JPS581277A - Character characteristic pickup system - Google Patents

Character characteristic pickup system

Info

Publication number
JPS581277A
JPS581277A JP56099931A JP9993181A JPS581277A JP S581277 A JPS581277 A JP S581277A JP 56099931 A JP56099931 A JP 56099931A JP 9993181 A JP9993181 A JP 9993181A JP S581277 A JPS581277 A JP S581277A
Authority
JP
Japan
Prior art keywords
character
contour
feature data
profile
characteristic
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
JP56099931A
Other languages
Japanese (ja)
Inventor
Tetsuji Morishita
森下 哲次
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.)
Fujitsu Ltd
Original Assignee
Fujitsu 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 Fujitsu Ltd filed Critical Fujitsu Ltd
Priority to JP56099931A priority Critical patent/JPS581277A/en
Publication of JPS581277A publication Critical patent/JPS581277A/en
Pending legal-status Critical Current

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V30/00Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
    • G06V30/10Character recognition
    • G06V30/18Extraction of features or characteristics of the image
    • G06V30/1801Detecting partial patterns, e.g. edges or contours, or configurations, e.g. loops, corners, strokes or intersections
    • G06V30/18076Detecting partial patterns, e.g. edges or contours, or configurations, e.g. loops, corners, strokes or intersections by analysing connectivity, e.g. edge linking, connected component analysis or slices
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V30/00Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
    • G06V30/10Character recognition

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Theoretical Computer Science (AREA)
  • Character Discrimination (AREA)

Abstract

PURPOSE:To pick up the characteristic of profile of a character and to surely discriminate similar characters with similar characteristics, by taking difference of interval from a character frame to a character line at each adjacent scanning lines and obtaining a characteristic pattern at each scanning line with the difference. CONSTITUTION:A video signal formed at a video signal forming section 1 is scanned and a profile characteristic is picked up at a profile pickup section 2 to obtain a profile characteristic element train. The train is smoothed at a smoothing processing 3 to obtain a profile characteristic pattern. The character profile characteristic element at each scanning line of the pattern is stored in a profile characteristic data storage memory 7 of a differential circuit section 4 as the profile characteristic data. The address of the memory 7 is instructed with a counter 6 in this state, the differential value at each scanning line is obtained with the 1st and 2nd FFs 8 and 9 and a subtraction circuit 10 and stored in a sequential differential characteristic data storage memory 11. The output of the section 4 is inputted to a normalizing circuit 5, the profile characteristic of the character is picked up, and similar characters with similar characteristics can surely be discriminated.

Description

【発明の詳細な説明】 本発明は、文字特徴抽出方式に関するものであって、特
に例えば手書き文字の輪郭特徴を抽出することによ)辞
書ファイルと照合して文字認識をする場合に、隣接走査
線毎に文字枠から文字線までの距離の差分なとシ、これ
を各走査線について求めることによ)文字の翰郭鯖の特
徴を抽出しようとするものに関する。
DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a character feature extraction method, and in particular, when performing character recognition by comparing with a dictionary file (for example, by extracting outline features of handwritten characters), adjacent scanning This method is concerned with extracting the characteristic features of characters (by calculating the difference in the distance from the character frame to the character line for each scanning line).

