JPS58195987A - Symbol and character detector - Google Patents

Symbol and character detector

Info

Publication number
JPS58195987A
JPS58195987A JP57078536A JP7853682A JPS58195987A JP S58195987 A JPS58195987 A JP S58195987A JP 57078536 A JP57078536 A JP 57078536A JP 7853682 A JP7853682 A JP 7853682A JP S58195987 A JPS58195987 A JP S58195987A
Authority
JP
Japan
Prior art keywords
points
small area
feature
feature points
point
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
JP57078536A
Other languages
Japanese (ja)
Inventor
Takashi Tsunekawa
尚 恒川
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.)
Toshiba Corp
Original Assignee
Toshiba Corp
Tokyo Shibaura 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 Toshiba Corp, Tokyo Shibaura Electric Co Ltd filed Critical Toshiba Corp
Priority to JP57078536A priority Critical patent/JPS58195987A/en
Publication of JPS58195987A publication Critical patent/JPS58195987A/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
    • G06V30/18019Detecting partial patterns, e.g. edges or contours, or configurations, e.g. loops, corners, strokes or intersections by matching or filtering
    • 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)
  • Image Analysis (AREA)

Abstract

PURPOSE:To detect symbol and character through adequate and simple processing by extracting feature point of line picture constituting a figure and detecting the existence of a symbol or character in a small area from the number of feature point of the line picture in the small area. CONSTITUTION:Each segment of a drawing pattern stored in a one-frame memory is made into a thin line and feature point of the line picture is extracted. Information on the feature points are end point, flection point, branch point, etc., of line picture shown by (a)-(f) and pieces of information on those feature points are stored as data 1 in a memory 1 successively corresponding to positions. Data 0 is given at positions other than the feature points. Thus, feature data are written at respective address locations (i) and (j) of the memory 1 corresponding to input figures MxN to form a feature picture.

Description

【発明の詳細な説明】 〔発明の技術分野〕 本発明は図面中のシンがルマータや文字を適確に検出す
ることのできるシンがル・文字検出装置に関する。
DETAILED DESCRIPTION OF THE INVENTION [Technical Field of the Invention] The present invention relates to a character detection device capable of accurately detecting lumata and characters in drawings.

〔発明の技術的背景とその問題点1 図面情報を計算機システムに7フイルし、図面編集等を
行わんとする場合、上記図面中の各徨シンノルやこれら
のシンゲルを結ぶ結合線分およびこれらに付された文字
をそれぞれ別個に識別して入力することが重要となる。
[Technical Background of the Invention and its Problems 1] When you want to upload drawing information to a computer system and edit the drawing, you can use It is important to identify and input each of the attached characters separately.

つ★〕、図面を構成する細索をその機能毎に分類するこ
とによって、必要とする図面情報を適確に入力すること
が可能となる。そこで従来一般的には、一分の長さから
シンールとその結合線とを識別し、シンールや文字等を
独立に検出する仁とが打われている。仁れは通常の図面
にあっては、接ivCを表わす一分が比較的長く、ま友
シンがルや文字を構成する一分が比較的短いと首う一般
的性質を利用して行われるものである。ところが、1閣
の種類によりては、例えば第1図(a)に示すように、
シンIルを構成する一分に比して、その振続−の力が短
い仁ともしにしば存在し、上配線分艮だけでその識別を
行うには無理があった。これ故、図面中のシンールや文
字を適確に検出することが困難であった。を友、このよ
うにして線分長を求める為には、着目し九−分を辿ると
言う処理が必要とな9、高速処理に同かないという間−
もめった。
★] By classifying the cords that make up a drawing according to their functions, it becomes possible to accurately input the necessary drawing information. Conventionally, therefore, it has been common practice to identify lines and their connecting lines based on the length of one minute, and to detect lines, characters, etc. independently. Nire is done by taking advantage of the general property that in normal drawings, the one minute representing the conjunct ivC is relatively long, and the one minute forming the letters and letters is relatively short. It is something. However, depending on the type of cabinet, for example, as shown in Figure 1 (a),
Compared to the one minute that makes up the Shin Iru, there are some that have a short duration of power, and it is impossible to identify them just by analyzing the upper wiring. Therefore, it has been difficult to accurately detect symbols and characters in drawings. In order to find the line segment length in this way, it is necessary to focus on the line segment length and trace the 9-segment.9, which is not equivalent to high-speed processing.
I struggled.

