JPH0444187A - Character recognizing device - Google Patents
Character recognizing deviceInfo
- Publication number
- JPH0444187A JPH0444187A JP2152244A JP15224490A JPH0444187A JP H0444187 A JPH0444187 A JP H0444187A JP 2152244 A JP2152244 A JP 2152244A JP 15224490 A JP15224490 A JP 15224490A JP H0444187 A JPH0444187 A JP H0444187A
- Authority
- JP
- Japan
- Prior art keywords
- projection
- character
- characters
- projection value
- value
- 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
Links
- 238000010586 diagram Methods 0.000 description 6
- 238000000605 extraction Methods 0.000 description 2
- 238000000034 method Methods 0.000 description 2
- 238000000926 separation method Methods 0.000 description 2
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
Abstract
Description
【発明の詳細な説明】
（産業上の利用分野）
本発明は、文字認識装置に関し、特に英文イタリック体
等の斜体文字を切り出して確実に認識することができる
文字認識装置に関する。DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to a character recognition device, and more particularly to a character recognition device that can cut out and reliably recognize italic characters such as English italics.
（従来の技術）
従来の文字認識装置では、第５図に示すように斜体文字
でない通常の正体文字Ｅｌ、 ・・・を認識する場合
、行方向と直角に投影値を算出して得られたパターンＣ
Ｉ、 ・・・を用いることにより文字を認識していた
。(Prior art) As shown in Fig. 5, in conventional character recognition devices, when recognizing normal regular characters El, . pattern C
Characters were recognized by using I, ....
（発明が解決しようとする課題）
しかしながら、英文イタリック体等の斜体文字を読み取
ろうとする場合、従来の正体文字と同様に行方向と直角
の方向に投影したのでは隣あう文字が重なって投影され
てしまい、１文字ごとのパターンを抽出することが不可
能である。このため、斜体文字を文字認識装置により認
識させる場合は、第６図に示すように投影方向を斜体文
字Ｆｌ。(Problem to be Solved by the Invention) However, when trying to read italic characters such as English italics, if the characters are projected in a direction perpendicular to the line direction in the same way as conventional regular characters, adjacent characters will overlap and be projected. Therefore, it is impossible to extract a pattern for each character. Therefore, when an italic character is recognized by a character recognition device, the projection direction is set to the italic character Fl as shown in FIG.
・・の傾斜方向とほぼ平行に投影して投影値を算出しパ
ターンＤＩ、 ・・を求め、しかも投影方向の角度を
予め最適なものに設定しておかねばならない。特に、投
影方向の角度設定が悪い場合は斜体文字Ｆｌ、 ・・
・の認識率が低下する問題があった。It is necessary to calculate the projection value by projecting substantially parallel to the inclination direction of . . . to obtain the pattern DI, . In particular, if the angle setting of the projection direction is bad, italic letters Fl, ・・
・There was a problem that the recognition rate decreased.
本発明は、上記の問題点を解決するためになされたもの
で、その目的とするところは英文イタリツク体等の斜体
文字を読み取ろうとする場合に、最適な投影角度により
高精度で認識することのできる文字認識装置を提供する
ことにある。The present invention has been made to solve the above problems, and its purpose is to recognize it with high accuracy using an optimal projection angle when attempting to read italic characters such as English italics. Our goal is to provide a character recognition device that can.
（課題を解決するための手段）
上記目的を達成するために、本発明は、対象文字を投影
して得られた投影値のパターンにもとづいて文字を認識
する文字認識装置において、投影方向を微小角度ずつ変
えながら認識対象行の各文字についてそれぞれ投影値を
算出する手段と、角度ごとに得られた投影値パターンの
なかから認識に最適な投影値パターンを選択する手段と
を備えたことを特徴とする。(Means for Solving the Problems) In order to achieve the above object, the present invention provides a character recognition device that recognizes characters based on a pattern of projection values obtained by projecting a target character. The present invention is characterized by comprising means for calculating a projection value for each character in a line to be recognized while changing the angle one by one, and means for selecting a projection value pattern most suitable for recognition from among the projection value patterns obtained for each angle. shall be.