例えば原稿などに記載された手書き文字を辞書ファイル
と照合して文学認識を行うときの特徴抽出方式には、文
字の複雑さを表現する線密度抽出方式へ文字の方向を表
現する文字コード抽出方式、文字の背景部分を表現する
面特徴抽出方式などがあるが、これらはみな文字全体か
らみ九S徴、すなわち大域的特徴を表現するものである
。そのためこれらの方式によ)抽出され九各特徴のみで
は大域的%黴の類似している類似文字を識別することは
難しい。そこで、特徴抽出の一つとして局所的特徴を反
映させることKより類似文字を識・別する文字輪郭特徴
抽出方式が使用されている。この文字輪郭抽出方式は、
第1図に示すように先ずビデオ信号作成部1にて手書き
文字を読取ることKよシ得られ九第2図(イ)に示す入
力ビデオ信号1−0を記憶し、これを輪郭抽山部2にお
いて第2図(イ)に矢印で示すようK例えば左から走査
して輪郭抽出を行う、この輪郭抽出とは、走査線a4が
文字線と最初にクロスするまでの距離L4を求めるもの
である。このように第2図(イ)に示す入力ビデオ信号
1−0を走査IMae−Nにより走査することによシ、
第2図(ロ)に示すように輪郭特I[要素Lo%L、が
求められる。なお文字線とクロスしない場合には右端文
字枠までの距離が輪郭特徴要素となる。このようにして
第2図(ロ)に示す輪郭特徴要素列が求められる。そし
てこれを、ノイズなどを除去するために平滑化処理部3
で平滑化する。これによ)上記輪郭特徴要素列は第2図
(/→に示す如き特徴パターンとなる。ところが、この
ような従来の方式では類似文字に関する識別が不充分と
なることがある。例えば第3図(イ)′に示す文字「墓
」は輪郭抽出によシ同図(口rの輪郭特徴ヤ素列となシ
、これを平滑化すると、同図(ハ)′に示す特徴パター
ンとなる。これによシ明らかなように、第2図(イ)K
示す文字「基」の同図(ハ)の特徴パターンと第3図(
イ)に示す文字「墓」の同図f→′の特徴パターンはほ
とんど一致し、両者を識別できない。
For example, when performing literary recognition by comparing handwritten characters written on a manuscript with a dictionary file, the feature extraction methods include the linear density extraction method, which expresses the complexity of the characters, and the character code extraction method, which expresses the direction of the characters. , a surface feature extraction method that expresses the background part of a character, etc., but all of these methods express the nine S features, that is, global features, from the entire character. Therefore, it is difficult to identify similar characters that are globally similar using only the nine features extracted using these methods. Therefore, as one feature extraction method, a character contour feature extraction method is used that identifies and distinguishes similar characters by reflecting local features. This character outline extraction method is
As shown in FIG. 1, the input video signal 1-0 obtained by reading handwritten characters in the video signal generating section 1 and shown in FIG. 2, perform contour extraction by scanning from the left, for example, as shown by the arrow in FIG. be. In this way, by scanning the input video signal 1-0 shown in FIG.
As shown in FIG. 2(b), the contour characteristic I [element Lo%L] is obtained. Note that when the character line does not cross the character line, the distance to the rightmost character frame becomes the contour feature element. In this way, the contour feature element sequence shown in FIG. 2(b) is obtained. This is then processed by a smoothing processing unit 3 to remove noise etc.
Smooth with . As a result, the outline feature element sequence becomes a feature pattern as shown in FIG. The character ``tomb'' shown in (a)' is extracted by contour extraction and becomes the contour feature of the mouth r in the same figure.If this is smoothed, it becomes the feature pattern shown in (c)' in the same figure. As is clear from this, Fig. 2 (a) K
The characteristic pattern of the character "Ki" shown in the same figure (c) and the characteristic pattern in figure 3 (
The characteristic patterns of the character "grave" shown in b) from f→' in the same figure almost match, and the two cannot be distinguished.

したがって本発明は、以上のように走査線が文字線とり
はスするまでの距離を各走査線について求める文字輪郭
抽出方式では文字の輪郭の局所的%徴は異るが大域的特
徴は類似する類似文字を識別できない問題点を改善する
ために、走査線毎に文字線とクロスするまでの距離を検
出し、1iiI接走査線毎に順次この検出した距離の差
分をとシ、各走査線についてこの差分の特徴パターンを
求めることによシ類似文字の識別に有効な4I徴を抽出
しようとするものである。その丸めに本発明の文学輪郭
特徴抽出方式は、文字の輪郭を抽出する文字輪郭抽出手
段と、輪郭特徴データが保持される輪郭特徴データ保持
手段と、上記翰郭特黴データ保持手段からデータを読出
す読出し手段と、差分作成回路を備え、上記文字輪郭抽
出手段によシ抽出された文字の輪郭特徴データを上記輪
郭特徴データ保持手段に格納し、この輪郭特徴データ保
持手段に格納された文字の輪郭特徴データを上記続出し
手段によシ順次絖出し、この続出された順に順次文字の
輪郭特徴データの差分を上記差分作成回路によシ作成す
ることを特徴とするものである。
Therefore, according to the present invention, in the character contour extraction method that calculates the distance until the scanning line separates from the character line for each scanning line as described above, the local percentage characteristics of the character contour are different, but the global characteristics are similar. In order to improve the problem of not being able to identify similar characters, the distance to which the character line crosses is detected for each scanning line, and the difference between the detected distances is calculated for each tangent scanning line, and for each scanning line. By finding the feature pattern of this difference, the 4I features that are effective for identifying similar characters are extracted. To round this off, the literary contour feature extraction method of the present invention includes a character contour extraction means for extracting the contours of characters, a contour feature data holding means for holding contour feature data, and data from the above-mentioned Hanguo special mold data holding means. The apparatus includes a readout means for reading and a difference creation circuit, and stores the outline feature data of the character extracted by the character outline extraction means in the outline feature data storage means, and stores the character outline feature data extracted by the character outline extraction means in the outline feature data storage means. The present invention is characterized in that the contour feature data of the characters are sequentially generated by the successive output means, and the differences of the contour characteristic data of the characters are generated by the difference generation circuit in the order in which they are successively generated.