一万、文字検出等にあって線、その狐立性を調べること
が行われている。然し乍ら@1図(klに示すよう(文
字/量ターンが一分に近縁して記されている場合等、そ
の檄文状態を確実に検出することが非富に困難であると
舊う問題があった。
10,000 times, in character detection, etc., lines and their erect nature are investigated. However, as shown in Figure @1 (kl), there is a problem that it is extremely difficult to reliably detect the state of the invocation, such as when characters/quantity turns are written closely related to minutes. there were.

〔発明の目的〕[Purpose of the invention]

本発明はこのようなφ情を考慮してなされ喪もので、そ
の目的とするところは、図面中のシンダルマークや文字
を結合−分と明確に区別して適確に検出することのでき
る実用性の為いシンプル・文字検出装置を提供すること
にある。
The present invention has been made in consideration of such φ circumstances, and its purpose is to provide practicality in which it is possible to accurately detect cinder marks and characters in drawings by clearly distinguishing them from combinations. The purpose of this invention is to provide a simple character detection device for.

〔発明の概費〕[Outline of invention cost]

本発明i!図面を構成する線素の端点、分岐点、屈曲点
等の森−%像点を抽出し、小領域における上配解画特徴
点の数から上記小領域におするシンIル・文字の存在の
有無を検出するようにL5たものである。
This invention i! Extract image points such as end points, branching points, and bending points of line elements that make up the drawing, and determine the existence of symbols and characters in the small area based on the number of feature points in the superimposed image in the small area. L5 is designed to detect the presence or absence of the .

〔発明の効果゛〕′ 従って本兄明によれば、図面を構成する線分長に代えて
、所定の小領域におけるlIi!l!ii%像点数よシ
シンlル・文字の検出を行うので、高速に、且つ精度良
く上記検出を行い得る。しかも、結合線がシンプル・文
字の線分長よシ短い複雑な図面であっても、上記シンー
ル・文字を適確に、且つ簡易な処理によって検出できる
ので、その実用的利点は絶大である。
[Effect of the Invention]' Therefore, according to the present invention, instead of the length of the line segments composing the drawing, the lIi! l! Since letters and characters are detected based on the number of image points, the above detection can be performed at high speed and with high precision. Furthermore, even in complex drawings where the connecting lines are shorter than the line segment lengths of simple characters, the simple characters can be detected accurately and through simple processing, so the practical advantage is enormous.

〔発明の実施例〕[Embodiments of the invention]

以下、図vIJを参照して本発明の一実施例につき説明
する。
An embodiment of the present invention will be described below with reference to Figure vIJ.

凧2図#′i冥施例装置の概略構成図であシ、第3図は
その処理フローを示す図である。入力処理に供される図
面は、例えに所定の用4iK描かれたものであって、I
TV+CCDセンサ等の光電に換装W(図示せず)を用
いて撮儂入力され、2値化処理されたのち1フレームメ
モリに順次格納される。このlフレー五メモリに格納さ
れ九図面・母ターンの格線分に対して細線化処理が施さ
れ、その113!lI!t1%値点が抽出される。この
−画特徴点は、−例えば第4図(a)〜(f)に示すよ
う1こ、線画の端点や屈曲点、分岐点等の特徴からなる
もので、その特徴点の情報は、例えばr−タ″″l”と
してメモリ11に順次位置対応して記憶される。尚、上
記特徴点以外の位置には、f−タ@″0”が与えられる
。このようにして、入力図III(MXN)に相尚した
メモリ11の各アドレス位置(trj、)には、上記特
徴データがそれぞれ書き込まれ、ここに特徴画倫が形成
されることになる。
FIG. 2 is a schematic configuration diagram of the kite implementation device, and FIG. 3 is a diagram showing its processing flow. The drawings used for input processing are, for example, drawn in 4iK for a predetermined purpose.
Images are input to a photoelectric device such as a TV+CCD sensor using a replacement W (not shown), and after being binarized, they are sequentially stored in one frame memory. Thinning processing is applied to the grid line segments of the nine drawings and mother turns stored in the memory of this l frame, and the 113! lI! The t1% value point is extracted. These drawing feature points consist of features such as end points, bending points, and branching points of the line drawing, as shown in FIGS. 4(a) to 4(f), for example. r-data ""l" are stored in the memory 11 in sequential positional correspondence. Note that f-ta@"0" is given to positions other than the above-mentioned feature points. In this way, the input diagram III The above feature data is written in each address position (trj,) of the memory 11 corresponding to (MXN), and a feature image is formed here.