（作 用）
本発明においては、投影値を算出する際に投影方向を微
小角度ずつ変えながら認識対象行の各文字についてそれ
ぞれ投影値が算出され、角度ごとに得られた投影値パタ
ーンのなかから認識に最適な投影値パターンが選択され
ることにより、斜体文字を高精度で認識することができ
る。(Function) In the present invention, when calculating the projection value, the projection value is calculated for each character in the line to be recognized while changing the projection direction by small angles, and from among the projection value patterns obtained for each angle. By selecting the optimum projection value pattern for recognition, italic characters can be recognized with high precision.
（実施例） 以下、図に沿って本発明の詳細な説明する。(Example) The present invention will be described in detail below with reference to the drawings.
第１図は本発明に係る文字認識装置の文字切り出し動作
の一例を示すフローチャートである。このフローチャー
トにもとづいて、第２図の投影方向設定の説明図を参照
しながら動作を説明する。FIG. 1 is a flowchart showing an example of a character extraction operation of the character recognition device according to the present invention. Based on this flowchart, the operation will be explained with reference to an explanatory diagram of projection direction setting in FIG.
先ず、文字認識装置の動作が開始されると、入力画像の
なかから認識対象となる文書画像が抽出される（ステッ
プ１０１）。First, when the character recognition device starts operating, a document image to be recognized is extracted from an input image (step 101).
次いで、抽出した文書画像を横方向に投影して行を抽出
する（ステップ１０２）。第３図は、ここで抽出された
行の一例を示し、英文イタリック体からなる斜体文字を
含む投影領域Ｓが抽出される。Next, the extracted document image is projected in the horizontal direction to extract lines (step 102). FIG. 3 shows an example of the lines extracted here, and a projection area S including italic characters made of English italics is extracted.
次に、投影領域Ｓの長手方向、すなわち第２図のＸ方向
の中心線ｌＣ上に投影の基準位置Ｑ （ｘ）が設定され
る（ステップ１０３）。この投影の基準位置Ｑ（ｘ）は
左端から設定が開始され、右方向に順次、画素単位で移
動される。Next, a projection reference position Q (x) is set in the longitudinal direction of the projection area S, that is, on the center line 1C in the X direction in FIG. 2 (step 103). Setting of the reference position Q(x) for this projection is started from the left end and is sequentially moved to the right pixel by pixel.
基準位置Ｑ（に）が設定されると、その位置を通る投影
線の傾斜角度θ８を第２図に示すように時針方向に一定
角度ずつずらしながら、その投影線上の黒画素数をカウ
ントした投影値Ｐ、を求めてメモリに格納する（ステッ
プ１０４〜１０６）。Once the reference position Q is set, the projection is performed by counting the number of black pixels on the projection line while shifting the inclination angle θ8 of the projection line passing through that position by a fixed angle in the direction of the hour hand as shown in Figure 2. The value P is determined and stored in memory (steps 104 to 106).
投影値Ｐ、の算出が所定の傾斜角度θ７まで終了すると
（ステップ１０７肯定）、基準位置Ｑ （ｘ）を右方向
にΔＸ、具体的にはＩ画素分ずらして、同様に投影線の
傾斜角度θ、をずらしながら、投影値Ｐ、を求めてメモ
リに格納する（ステップ１０９否定、ステップ１１０、
ステップ１０４〜１０８）。When the calculation of the projection value P is completed up to the predetermined inclination angle θ7 (step 107 is affirmative), the reference position Q (x) is shifted to the right by ΔX, specifically I pixels, and the inclination angle of the projection line is similarly calculated. While shifting θ, calculate the projection value P and store it in the memory (step 109 negative, step 110,
Steps 104-108).
これらの投影値Ｐ、の算出が投影領域Ｓの右端まで終了
すると（ステップ１０９肯定）、メモリに格納されてい
る投影値Ｐ８を左端のｉ−１から順に読み出してその変
化量Ｙｎを算出する（ステップ１１１〜１１４）。When the calculation of these projection values P, is completed up to the right end of the projection area S (Yes at step 109), the projection values P8 stored in the memory are read out in order from i-1 at the left end, and the amount of change Yn is calculated ( Steps 111-114).
