Deprecated: The each() function is deprecated. This message will be suppressed on further calls in /home/zhenxiangba/zhenxiangba.com/public_html/phproxy-improved-master/index.php on line 456
JPH0149999B2 - - Google Patents
[go: Go Back, main page]

JPH0149999B2 - - Google Patents

Info

Publication number
JPH0149999B2
JPH0149999B2 JP57045847A JP4584782A JPH0149999B2 JP H0149999 B2 JPH0149999 B2 JP H0149999B2 JP 57045847 A JP57045847 A JP 57045847A JP 4584782 A JP4584782 A JP 4584782A JP H0149999 B2 JPH0149999 B2 JP H0149999B2
Authority
JP
Japan
Prior art keywords
character
strokes
distance
straightness
characters
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.)
Expired
Application number
JP57045847A
Other languages
Japanese (ja)
Other versions
JPS58163080A (en
Inventor
Tetsuji Morishita
Yasuhiko Yoshinaga
Koya Fujita
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 JP57045847A priority Critical patent/JPS58163080A/en
Publication of JPS58163080A publication Critical patent/JPS58163080A/en
Publication of JPH0149999B2 publication Critical patent/JPH0149999B2/ja
Granted legal-status Critical Current

Links

Classifications

    • GPHYSICS
    • G06COMPUTING OR CALCULATING; COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00Arrangements for image or video recognition or understanding
    • G06V10/10Image acquisition
    • G06V10/17Image acquisition using hand-held instruments

Landscapes

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

Description

【発明の詳細な説明】[Detailed description of the invention]

