。 (9)最初の文字ピッチは既知の長さである特許請求の
範囲第(8)項記載の文字切出し装置。 (10)最初の文字ピッチは投影信号よシ推定される長
さである特許請求の範囲第8項記載の文字切出し装置。 (11)投影信号は文字行と直交する方向に丈高の高さ
で決まる範囲内を走査して検出される文字ラインかブラ
ンクラインかの2値信号であシ、頻度分布の特徴位置は
頻度が最小となる位置とする特許請求の範囲第(9)項
及び第(10)項記載の文字切出し装置。 (12)投影信号は文字行と@又する方向にマスクを当
てて走査し抽出されるブランクライン又は文字の端かそ
れ以外かの2値信号であシ、頻度分布の特徴位置は頻度
が最小となる位置とする特許請求の範囲第(9)項及び
第(lの項記載の文字切出し装置。 (13)切出し開始位置から文字ピッチ長ごとに決まる
位置を個々の文字切出し位置とする特許請求の範囲第即
)項又は第(12)項記載の文字切出し装装置・ (10切出し開始位置から文字ピッチ長で決まる位置の
周囲で投影信号tNべ検出されるブランクラインを個々
の文字切出し位置とする特許請求の範囲第(1)項又は
第(12)項記載の文字切出し装置・
[Scope of Claims] (In a character cutting device that determines a character cutting position using a projection signal obtained by projecting a line image of 1121 m, means for determining individual character extraction positions using the extraction start position and character pitch; and t-
A character cutting device that uses percentage marks. (2) The character pitch is a known length.Claim No. fl)Claim 6: A self-contained character cutting device. (3) The character pitch is the length of the projected signal. (4) The projection No. 11 is the height of the character in the direction orthogonal to the character line. It is a binary signal of a character line or a blank line detected by scanning within a determined range, and the characteristic position of the frequency distribution is the position where the frequency is minimum. Character cutting device described in section 3). (5) The projection signal is a binary signal indicating whether it is a blank line or the edge of a character or not, which is extracted by scanning with a mask applied in a direction perpendicular to the character line. Character cutting device described in section 3). (6) The character cutting device according to claim (4) or (5), wherein the character cutting position is set at a position ft'e determined for each character pitch length from the cutting start position. (7) The method according to claim (4) or (5), wherein the blank line detected by examining the projection signal around the position determined by the character pitch length from the cutting start position is the individual character cutting position. Character cutting device. (8) Character cutting position using projection No. 1g obtained by projecting a binary line image? ! -Cut out desired characters. In the apparatus, the cutout start position is estimated from the characteristic position of the frequency distribution obtained by folding the projection signal at a position away from the end of the character string and the width of the character pitch, and the position from the end of the character string. means for convolving the projected signal sound with each width by increasing or decreasing the character pitch length from the valley position around the cutting start position, and convolving the projected signal sound with each width of the valley position around the cutting start position, and convolving the projected signal sound with the minimum frequency of the characteristic position among the frequency distributions obtained. Means for determining the corrected cutting start position from the % minute position. A character cutting device characterized in that the character cutting device has a means for determining a character cutting position f using the obtained cutting start position and the character pitch, and determining the cutting position f for each character using the obtained cutting start position and the character pitch. . (9) The character cutting device according to claim (8), wherein the initial character pitch has a known length. (10) The character segmentation device according to claim 8, wherein the initial character pitch is a length estimated from the projection signal. (11) The projection signal is a binary signal of character line or blank line detected by scanning within the range determined by the height in the direction orthogonal to the character line, and the characteristic position of the frequency distribution is the frequency A character cutting device according to claims (9) and (10), in which the position is the minimum. (12) The projection signal is a binary signal of the blank line or the edge of the character or other points extracted by scanning with a mask applied in the direction opposite to the character line, and the characteristic position of the frequency distribution has the minimum frequency. A character cutting device according to claims (9) and (l) in which the position is set as follows. (13) A patent claim in which the position determined for each character pitch length from the cutting start position is the position for cutting out each character. The character cutting device described in item (1) or item (12) (10) The blank line detected by the projection signal tN around the position determined by the character pitch length from the cutting start position is determined as the individual character cutting position. A character cutting device according to claim (1) or (12)
JP59002910A 1984-01-11 1984-01-11 Device for segmenting character Pending JPS60146376A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP59002910A JPS60146376A (en) 1984-01-11 1984-01-11 Device for segmenting character

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
JP59002910A JPS60146376A (en) 1984-01-11 1984-01-11 Device for segmenting character

Publications (1)

Publication Number Publication Date
JPS60146376A true JPS60146376A (en) 1985-08-02

Family

ID=11542510

Family Applications (1)

Application Number Title Priority Date Filing Date
JP59002910A Pending JPS60146376A (en) 1984-01-11 1984-01-11 Device for segmenting character

Country Status (1)

Country Link
JP (1) JPS60146376A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPS62103777A (en) * 1985-10-31 1987-05-14 Toshiba Corp Character stretch-breaking system

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPS5750076A (en) * 1980-09-10 1982-03-24 Toshiba Corp Character reader
JPS5751146A (en) * 1980-09-09 1982-03-25 Nippon Telegr & Teleph Corp <Ntt> Preparation of raw material for fluoride optical fiber
JPS57137972A (en) * 1981-02-20 1982-08-25 Nec Corp Character out position detecting method
JPS57189274A (en) * 1981-05-15 1982-11-20 Nippon Telegr & Teleph Corp <Ntt> Character cutout device

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPS5751146A (en) * 1980-09-09 1982-03-25 Nippon Telegr & Teleph Corp <Ntt> Preparation of raw material for fluoride optical fiber
JPS5750076A (en) * 1980-09-10 1982-03-24 Toshiba Corp Character reader
JPS57137972A (en) * 1981-02-20 1982-08-25 Nec Corp Character out position detecting method
JPS57189274A (en) * 1981-05-15 1982-11-20 Nippon Telegr & Teleph Corp <Ntt> Character cutout device

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPS62103777A (en) * 1985-10-31 1987-05-14 Toshiba Corp Character stretch-breaking system

Similar Documents

Publication Publication Date Title
CN101251892B (en) Method and apparatus for cutting character
CN1332348C (en) Blocks letter Arabic character set text dividing method
US8201742B2 (en) Barcode processing apparatus and barcode processing method
US4878124A (en) Image inclination detecting method and apparatus
JPS6077279A (en) Initiation of character image
US7680329B2 (en) Character recognition apparatus and character recognition method
EP0248262B1 (en) Apparatus and method for detecting character components on a printed document
US4887301A (en) Proportional spaced text recognition apparatus and method
JPS60146376A (en) Device for segmenting character
JP5041775B2 (en) Character cutting method and character recognition device
JP3092576B2 (en) Character recognition device
JP3954246B2 (en) Document processing method, recording medium storing document processing program, and document processing apparatus
JP4192886B2 (en) Tamper detection system, tamper detection device, threshold determination device, tamper detection method, threshold determination method
JP4158762B2 (en) Tamper detection threshold determination system, evaluation target data creation device, threshold determination device, evaluation target data creation method, threshold determination method
JP2813601B2 (en) Tabular document recognition device
JPS60132281A (en) Character separating device
JP2009053931A (en) Document image processor and document image processing program
JPH0782524B2 (en) Optical character reader
JP2008123181A (en) Musical score recognition device and program
JPH07192087A (en) Optical character reader
KR101495656B1 (en) A method of recognizing magnetic ink character
JP2000207490A (en) Character segmenting device and character segmenting method
JPH05128308A (en) Character recognition device
JPH0467674B2 (en)
JPH0221385A (en) Printer