この変化量Ｙｎとしては、
Ｙｎ＝Σｌ Ｐｎ（ｘ＋Δｘ） −Ｐｎ（ｘ） ｌ −
（１）また、
Ｐｎ（ｘ）−Ｐｔｈ＞Ｏの時、Ｐｎ ’ （ｘ）−１−
（２）Ｐｎ（ｘ）−Ｐｔｈ≦０の時、Ｐｎ ’ （ｘ）
−〇 ・＝（３１（ただし、ｐｔｈは予め設定しておい
た闇値）として、投影値Ｐ、を２値化し、値が１の部分
をＷｉ（ｉ＝１＝Ｎｗ）とし、値がＯの部分を５ｊ（ｊ
＝１〜Ｎｓ）とする。なおここでＮは自然数を意味する
。This amount of change Yn is: Yn=Σl Pn(x+Δx) −Pn(x) l −
(1) Also, when Pn(x)-Pth>O, Pn' (x)-1-
(2) When Pn(x)-Pth≦0, Pn' (x)
−〇 ・=(31 (however, pth is the darkness value set in advance), the projection value P is binarized, the part where the value is 1 is set as Wi (i=1=Nw), and the value is O 5j (j
=1 to Ns). Note that N here means a natural number.
ここで、
Ｙｎ−ΣＳｊ・・・・・・（４）
とする。すなわち、この場合の変化量Ｙｎは、値がＯと
判別された投影線区間の総和である。Here, it is assumed that Yn-ΣSj (4). That is, the amount of change Yn in this case is the sum of the projection line sections whose value is determined to be O.
同様に、
Ｙｎ＝Ｎｓ・・・・・・・・・（５）
すなわち、この場合の変化量Ｙｎは、値が０である連続
した投影線区間の個数である。Similarly, Yn=Ns (5) That is, the amount of change Yn in this case is the number of continuous projection line sections whose value is 0.
これらの変化量Ｙｎを、それぞれθの値を変えながら算
出して、変化量Ｙｎが最大になるθの値を求める（ステ
ップ１１５）。These amounts of change Yn are calculated while changing the value of θ, and the value of θ that maximizes the amount of change Yn is determined (step 115).
変化量Ｙｎを最大にする投影線の傾斜角度θが得られる
と、その傾斜角度θにより投影して得られた投影パター
ンを用いて、文字の切り出し処理がおこなわれて（ステ
ップ１１６）、処理が終了する。When the inclination angle θ of the projection line that maximizes the amount of change Yn is obtained, character cutout processing is performed using the projection pattern obtained by projecting at the inclination angle θ (step 116). finish.
なお、（１１式で用いるＰｎが、０に近い所定の値以下
である場合は、オフセット補正として全て０に補正する
処理を加えることにより、変化量Ｙｎの算出値を強調す
ることができる。Note that (if Pn used in equation 11 is less than or equal to a predetermined value close to 0), the calculated value of the amount of change Yn can be emphasized by adding a process of correcting all to 0 as offset correction.
第４図は、傾斜角度θ。を９０度から７４度まで２度お
きに変えた場合に、上部に示す斜体文字を投影して得ら
れる９種類の投影値のパターンを示す図である。文字間
の分離の点から見比べると７８度および８０度が最もは
っきり分離されている。FIG. 4 shows the inclination angle θ. FIG. 9 is a diagram showing patterns of nine types of projection values obtained by projecting the italicized characters shown at the top when the angle is changed every two degrees from 90 degrees to 74 degrees. In terms of separation between characters, 78 degrees and 80 degrees are the most clearly separated.
この発明では、文字間の分離の明確さや、パターンの特
徴の出具合から最適の傾斜角度θ７が選択されるが、こ
のとき選択される傾斜角度θは、必ずしも、文字形状自
体の傾斜角度と一致するものではなく、斜体文字を認識
するためにパターンの特徴が最もよくあられれる傾斜角
度が選択される。In this invention, the optimum inclination angle θ7 is selected based on the clarity of the separation between characters and the appearance of the characteristics of the pattern, but the inclination angle θ selected at this time does not necessarily match the inclination angle of the character shape itself. Instead, the tilt angle that best reveals the pattern features for recognizing italic characters is selected.
このようにして、最適な傾斜角度θが選択されて文字切
り出しがおこなわれたことにより、以後の文字認識処理
における認識率を向上させて文字〜 ７ −
認識装置そのものの性能を向上させることができる。In this way, by selecting the optimal inclination angle θ and character extraction, it is possible to improve the recognition rate in subsequent character recognition processing and improve the performance of the character recognition device itself. .