しかして、このメモリ11に対して、アドレスコントロ
ーラ12に19、(mXm)jli素からなる小領域1
1が設定される。この小領域13の大きさは、図面に描
かれるシンがルマーりや文字の大きさ一応じて定めるも
のであり、該小領域13は上記メモリJJO全領域に対
して走査されるようになっている。この小領域13のメ
毫り11に対する走査位置指定は、例えば上記小領域1
3の左上端アドレスによって示されるようになっている
Therefore, for this memory 11, the address controller 12 has a small area 1 consisting of 19, (mXm)jli elements.
1 is set. The size of this small area 13 is determined depending on the outline and the size of characters drawn in the drawing, and this small area 13 is designed to scan the entire area of the memory JJO. . The scanning position designation for the image 11 of this small area 13 is, for example,
It is indicated by the upper left end address of 3.

このようにしてメ毫す11を走査される小領域13にお
ける前記特徴画偉の情報は計数回路14に順次続出され
、(mXm)なる小領域13における前記線画特徴点の
数が計数される。
Information on the feature points in the small area 13 scanned in this way is sequentially sent to the counting circuit 14, and the number of line drawing feature points in the small area 13 (mXm) is counted.

そして、この計数され良縁−特値点の数llfは比[器
JjK4がれ、B値しジスタ1−に設定されたm値−と
大小比較されている。この比較により mf≧− なるとき、販小領域11に存在する図形)譬ターンが/
/dtルマークあるい社文字であると判定されるーそし
て、その判定結果は、シンがル位置に2僧回路17に供
給され、上記小領域JJのメモリ11に対する画面位置
に対応して記憶される。以上の処理がメモ17 J J
の全m11m領域に亘って走査される各小領域13毎に
それぞれ行われる。
Then, the number of counted special value points is compared in magnitude with the m value set in the B value register 1. As a result of this comparison, when mf≧-, the figure existing in the small area 11) is /
/dt is determined to be a mark or a company character - and the result of the determination is supplied to the second circuit 17 at the position where it is written, and is stored in correspondence with the screen position of the above-mentioned small area JJ in the memory 11. . The above process is memo 17 J J
This is performed for each small area 13 that is scanned over the entire m11m area.

ところで、−画物徴点は先に簡単に説明したように、線
素の端点や屈曲点、分岐点等によって表わされる。即ち
、第4図(a)に示す4分岐、同図(blに示す3分岐
、同図(@)に示す屈曲、更には同図(dlに示す真直
、同図(・)に示す端点、そして同図(f)に示す孤立
等として表わされる。このような−画特徴点は、図面を
形成する線素・fターンの基本的費累を為すもので、こ
のような特徴点はシンがル1−りや文字において通常集
中的に多く存在する。しかして、成る小領域IJについ
て着目すれば、シンがル十文字が存在する小領域にあっ
ては上記特命点が多く存在することになる。逆にシンが
ル十文字が含まれない場合や、その一部分のみが含まれ
るような場合には、上に2%像点の数が比較的少なくな
ると盲使って、成る小領域13における特徴点の数を計
数し、その数11fと、適正設定され九閾値−とを比較
すれば、咳小領域IJにシンプル・文字が含まれるか否
かを明確に識別することが可能となる。
By the way, as briefly explained above, -picture feature points are represented by end points, bending points, branching points, etc. of line elements. That is, the four branches shown in FIG. Then, it is expressed as an isolated figure as shown in (f) of the same figure.Such - image feature points form the basic accumulation of line elements and f-turns that form the drawing, and such feature points are Usually, there are many concentrated points in R1-Riya characters.However, if we pay attention to the small area IJ consisting of R1 and Riya characters, we will find that there are many special points mentioned above in the small area where Shin-Ru Jumonji exists. On the other hand, if the thin line does not include the Jumonji, or if only a part of it is included, the number of 2% image points on the top is relatively small, and the number of feature points in the small area 13 is blindly used. By counting the number and comparing the number 11f with a properly set nine threshold value, it becomes possible to clearly identify whether or not the cough small area IJ includes simple characters.

尚、関連したli!1画特徴点の抽出は、例えば画本対
応した第5−に示すマスクを用いて、各画亀r−夕の論
理処理を第6図に示す如く行えばよい。この%像点抽出
には、1点鳶の周囲8点に層目する8点連結方式と、そ
の上下左右04点に層目する4点連結方式とが与えられ
るが、4点遵結によってこれを行う場合には次のようl
k論理処理を行うようにすればよい。
In addition, related li! Extraction of one-stroke feature points can be carried out by performing logical processing for each picture book as shown in FIG. 6, using, for example, the mask shown in No. 5 corresponding to the picture book. For this % image point extraction, there is an 8-point connection method that connects 8 points around one point, and a 4-point connection method that connects 04 points above, below, left, and right. If you want to do this, do the following:
It is sufficient to perform k logical processing.

