JP2002288668A - Curve linear transformation method - Google Patents

Curve linear transformation method

Info

Publication number
JP2002288668A
JP2002288668A JP2001130376A JP2001130376A JP2002288668A JP 2002288668 A JP2002288668 A JP 2002288668A JP 2001130376 A JP2001130376 A JP 2001130376A JP 2001130376 A JP2001130376 A JP 2001130376A JP 2002288668 A JP2002288668 A JP 2002288668A
Authority
JP
Japan
Prior art keywords
curve
deformation
changing
transformation
degree
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.)
Withdrawn
Application number
JP2001130376A
Other languages
Japanese (ja)
Inventor
Yoshinobu Takeuchi
良亘 竹内
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.)
Individual
Original Assignee
Individual
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 Individual filed Critical Individual
Priority to JP2001130376A priority Critical patent/JP2002288668A/en
Publication of JP2002288668A publication Critical patent/JP2002288668A/en
Withdrawn legal-status Critical Current

Links

Landscapes

  • Character Discrimination (AREA)
  • Image Generation (AREA)
  • Image Analysis (AREA)

Abstract

PROBLEM TO BE SOLVED: To produce a transformed curve, in a state of a transformed degree of the curve agreeing with an identification degree of recognition, and to combine the curves of the same transformation degree, when the transformation curves having the same transformation degree are combined. SOLUTION: A correlation function used conventionally as criterion in a identification method, is applied as the transformation degree. One reference curve is determined to realize three means for producing the transformation curve having the predetermined correlation function with the reference curve. That is, (1) a transformation method by changing an inclination, without changing the length of each intercept, (2) a transformation method by changing the length without changing the inclination of each intercept, and (3) a combination method for equally producing an aimed transformation curve by combining both, when the curve is represented by connecting divided intercepts.

Description

【発明の詳細な説明】DETAILED DESCRIPTION OF THE INVENTION

【0001】[0001]

【発明が属する技術分野】本発明は、手書き文字認識方
法および図形認識方法において、それらの基本技術であ
る曲線を変形する方法に関し、特に学習およびテストに
用いるデータの生成方法に関するものである。
BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method of deforming a curve, which is a basic technique of a handwritten character recognition method and a graphic recognition method, and more particularly to a method of generating data used for learning and testing.

【0002】[0002]

【従来の技術】曲線の変形法は、パターン認識における
基本技術のひとつである。それは、二つの曲線同士のマ
ッチングに用いられる。一方は未知入力パターンであ
り、他方は認識辞書パターンである。どちらか一方が変
形される。曲線の変形法は大きく四つに分類される。
(1)正規分布乱数を用いる方法。曲線のパラメータま
たは座標を正規分布乱数により変動させる方法である。
(2)適当に選択されたいくつかのパターンを融合する
方法。凸集合の考え方により座標またはパラメータを合
成する。(3)変形規則を与えて曲線上の点を目標点へ
移動させる方法。変形規則として非線形な関数、ファジ
ィ関数が用いられる。モーフィングにも適用される。こ
れらの起こり得る変形は、変形ベクトル場という量で一
括して扱われる。(4)曲線のエネルギーを利用する方
法。曲線の満たすべき物理方程式を設定して拘束条件を
満たすように変形する。
2. Description of the Related Art Curve deformation is one of the basic techniques in pattern recognition. It is used for matching two curves. One is an unknown input pattern and the other is a recognition dictionary pattern. Either one is deformed. Curve deformation methods are roughly classified into four.
(1) A method using a normally distributed random number. This is a method in which parameters or coordinates of a curve are changed by a normally distributed random number.
(2) A method of fusing several appropriately selected patterns. The coordinates or parameters are synthesized according to the concept of a convex set. (3) A method of moving a point on a curve to a target point by giving a deformation rule. A non-linear function or fuzzy function is used as the deformation rule. Also applies to morphing. These possible deformations are treated collectively in quantities called deformation vector fields. (4) A method using the energy of a curve. A physical equation to be satisfied by the curve is set and deformed so as to satisfy the constraint condition.