本発明を一実施例にもとづき詳述するに先立ち本発明の
動作MK理を第5図および第6図にもとづき説明する。
Before explaining the present invention in detail based on one embodiment, the operation MK principle of the present invention will be explained based on FIGS. 5 and 6.

例えば第5図において、上記第1図および館2図で説明
したように文字「基」の入力ビデオ信号1−0から輪郭
抽出を行い、さらにこれを平滑化を施して第5図e→に
示す特徴パターンを求め、それから次のようkして差分
を求める。すなわち、走査11ia4+sの平滑化し九
輪郭特徴要素iA+1とその1つ先の走査線G4の平滑
化した輪郭特徴要素t4の差分’4+t−’4を求め、
これを各走査線60〜番、について求めて第5図に)K
示す差分特徴パターンを得る。これに正規化処理を行い
第5図(ホ)の差分特徴パターンを得る。このようにし
て求めた#15図(ホ)の正規化した差分特徴パターン
と同図(ハ)の平滑化し九輪郭特徴要素列とを比較する
と、後者の輪 □郭特徴l!累列における変化の大きい
部分はそれに対応して拡大されて前者の差分%徴パター
ンに表われることがわかる。同様にして第6図に示すよ
うに文字「墓」の入力ビデオ信号1−0’について同図
呵に示す差分特徴パター/、さらKは同図(ホ)′に示
す正規化し友差分特徴パターンを得ることができる。と
のようKして得られた文字「基」と「墓」との差分4I
微パタ一ン第5図に)と第6図に)′とを比較してみる
と、上記に)′のパターンには上記に)のパターンにな
いビークtが中央にあられれていることがわかる。これ
は115図(ロ)および第6図(ロ)′における輪郭特
徴要素列において文字「基」と「墓」の中間部の窪みの
有無に起因する文字輪郭特徴要素列LP−島とLS %
 L’Bのわずか表相道が山記したように差分をとるこ
とKよ〕第5図および第6図のに)とに)′に示される
如くはつきp区別できるように41微抽出が行われたこ
とにもとづく吃のである。そして第5図(ホ)と第6図
(ホ)′に示されるように正規化処理されることによシ
ピークp′が明確に表わされ、この4I徴の相違が一層
明瞭になる。
For example, in FIG. 5, the outline is extracted from the input video signal 1-0 of the character "base" as explained in FIG. 1 and FIG. Find the characteristic pattern shown, and then calculate the difference by k as follows. That is, find the difference '4+t-'4 between the smoothed nine contour feature element iA+1 of the scan 11ia4+s and the smoothed contour feature element t4 of the next scanning line G4,
This is calculated for each scanning line number 60~ and shown in Figure 5)K
Obtain the differential feature pattern shown. This is subjected to normalization processing to obtain the differential feature pattern shown in FIG. 5 (E). Comparing the normalized differential feature pattern in Figure #15 (E) obtained in this way with the smoothed nine contour feature element sequence in Figure (C), we find that the latter's contour □Contour feature l! It can be seen that portions with large changes in the series are correspondingly enlarged and appear in the former difference percentage pattern. Similarly, as shown in FIG. 6, for the input video signal 1-0' of the character "grave", the differential feature pattern shown in FIG. can be obtained. Difference between the characters ``Ki'' and ``Tomb'' obtained by ``K'' 4I
Comparing the fine patterns shown in Figure 5) and Figure 6)', it can be seen that the pattern in )' has a beak t in the center that is not present in the pattern in ). Recognize. This is due to the presence or absence of a depression in the middle of the characters "base" and "tomb" in the outline feature element strings in Figures 115 (b) and 6 (b)'.
As shown in Figures 5 and 6, 41 micro-extractions are obtained to be able to distinguish between p as shown in Figures 5 and 6. The stuttering is based on what has been done. Then, as shown in FIGS. 5(e) and 6(e)', the peak p' is clearly expressed by the normalization process, and the difference in the 4I characteristics becomes even clearer.