(1) 発明の技術分野 本発明は実時間手書き文字認識装置に係り、と
くに曲線の部分の多いひらがな等の文字を高精度
に近似できる特徴抽出方式に関するものである。 (2) 従来技術と問題点 従来の特徴抽出方式としては、たとえば小高、
荒川、増田、「ストロークの近似点になる手書き
文字のオンライン認識」、電子通信学会論文集
80/2Vol J63DNo.2に発表されたように、文字の
ストロークの長さや曲がりとは無関係に、画数に
よつて代表点数が固定的に決められている。この
ため第2図に示す構成は非常に簡単化され、タブ
レツト1からの入力データは代表点数が画数によ
り固定化された特徴抽出回路2で特徴がマツチン
グ回路3に送られ、辞書4から読出された候補文
字と照合され距離がとられ判定が行なわれる。し
かし、このような方法では第1図に示すように、
曲りのあるストロークをもつ文字「わ」を入力
した場合、辞書から読出される候補文字は「わ」
と「れ」であり、との距離のでる部分
はのαとのA間の距離であり、との距離
のでる部分はのbとのB′間の距離である。
このように、直線部分から計算された距離も、曲
線部分から計算された距離も同等に扱われてしま
うから認識率が悪くなる。 (3) 発明の目的 本発明の目的は曲線の部分の多い文字を高精度
に近似できる特徴抽出方式を提供することであ
る。 (4) 発明の構成 前記目的を達成するため、本発明の文字の特徴
抽出方式は実時間手書き文字認識装置において、
曲線部分の多い文字の特徴抽出をする際、ストロ
ークの直線度を計測する手段と、少なくとも該手
段により計測された直接度に応じて抽出すべき代
表点数を決定する手段とを設けたことを特徴とす
るものである。 (5) 発明の実施例 手書文字、とくにひらがなの場合には“う−
ら”、“わ−れ−ね”のように主に曲線部分に識別
のための情報が集中している。これに対し、従来
は直線部分も曲線部分も全く同じ個数の代表点で
近似していた。しかし、これでは本来本質的でな
い“う−ら”の“ヽ”の位置あるいは“わ−れ−
ね”の“1”の位置の変化により特徴が変つてし
まい、本来識別のために有効な曲線部分の差異が
見えにくい。 そこで、本発明の原理は、代表点を抽出するに
先立ちストロークの直線度を計算し、直線に近い
ストロークは少ない点数で、曲線に近いストロー
クは多くの点数で近似することによつて曲線スト
ロークに重みをつけ、ひらがななどの曲線部分の
多い文字を高精度で近似し認識率を高めるように
したものである。 第3図a,bは上述の原理に従う実施例の説明
図であり、同図aは構成図、同図bは要部の詳細
説明図である。 同図aにおいて、タブレツト1から入力された
文字情報は座標点系列として、直線度計測回路1
1に入力される。ここで次に示す方法で直線度が
計算される。 すなわち、同図bに示すように、曲線のストロ
ークS上代表点数のベクトルP1,P2,…,Po
すると、直線度Csを次のように定義する。ストロ
ークSの延べ距離Lは L=o-1i=1 |P→o-1−P→o| (1) ストロークSのP→1→P→nまでの直線距離lは l=|P→o−P→1| (2) とした場合、直線度Csは(1)、(2)より Cs=l/L (3) である。Csは0≦Cs≦1の実数をもち1に近いほ
ど直線に近い。 この直線度Csと代表点数の1例は次表に示され
る。
(1) Technical Field of the Invention The present invention relates to a real-time handwritten character recognition device, and in particular to a feature extraction method that can accurately approximate characters such as hiragana, which have many curved parts. (2) Conventional techniques and problems Conventional feature extraction methods include, for example, Odaka,
Arakawa, Masuda, “Online recognition of handwritten characters using stroke approximation points”, Proceedings of the Institute of Electronics and Communication Engineers.
As announced in 80/2 Vol J63D No. 2, the number of representative points is fixedly determined by the number of strokes, regardless of the length or curve of the stroke of the character. For this reason, the configuration shown in FIG. 2 is extremely simplified, and the input data from the tablet 1 is sent to the feature extraction circuit 2 in which the number of representative points is fixed according to the number of strokes, and the features are sent to the matching circuit 3, where they are read out from the dictionary 4. The candidate character is compared with the candidate character, the distance is taken, and a determination is made. However, with this method, as shown in Figure 1,
If you input the character "wa" with a curved stroke, the candidate character read from the dictionary is "wa".
The distance between and is the distance between α and A, and the distance between and is the distance between b and B'.
In this way, the distance calculated from a straight line portion and the distance calculated from a curved portion are treated equally, resulting in a poor recognition rate. (3) Purpose of the Invention The purpose of the present invention is to provide a feature extraction method that can approximate characters with many curved parts with high precision. (4) Structure of the Invention In order to achieve the above object, the character feature extraction method of the present invention includes a real-time handwritten character recognition device that
When extracting features of characters with many curved parts, the present invention is characterized by providing means for measuring the straightness of strokes and means for determining the number of representative points to be extracted according to at least the directness measured by the means. That is. (5) Embodiments of the invention In the case of handwritten characters, especially hiragana, “u-
Information for identification is mainly concentrated in curved parts, such as "ra" and "ware-ne."In contrast, in the past, straight and curved parts were approximated by exactly the same number of representative points. However, this does not mean that the position of ``ヽ'' in ``Ura'' or ``Ware'' is not essential.
The feature changes due to the change in the position of "1" in "ne", making it difficult to see the difference in the curved portion that is originally effective for identification. Therefore, the principle of the present invention is that before extracting the representative point, the straight line of the stroke The strokes that are close to straight lines are approximated by a small number of points, and the strokes that are close to curves are approximated by a large number of points, thereby giving weight to curved strokes and approximating characters with many curved parts, such as hiragana, with high precision. 3A and 3B are explanatory diagrams of an embodiment according to the above-mentioned principle, where FIG. 3A is a configuration diagram and FIG. In FIG.
1 is input. Here, linearity is calculated using the following method. That is, as shown in FIG. 2B, if vectors P 1 , P 2 , . . . , P o are representative points on the stroke S of the curve, then the straightness C s is defined as follows. The total distance L of the stroke S is L= o-1i=1 |P→ o-1 −P→ o | (1) The straight-line distance l from P→ 1 →P→n of the stroke S is l=|P → o −P→ 1 | (2) When it is set, the linearity C s is C s =l/L (3) from (1) and (2). C s has a real number of 0≦C s ≦1, and the closer it is to 1, the closer it is to a straight line. An example of this straightness C s and the number of representative points is shown in the following table.