なお、実施例では、認識対象行の全てについて、画素ご
とに投影位置を移動し、それぞれの位置で傾斜角度を順
次変えて投影値を算出処理していたが、次に抽出される
行についても同様な処理を実行して最適な傾斜角度を算
出処理することもできるが、同一の書体からなる文字画
像であることが予めわかっていれば、最初の行で求めた
傾斜角度を採用して、以後の行については最適な傾斜角
度の算出を省略し、切り出した行についてただちに投影
値の算出を開始することも可能である。In addition, in the embodiment, the projection position was moved pixel by pixel for all recognition target rows, and the projection value was calculated by sequentially changing the tilt angle at each position. It is possible to perform similar processing to calculate the optimal tilt angle, but if it is known in advance that the character images are composed of the same font, the tilt angle determined in the first line is used, It is also possible to omit calculation of the optimal inclination angle for subsequent rows and immediately start calculation of projection values for the cut out rows.
同様に、最適な傾斜角度の算出処理を、認識対象の文字
画像全部について実行せずに、一部の行を選択抽出し、
その行についてのみ実行したり、あるいは各行ごとにそ
の一部の区間を抽出して実行するごともできる。Similarly, the process of calculating the optimal inclination angle is not performed on the entire character image to be recognized, but only some lines are selected and extracted.
You can execute only that line, or you can extract and execute a part of each line.
なお、上記処理中の基準位置Ｑ （Ｘ）を順次画素単位
でＸ方向にずらしながらその位置における投影線の傾斜
角度θ８を一定角度ずつずらしながら、それぞれの角度
の投影線上の黒画素数をカウントする処理は、ソフトウ
ェア処理でも、専用のハードウェア回路のいずれによっ
ても実現できる。Note that while the reference position Q (X) during the above processing is sequentially shifted in the X direction pixel by pixel, the inclination angle θ8 of the projection line at that position is shifted by a fixed angle, and the number of black pixels on the projection line at each angle is counted. The processing can be realized by either software processing or dedicated hardware circuitry.
（発明の効果）
以上述べたように本発明によれば、投影値を算出する際
に投影方向を微小角度ずつ変えながら認識に最適な投影
値パターンを求めてから、文字認識をおこなうため、英
文イタリック体等の斜体文字を高精度で認識することが
可能となり、文字認識装置の認識性能を向上させること
ができる。(Effects of the Invention) As described above, according to the present invention, when calculating projection values, the projection value pattern optimal for recognition is obtained by changing the projection direction by small angles, and then character recognition is performed. It becomes possible to recognize oblique characters such as italics with high accuracy, and the recognition performance of the character recognition device can be improved.
第１図は本発明の一実施例の処理動作を示すフローチャ
ートであり、第２図は投影方向の設定を示す説明図、第
３図は抽出された斜体文字１行分を含む投影領域画像、
第４図は傾斜角度を変えた場合に斜体文字を投影して得
られる各種投影値のパターン図、第５図は従来の斜体文
字でない通常の正体文字を認識する際の投影の説明図、
第６図は斜体文字を認識する従来の文字認識装置におけ
る投影の説明図である。
Ｓ・・・投影領域 ｌｃ・・・抽出行の中心線Ｑ （ｘ
）・・・投影の基準位置 θ直・・・傾斜角度Ｐ、・・
・投影値
Ｏ
Ｏ
本ｉ
ｊｊｊｊ
」
」
」
」FIG. 1 is a flowchart showing the processing operation of an embodiment of the present invention, FIG. 2 is an explanatory diagram showing the setting of the projection direction, and FIG. 3 is a projection area image including one line of extracted italic characters,
Fig. 4 is a pattern diagram of various projection values obtained by projecting an italic character when the inclination angle is changed, and Fig. 5 is an explanatory diagram of projection when recognizing normal normal characters that are not conventional italic characters.
FIG. 6 is an explanatory diagram of projection in a conventional character recognition device that recognizes italic characters. S... Projection area lc... Center line Q (x
)...Reference position of projection θ straight...Inclination angle P,...