次に本発明の一実施例を第4図および第7図にもとづき
第5図を参照しっつ四明する。
Next, an embodiment of the present invention will be explained based on FIGS. 4 and 7 with reference to FIG. 5.

第4図は本実施例の構成の概略図、第7図はその差分回
路部4の詳細図である。
FIG. 4 is a schematic diagram of the configuration of this embodiment, and FIG. 7 is a detailed diagram of the differential circuit section 4. In FIG.

図中、他図と同符号部は同一構成部分を示すものであっ
て、4は差分回路部、5は正規化回路部、6けカウンタ
、7は輪郭特徴データ格納メモリ、8け第1フリツプ7
0ツブ、9は第2フリツプ70ツブ、10は減算回路、
11は差分特徴データ格納メモリである。
In the figure, the same reference numerals as in other figures indicate the same components, and 4 is a differential circuit section, 5 is a normalization circuit section, 6-digit counter, 7 is a contour feature data storage memory, and 8-digit first flip. 7
0 block, 9 is the second flip 70 block, 10 is the subtraction circuit,
11 is a differential feature data storage memory.

差分回路部4tlj:、平滑化処理部3で平滑化された
各走査線毎の輪郭特徴要素について、上記の如く、その
隣接した輪郭特徴要素との差分を作成するものであって
、第7図に示す如く、カウンタ6、輪郭特徴データ格納
メモリ7、第17リツプ7crツブ8、第2フリツプフ
ロツプ9、減算回路10゜差分特徴データ格納メモリ1
1等から構成されている― ここでカウンタ6は、輪郭特徴データ格納メモリ7およ
び差分特徴データ格納メモリ11のアドレスを指示し、
輪郭特徴データ格納メモリ7に格納されているデータを
読出すとともに差分q+徴データ格納メモリ11に輪郭
特徴データの差分データを書込めるようにしたものであ
る。
Difference circuit section 4tlj: For the contour feature element of each scanning line smoothed by the smoothing processing section 3, it creates a difference between the contour feature element and the adjacent contour feature element as described above. As shown in the figure, a counter 6, a contour feature data storage memory 7, a 17th lip 7CR tube 8, a second flip-flop 9, a subtraction circuit 10, and a differential feature data storage memory 1.
Here, the counter 6 indicates the addresses of the contour feature data storage memory 7 and the difference feature data storage memory 11,
The data stored in the contour feature data storage memory 7 can be read out, and the difference data of the contour feature data can be written into the difference q+characteristic data storage memory 11.

輪郭特徴データ格納メモリ7/Ii、第5図ヒjに示す
各走査線毎の文字枠から文字IMtでの距離の平滑化し
た輪郭特徴要素を表わす輪郭特徴データをあらかじめ格
納しているものである。
Contour feature data storage memory 7/Ii stores in advance contour feature data representing smoothed contour feature elements of the distance from the character frame to the character IMt for each scanning line shown in FIG. .

第17リツプフロツプ回路8は、上記差分データを作る
ときの被減算値となる輪郭特徴データを保持するもので
あって、輪郭特徴データ格納メモリ7から読出された特
徴データを保持する。そして1クロツク遅れてこの輪郭
特徴データを出力し、このデータを第27リツプフロツ
プ9および減算回路10に伝達するものである。
The seventeenth lip-flop circuit 8 holds the contour feature data that becomes the subtracted value when creating the difference data, and holds the feature data read out from the contour feature data storage memory 7. Then, this contour feature data is output with a delay of one clock, and this data is transmitted to the 27th lip-flop 9 and the subtraction circuit 10.

第2フリツプ70ツブ9は、上記差分データを作るとき
の減算値となる輪郭特徴データを保持するデータ保持手
段であって、第17リツプ70ツブ8よ多出力された輪
郭特徴データを保持し、これをさらに1クロツク遅れて
出力し、減算回路10に伝達する庵のである。
The second flip 70 tube 9 is a data holding means for holding contour feature data that is a subtraction value when creating the difference data, and holds the contour feature data output more than the 17th flip 70 tube 8, This signal is further delayed by one clock and is outputted and transmitted to the subtraction circuit 10.