(4分岐) A−1・C−D−1 (3分岐) A−8・C−D冨1 ム曇B働C−D−1 ム・B−C−D=1 A・ト(@Dxl (真直)  ム・lφC−D−1 A−B−C−D=1 ()日4曲 )       A−B−C−D=1A−
B−C−D=1 A−IS−C−D=1 A ・ 1−C−D−1 (端点)  ム−B−C−D−1 ム ・ B−C−D−1 (孤立)   A−B−C−D=1 しかして、これらの%徴のうち、例えば第6図に示すa
iim回路にて、真Ilを除く5つの一形Iクターンを
検出対象とすることにより、その%像点の情報を効果的
に、且つ適確に検出してシンプル・文字の検出に供する
ことが可能となる。
(4 branches) A-1・C-D-1 (3 branches) A-8・C-D 1 Mu cloud B work C-D-1 Mu・B-C-D=1 A・To (@Dxl (Mahono) Mu・lφC-D-1 A-B-C-D=1 ()day 4 songs) A-B-C-D=1A-
B-C-D=1 A-IS-C-D=1 A ・ 1-C-D-1 (End point) Mu-B-C-D-1 Mu ・ B-C-D-1 (Isolated) A -B-C-D=1 However, among these percentage signs, for example, a shown in FIG.
By using the iim circuit to detect five monomorphic I patterns excluding true Il, it is possible to effectively and accurately detect the information of the % image point and use it for simple character detection. It becomes possible.

以上説明したように本発明によれは、小領域における#
M画画像像点数から上記小領域にシンゲルまたは文字が
含まれるか1否かを識別するので、そのシンプル・文字
検出が簡易で69、且つ適確である。しかも図形Δター
ンが複雑であっても、その検出を確実に行い得、従来方
式に比してその夾用的利点線極めて多い。しかもマスク
処理によって簡易に特命点を検出し、その叙ki!′を
叙したのち閾値と比較すれはよいので、尚連処理がoJ
舵てあり、且つ−・−ドウニア構成も簡単であるとどう
効果を奏する。
As explained above, according to the present invention, #
Since it is determined from the number of image points of the M image whether a singel or a character is included in the small area or not, the simple character detection is simple, 69, and accurate. Furthermore, even if the figure Δturn is complex, it can be detected reliably, and its advantages over conventional methods are extremely large. Moreover, the special mission point can be easily detected by mask processing, and the mission point can be easily detected! ′ and then compare it with the threshold value, so the Nao-ren processing is oJ
What effect will it have if it has a rudder and has a simple Dounia configuration?

向、本発明は上記実施例に限定されるものではない。例
えば%像点検出を(kXJ )の領域1Cついて行うよ
うにしてもよく、%徴の定義も仕体に応じて矩めればよ
いものである。また比11&+11M#についても、処
理対象とする図面に応じてxEjCIれはよい。賛する
に本発明はその簀旨を逸脱しない範囲で億々変形して実
施することができる。
However, the present invention is not limited to the above embodiments. For example, the % image point detection may be performed for the area 1C of (kXJ), and the definition of the % mark may also be rectified according to the type. Also, regarding the ratio 11&+11M#, the ratio of xEjCI is good depending on the drawing to be processed. Admittedly, the present invention can be modified and implemented in numerous ways without departing from its spirit.

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

第1図(a) (blは入力−向の例を示す図、第2図
Vよ本発明の実施例装置の概略構成図、第3図は実施例
装置の処理フローを示す図、wJ4図−)〜(f)&ユ
縁画特徴点の例を示す図、第5図は特似点抽出のマスク
例を示す図、第6図は籍像点抽出の−理回路を示す図で
ある。 11・・・メモリ、12・・・アドレス・コントローラ
、13・・・小領域、14・・・計数回路、Jj・・・
比軟器、16・・・kl櫨レしスタ、11・・・シンゲ
ル位置ルC:憧回路。 113m人代理人  弁理士 鈴 江 武 該第1図 (a) 第2図
Fig. 1(a) (bl is a diagram showing an example of the input direction, Fig. 2 V is a schematic configuration diagram of the embodiment device of the present invention, Fig. 3 is a diagram showing the processing flow of the embodiment device, wJ4 diagram Figure 5 is a diagram showing an example of a mask for extracting special points; Figure 6 is a diagram showing a logic circuit for extracting image points. . 11...Memory, 12...Address controller, 13...Small area, 14...Counting circuit, Jj...
Hi Souki, 16...kl Hajira Resta, 11...Singel position C: Adoring circuit. 113m agents Patent attorney Takeshi Suzue Figure 1 (a) Figure 2