・Projected value
Claims (1)
いて文字を認識する文字認識装置において、 投影方向を微小角度ずつ変えながら認識対象行の各文字
についてそれぞれ投影値を算出する手段と、 角度ごとに得られた投影値パターンのなかから認識に最
適な投影値パターンを選択する手段と、を備えたことを
特徴とする文字認識装置。[Claims] In a character recognition device that recognizes characters based on a pattern of projection values obtained by projecting a target character, the projection value is determined for each character in a line to be recognized while changing the projection direction by small angles. A character recognition device comprising: means for calculating; and means for selecting a projection value pattern most suitable for recognition from among projection value patterns obtained for each angle.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2152244A JPH0444187A (en) | 1990-06-11 | 1990-06-11 | Character recognizing device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2152244A JPH0444187A (en) | 1990-06-11 | 1990-06-11 | Character recognizing device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| JPH0444187A true JPH0444187A (en) | 1992-02-13 |
Family
ID=15536249
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| JP2152244A Pending JPH0444187A (en) | 1990-06-11 | 1990-06-11 | Character recognizing device |
Country Status (1)
| Country | Link |
|---|---|
| JP (1) | JPH0444187A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH06301781A (en) * | 1993-02-03 | 1994-10-28 | Internatl Business Mach Corp <Ibm> | Method and equipment for image transformation for pattern recognition by computer |
| US5872725A (en) * | 1994-12-05 | 1999-02-16 | International Business Machines Corporation | Quasi-random number generation apparatus and method, and multiple integration apparatus and method of function f |
| JP2013171309A (en) * | 2012-02-17 | 2013-09-02 | Omron Corp | Character segmentation method, and character recognition device and program using the same |
-
1990
- 1990-06-11 JP JP2152244A patent/JPH0444187A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH06301781A (en) * | 1993-02-03 | 1994-10-28 | Internatl Business Mach Corp <Ibm> | Method and equipment for image transformation for pattern recognition by computer |
| US6246793B1 (en) | 1993-02-03 | 2001-06-12 | International Business Machines Corp. | Method and apparatus for transforming an image for classification or pattern recognition |
| US5872725A (en) * | 1994-12-05 | 1999-02-16 | International Business Machines Corporation | Quasi-random number generation apparatus and method, and multiple integration apparatus and method of function f |
| JP2013171309A (en) * | 2012-02-17 | 2013-09-02 | Omron Corp | Character segmentation method, and character recognition device and program using the same |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110688930B (en) | Face detection method and device, mobile terminal and storage medium | |
| EP0843275A2 (en) | Pattern extraction apparatus and method for extracting patterns | |
| JPH05346970A (en) | Document recognizing device | |
| CN102622593A (en) | Text recognition method and system | |
| JP2011243201A (en) | Method and system for preprocessing image for optical character recognition | |
| CN102063621B (en) | Method and device for correcting geometric distortion of character lines | |
| US20050281464A1 (en) | Particular image area partitioning apparatus and method, and program for causing computer to perform particular image area partitioning processing | |
| JP5906788B2 (en) | Character cutout method, and character recognition apparatus and program using this method | |
| CN110516655B (en) | Chinese character image stroke processing method and system | |
| JPH0444187A (en) | Character recognizing device | |
| JPH11167455A (en) | Hand form recognition device and monochromatic object form recognition device | |
| KR20020078663A (en) | Patched Image Alignment Method and Apparatus In Digital Mosaic Image Construction | |
| CN108460386A (en) | Character picture cutting method, device, equipment and storage medium | |
| JP2000076378A (en) | Character recognizing method | |
| CN111563511A (en) | Method and device for intelligently framing questions, electronic equipment and storage medium | |
| CN202533964U (en) | Text recognition system | |
| US6034702A (en) | Character forming apparatus | |
| KR102450872B1 (en) | Dot pattern and method for recognizing dot pattern | |
| CN110503622B (en) | Image global positioning optimizing splicing method based on positioning data | |
| CN112580638B (en) | Text detection method and device, storage medium and electronic equipment | |
| JP2624559B2 (en) | Character recognition device | |
| JPH1196379A (en) | Eye position detector | |
| JP2977230B2 (en) | Character extraction method | |
| JPH08194773A (en) | Method and device for processing picture | |
| JP2010039615A (en) | Character recognition method and character recognition apparatus |