減算回路10は、上記差分データを作成する回路であっ
て、第1フリツプ70ツブ8から伝達された輪郭%砿デ
ータから興2ノリップフロッ”プ9から伝達された輪郭
特徴データの差を作成するものである。
The subtraction circuit 10 is a circuit that creates the difference data, and creates the difference between the contour data transmitted from the first flip flop 70 and the contour feature data transmitted from the second flip flop 9. It is.

差分%徽データ格納メモリ11は、上記減算回路10で
作成した差分データを上記カウンタ6からのアドレスの
指示にしたがって格納するものである。
The difference percentage data storage memory 11 stores the difference data created by the subtraction circuit 10 in accordance with the address instruction from the counter 6.

正規化回路部5は、上記差分特徴データ格納メモリ11
に格納されている部分データを辞書ファイルと照合する
ときに照合し易くするために正規化するものである。
The normalization circuit unit 5 includes the differential feature data storage memory 11
This normalizes the partial data stored in the dictionary file to make it easier to compare it with the dictionary file.

次に餓4図および第7図に示す本実施例の動作を説明す
る。
Next, the operation of this embodiment shown in FIGS. 4 and 7 will be explained.

ビデオ信号作成部lにて作成され九ビデオ信号1−0を
第5図(イ)に示すように左から走査して輪郭舶山部2
で輪郭%微を抽出し、N図(ロ)に示す輪郭特徴要素列
を求める。さらkこれを平滑化処理部3で平滑化を行い
同図(ハ)の輪郭特徴パターンを求める。そしてこの輪
郭特徴パターンの各走査線毎の文字輪郭特徴I!素を、
輪郭特徴データとして予め輪郭特徴データ格納メモリ7
・に格納しておく。
The nine video signals 1-0 created by the video signal creation unit 1 are scanned from the left as shown in FIG.
The contour percentage is extracted and the contour feature element sequence shown in Figure N (b) is obtained. Furthermore, this is smoothed by the smoothing processing section 3 to obtain the contour feature pattern shown in FIG. And the character contour feature I for each scanning line of this contour feature pattern! The basics,
Contour feature data storage memory 7 in advance as contour feature data
・Store it in .

このような状態でカウンタ6を19TART伯号によジ
スタートさせること、輪郭4I微データ格納メモリ7の
アドレスが指示されてずでに格納ずみの第5図(ハ)に
示す走査線aoに対する輪郭特徴データJ・が続出され
、このデータ1・が第17リツプフロツプ8に伝達され
る。ついで次のクロック信号によシデータL・は第2フ
リツプフロツプ9および減算回路10に伝達され、同時
にカウンタ6から次のアドレスを指示されて26目の走
査11Aasに対する輪郭特徴データjlが続出され、
これは第1フリツプフロツプ8に伝達されこむに保持さ
れる。またこれと同時にこの段階ではデータを保持して
いない第27リツプフロツ/9から「0」が減算回路l
OK伝達される。減算回路10では第17リツプフロツ
プ8から伝達された上記データJ−とiK2フリップフ
ロップ9から伝達され九「0」とによシ1・−〇の差分
値が求められる。そしてさらに次のクロック信号によシ
この差分値1・−0はカウンタ6によシ指示されたアド
レスにし九がって差分特徴データメモリIIK格納され
る。これと同時に上記と同様に第27リツプフロツプ9
に保持された上記データLOは減算回路10に伝達され
、第17リツプフロツプ8に保持された上記データーL
1は第2フリツプフロツプ8および減算回路10に伝達
され、減算回路lOでは今度は1.1−1oの差分値が
求められる。
In such a state, the counter 6 is restarted by the number 19 TART, and the contour 4I for the scanning line ao shown in FIG. Characteristic data J. is successively output, and this data 1. is transmitted to the seventeenth lip-flop 8. Then, in response to the next clock signal, the data L. is transmitted to the second flip-flop 9 and the subtraction circuit 10, and at the same time, the next address is instructed from the counter 6, and the contour feature data jl for the 26th scan 11Aas is successively output.
This is transmitted to the first flip-flop 8 and held there. At the same time, "0" is subtracted from the 27th lip float/9, which does not hold data at this stage.
OK is communicated. The subtraction circuit 10 calculates the difference between the data J- transmitted from the 17th flip-flop 8, 9 "0" transmitted from the iK2 flip-flop 9, and 1.-0. Then, in response to the next clock signal, the difference values 1 and -0 are stored in the difference characteristic data memory IIK according to the address specified by the counter 6. At the same time, the 27th lip-flop 9
The data LO held in the 17th lip-flop 8 is transmitted to the subtraction circuit 10, and
1 is transmitted to the second flip-flop 8 and the subtraction circuit 10, and the subtraction circuit 10 then calculates the difference value of 1.1-1o.