【表】 1 1 1
[Table] 1 1 1

Claims (1)

【特許請求の範囲】[Claims] 1 実時間手書き文字認識装置において、曲線部
分の多い文字の特徴抽出をする際、ストロークの
直線度を計測する手段と、少なくとも該手段によ
り計測された直線度に応じて抽出すべき代表点数
を決定する手段とを設けたことを特徴とする文字
の特徴抽出方式。
1. In a real-time handwritten character recognition device, when extracting features of characters with many curved parts, a means for measuring the straightness of strokes and at least the number of representative points to be extracted are determined according to the straightness measured by the means. A character feature extraction method characterized by providing a means for extracting character features.
JP57045847A 1982-03-23 1982-03-23 Feature extracting system of character Granted JPS58163080A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP57045847A JPS58163080A (en) 1982-03-23 1982-03-23 Feature extracting system of character

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
JP57045847A JPS58163080A (en) 1982-03-23 1982-03-23 Feature extracting system of character

Publications (2)

Publication Number Publication Date
JPS58163080A JPS58163080A (en) 1983-09-27
JPH0149999B2 true JPH0149999B2 (en) 1989-10-26

Family

ID=12730600

Family Applications (1)

Application Number Title Priority Date Filing Date
JP57045847A Granted JPS58163080A (en) 1982-03-23 1982-03-23 Feature extracting system of character

Country Status (1)

Country Link
JP (1) JPS58163080A (en)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPS60217482A (en) * 1984-04-12 1985-10-31 Toshiba Corp Recognizer of character
CH665494A5 (en) * 1985-04-26 1988-05-13 Battelle Memorial Institute METHOD FOR THE DIGITAL STORAGE OF AN ANALOG CURVE AND OF TRACING A REPRESENTATIVE CURVE OF THIS ANALOG CURVE.
JPH04177485A (en) * 1990-11-07 1992-06-24 Matsushita Graphic Commun Syst Inc Character recognizing device

Also Published As

Publication number Publication date
JPS58163080A (en) 1983-09-27

Similar Documents

Publication Publication Date Title
JP3761937B2 (en) Pattern recognition method and apparatus, and computer control apparatus
Kshirsagar Key Frame Selection for One-Two Hand Gesture Recognition with HMM
JPH0149999B2 (en)
CN111126160B (en) Intelligent Chinese character structure evaluation method and system constructed based on five-stroke input method
KR100248601B1 (en) On-line character recognition method and device
JP2675303B2 (en) Character recognition method
JP3366443B2 (en) Character recognition method and device
JPH0210473B2 (en)
JPS5835674A (en) Extracting method for feature of online hand-written character
JPS61198381A (en) Calculation system of distance between strokes
JPS58186882A (en) Input device of handwritten character
JPS55138180A (en) Real-time character recognition unit with additional registering function
JPS57153387A (en) Character recognizing method
JPS5717087A (en) Recognizing method of on-line handwritten character
JPS6334682A (en) Character recognition system
JPH0313630B2 (en)
JPH01213772A (en) On-line character recognizing system
JPH0410115B2 (en)
JPS57182878A (en) Character recognizing device
JPH0474755B2 (en)
JPS642186A (en) Hand written character recognizing system
JPS5668879A (en) Real-time hand-written character recognition system
JPS63298494A (en) Pattern recognizing device
JPS60147891A (en) Character recognition device
JPH01213771A (en) On-line character recognizing system