Claims (2)

【特許請求の範囲】[Claims] (1)  用紙に描かれた図形を撮導入力してその図形
ノ譬ターンを記憶するノ々ターンメモリと、この記憶さ
れた図形/量ターンを構成する図形線分の線画特徴点を
それぞれ抽出する手段と、前記ノ9ターンメモリに対し
て設定し友小領域を前記ノ臂ターンメモリの全域に][
りて走査する手段と、上記小領域の各走査位置における
前記線画特徴点の数をそれぞれ計数する手段と、この計
数値と所定の閾値とを比較して前記小領域内のシンがル
・文字/ヤターンの存在を判定する子役とを具備したこ
とt−特徴とするシンール・文字検出装置、
(1) A non-turn memory that stores the parable turns of a figure drawn on a sheet of paper and extracts the line drawing feature points of the figure line segments that make up the stored figure/quantity turns. and a means for setting the above-mentioned 9-turn memory to set a friend small area in the entire area of the above-mentioned arm turn memory] [
a means for counting the number of line drawing feature points at each scanning position of the small area; and a means for counting the number of line drawing feature points at each scanning position of the small area; /A child actor for determining the presence of a character.
(2)図形線分の線画特徴点は、図形ノlターンの温点
、屈曲点、分岐点からなるものである特許請求の範囲第
1項記値のシンゲル・文字検出装置。
(2) The singel/character detection device according to claim 1, wherein the line drawing feature points of the graphic line segments are comprised of hot points, bending points, and branching points of the graphic nol turns.
JP57078536A 1982-05-11 1982-05-11 Symbol and character detector Pending JPS58195987A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP57078536A JPS58195987A (en) 1982-05-11 1982-05-11 Symbol and character detector

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
JP57078536A JPS58195987A (en) 1982-05-11 1982-05-11 Symbol and character detector

Publications (1)

Publication Number Publication Date
JPS58195987A true JPS58195987A (en) 1983-11-15

Family

ID=13664627

Family Applications (1)

Application Number Title Priority Date Filing Date
JP57078536A Pending JPS58195987A (en) 1982-05-11 1982-05-11 Symbol and character detector

Country Status (1)

Country Link
JP (1) JPS58195987A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPS61237181A (en) * 1985-04-12 1986-10-22 Sumitomo Electric Ind Ltd Optical reader
JPS61237172A (en) * 1985-04-15 1986-10-22 Mitsubishi Electric Corp Drawing registering system

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPS61237181A (en) * 1985-04-12 1986-10-22 Sumitomo Electric Ind Ltd Optical reader
JPH0679349B2 (en) * 1985-04-12 1994-10-05 住友電気工業株式会社 Optical reader
JPS61237172A (en) * 1985-04-15 1986-10-22 Mitsubishi Electric Corp Drawing registering system

Similar Documents

Publication Publication Date Title
CN104881665A (en) Chip character identification and verification method and apparatus
CN112749606A (en) Text positioning method and device
JPS58195987A (en) Symbol and character detector
JPH0338628B2 (en)
CN115860139A (en) Deep learning-based multi-scale ship target detection method
CN115035032A (en) Neural network training method, related method, device, terminal and storage medium
JPS58189783A (en) Pattern discriminator
JPH0260265B2 (en)
JP2503208B2 (en) Business card image processing method
CN114140391B (en) Method for realizing rapid detection of on-board display screen module based on machine vision
JPH04142676A (en) Binary image conversion method
Huo et al. Grid Row Parameter Identification Using Tensorflow Convolutional Neural Network
JP2870640B2 (en) Figure recognition method
JPH08137991A (en) Automatic registration method for identification display of electronic component
KR20170045701A (en) Method for reading graphical indicator, indicator structure and electronic apparatus thereof
JPS63115279A (en) Area extracting system
JPH0894335A (en) Image processing detector
JP2745763B2 (en) Wiring pattern inspection equipment
JP3395922B2 (en) Inspection method and inspection device
JPH0129643Y2 (en)
JPS62127987A (en) Method of checking printed board pattern
JPH04365045A (en) Reticle pattern inspecting device
CN115205575A (en) Circuit element labeling method, related method, equipment, terminal and storage medium
CN110728233A (en) Objective question template detection method based on component detection
JPH0769961B2 (en) Pattern inspection method