このとき同時に第1アリツブフロツプ8にはカウンタ6
によシ3番目のアドレスを指示されて続出された走査線
a意に対する輪郭特徴データ!意が保持されている。以
下同様にしてカウンタ6により読出され九アドレスにし
九がって走査線a2〜65に対する輪郭4I黴データt
、J、、が第17リツプフロツプ8および1クロツク遅
れて第27リツプフロツプ9に順次保持され、これら第
1フリツプ70ツブ8および第2フリツプフロツプ9か
ら伝達され九−接データ例えばJ4+1およびt4の差
分値JASs−14が減算回路10で求められ、この差
分値がカウンタ6によシ指示されたアドレスにしたがっ
て順次差分特徴データ格納メモ!jllK格納される。
At this time, at the same time, the counter 6 is displayed on the first arrival flop 8.
Contour feature data for the scanning line a that was successively produced when the third address was specified! intention is maintained. Thereafter, the contour 4I mold data t for the scanning lines a2 to 65 is read out by the counter 6 in the same way, and according to the 9th address.
. -14 is obtained by the subtraction circuit 10, and this difference value is sequentially stored in the difference characteristic data memo! according to the address specified by the counter 6. jllK is stored.

このようkして求められた差分特徴データは正規化回路
5によシ正規化され辞書ファイルと照合されて文字1g
縁が行われる。
The differential feature data obtained in this way is normalized by the normalization circuit 5, and compared with the dictionary file to obtain the character 1g.
The relationship is done.

以上説明した如く、結局本発明によれは、ii4接走前
走査線毎字枠から文字線までの距隨の差分をとり、これ
を各走査@i/Cついて求めることにより文字の輪郭%
黴を抽出するようにしたので、文字の輪郭の局所的4I
徴を艮〈抽出することができ、そのため文字の輪郭の局
所的特徴が異るが大域的咎微の@似する類似文字を確実
に置割することができる。
As explained above, according to the present invention, the difference in the distance from the character frame to the character line is calculated for each scanning line before ii4 tangent scanning, and this is calculated for each scan @i/C, thereby calculating the character outline%.
Since mold was extracted, the local 4I of the outline of the character
It is possible to extract the characteristics, and therefore it is possible to reliably place similar characters that have different local features of the outline of the characters but have slight global similarities.

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

嬉1図は従来の文字、輪郭特徴抽出方式の概略欽明図、
第2図および第8図はその具体例の原3!lI観明図、
第4図は本発明の一実施例の構成の概略図、第5図およ
び第6#Aはその具体例の原理説明図、第7図はその差
分回路図である。 図中、1はビデオ信号作成部、2は輪郭抽出処理部、3
は平滑処理部、4は差分回路部、5は正規化回路部、6
はカウンタ、7は輪郭特徴データ格納メモリ、8は第1
フリツプフロツプ、9は第2フリツプフロツプ、10は
減算回路、11は差分特徴データ格納メモリをそれぞれ
示す。 特許出願人  冨士通株式会社 代理人弁理士  山 谷 晧 榮
Figure 1 is a schematic diagram of Kinmei using the conventional character and contour feature extraction method.
Figures 2 and 8 are examples of the original 3! lI sightseeing map,
FIG. 4 is a schematic diagram of the configuration of an embodiment of the present invention, FIGS. 5 and 6#A are diagrams explaining the principle of the specific example, and FIG. 7 is a differential circuit diagram thereof. In the figure, 1 is a video signal generation section, 2 is a contour extraction processing section, and 3
is a smoothing processing section, 4 is a difference circuit section, 5 is a normalization circuit section, 6
is a counter, 7 is a contour feature data storage memory, and 8 is a first
9 is a flip-flop, 9 is a second flip-flop, 10 is a subtraction circuit, and 11 is a differential feature data storage memory. Patent applicant Fujitsu Co., Ltd. Representative patent attorney Akira Yamatani

Claims (1)

【特許請求の範囲】[Claims] (1)  文字の輪郭を抽出する文字輪郭抽出手段と、
輪郭特徴データが保持される輪郭特徴データ保持手段と
、上記輪郭特徴データ保持手段からデータを読出す続出
し手段と、差分作成回路を備え、上記文字輪郭抽出手段
によシ抽出された文字の輪郭特徴データを上記輪郭特徴
データ保持手段に格納し、この輪郭特徴データ保持手段
に格納された文字の輪郭特徴データを上記読出し手段に
より順次続出し、この読出された順に順次文字の輪廓特
徴データの差分を上記差分作成回路によシ作成すること
を特徴とする文字・$特徴抽出方式。
(1) Character contour extraction means for extracting the contours of characters;
A contour feature data holding means for holding contour feature data, a continuation means for reading out data from the contour feature data holding means, and a difference creation circuit are provided, and the character outline extracted by the character outline extraction means is provided. The feature data is stored in the outline feature data holding means, the outline feature data of the characters stored in the outline feature data holding means are sequentially read out by the reading means, and the differences in the outline feature data of the characters are sequentially read out in the order in which they are read out. A character/$ feature extraction method characterized in that the character/$ feature extraction method is created using the above-mentioned difference creation circuit.
JP56099931A 1981-06-27 1981-06-27 Character characteristic pickup system Pending JPS581277A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP56099931A JPS581277A (en) 1981-06-27 1981-06-27 Character characteristic pickup system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
JP56099931A JPS581277A (en) 1981-06-27 1981-06-27 Character characteristic pickup system

Publications (1)

Publication Number Publication Date
JPS581277A true JPS581277A (en) 1983-01-06

Family

ID=14260478

Family Applications (1)

Application Number Title Priority Date Filing Date
JP56099931A Pending JPS581277A (en) 1981-06-27 1981-06-27 Character characteristic pickup system

Country Status (1)

Country Link
JP (1) JPS581277A (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPS62177687A (en) * 1986-01-30 1987-08-04 Nec Corp Inclination detecting circuit for pattern outline
JPS63191282A (en) * 1987-02-03 1988-08-08 Nippon Denshi Kiki Kk Non-contact personal identification system
JPS63254574A (en) * 1987-04-10 1988-10-21 Nippon Denshi Kiki Kk Profile feature extraction system for face matching device

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPS62177687A (en) * 1986-01-30 1987-08-04 Nec Corp Inclination detecting circuit for pattern outline
JPS63191282A (en) * 1987-02-03 1988-08-08 Nippon Denshi Kiki Kk Non-contact personal identification system
JPS63254574A (en) * 1987-04-10 1988-10-21 Nippon Denshi Kiki Kk Profile feature extraction system for face matching device

Similar Documents

Publication Publication Date Title
JPH055146B2 (en)
JPH0467234B2 (en)
JPS6140684A (en) Contour tracking device
JPS581277A (en) Character characteristic pickup system
JP2000029983A (en) Document reader device
JPS58197581A (en) Method and device for recognizing character and figure
JP3090070B2 (en) Form identification method and device
JP2506071B2 (en) Contour tracking device
JP2670372B2 (en) Binarization method
JPS6254380A (en) Character recognizing device
JP2894111B2 (en) Comprehensive judgment method of recognition result in optical type character recognition device
JP2902694B2 (en) Optical character reader
JPH01119885A (en) Document reader
JPH07104940B2 (en) Figure recognition device
JPS625480A (en) Edge detection system
JPS63182780A (en) Picture processing method for drawing reading device
JPH01304497A (en) Method and device for score recognition
JPH0762873B2 (en) Contour tracking device
JP2643092B2 (en) Method and system for processing non-standard data located outside predefined fields on a document form
CN117315707A (en) Digital structured processing method, system, medium and equipment for track drawing
JPH0773273A (en) Pattern segmenting and recognizing method and its system
JPH11328399A (en) Figure extracting method and recording medium where program thereof is recorded
JPH07325921A (en) Device and method for extracting connected area
JPH0746363B2 (en) Drawing reader
JPH0461396B2 (en)