【0003】曲線の変形法は、人工的に曲線サンプルを
生成するのにも用いることができる。人工サンプルは次
のような場合に有用である。少数サンプルしか入手でき
ない場合、サンプル数を増す。学習サンプル集合とテス
トサンプル集合との独立性が怪しい場合、どちらかの代
用となる人工サンプル集合を提供する。人工サンプル生
成に使用できる曲線の変形法は、上記(1)と(2)で
ある。その理由は、可能なあらゆる変形曲線を生成でき
ることが必要だからである。(3)と(4)はあるパタ
ーンを他方のパターンへ近づける方法である。したがっ
て、ひとつ又はいくつかの標準の曲線からあらゆる変形
曲線を生成できない。(2)の方法は、(1)で生成さ
れた変形曲線の集合をもとにして、あらたな集合を作
る。
[0003] Curve deformation methods can also be used to artificially generate curve samples. Artificial samples are useful in the following cases. If only a few samples are available, increase the number of samples. When the independence between the training sample set and the test sample set is doubtful, an artificial sample set as a substitute for either is provided. Curve deformation methods that can be used for artificial sample generation are (1) and (2) above. The reason is that it is necessary to be able to generate all possible deformation curves. (3) and (4) are methods for bringing a certain pattern closer to the other pattern. Therefore, not every deformation curve can be generated from one or several standard curves. The method (2) creates a new set based on the set of deformation curves generated in (1).

【0004】(1)の方法では乱数が用いられる。そこ
では変形度を定義して、その制限範囲内で変形曲線を生
成する。その変形度は、認識の識別法とは結びついてい
なかった。すなわち、従来の変形度を使用して(1)の
方法により変形曲線のサンプル集合を生成したとき、そ
の集合は、認識の識別度で区別される曲線の集合とは直
接の関係をつけることができないという欠点があった。
従って、認識方法の学習に効果的な学習サンプルを準備
できないという問題点があった。
In the method (1), a random number is used. There, a degree of deformation is defined, and a deformation curve is generated within the limited range. The degree of deformation was not tied to the recognition discrimination method. That is, when a sample set of deformation curves is generated by the method (1) using the conventional deformation degree, the set may be directly related to the set of curves distinguished by the recognition discrimination degree. There was a disadvantage that it could not be done.
Therefore, there is a problem that an effective learning sample cannot be prepared for learning the recognition method.

【0005】また、(2)の方法では、同じ変形度を持
つ変形曲線同士を合成しても、合成曲線の変形度は異な
った値となってしまう欠点があった。
Further, the method (2) has a drawback that even if the deformation curves having the same deformation degree are combined, the deformation degrees of the combined curves have different values.

【0006】[0006]

【発明が解決しようとする課題】解決しようとする問題
点は、曲線の変形度が認識の識別度と一致しない点であ
り、また、同じ変形度を持つ変形曲線同士を合成したと
き、異なる変形度の曲線が合成されてしまう点である。
The problem to be solved is that the degree of deformation of the curve does not coincide with the discrimination degree of recognition, and when the deformed curves having the same degree of deformation are combined, different deformation levels are obtained. The point is that the degree curves are combined.

【0007】[0007]

【課題を解決するための手段】本発明は、識別法の判定
基準として古くから使用されている相関係数を変形度と
する。一つの標準の曲線を定め、それと決められた相関
係数を持つ変形曲線を生成する三つの手段を実現した。
それは、曲線を分割切片のつながりで表現したとき、
[1]各切片の長さを変えずに傾きを変えることによる
変形法と、[2]各切片の傾きを変えずに長さを変える
ことによる変形法と、[3]両者の合成から、目的とす
る変形曲線を万遍なく生成できる合成法とである。本発
明はこれらの手段を一つの方法として実現することを特
徴とする。
According to the present invention, a correlation coefficient which has been used for a long time as a criterion of a discrimination method is defined as a degree of deformation. Three means for defining one standard curve and generating a deformation curve with the determined correlation coefficient have been realized.
That is, when the curve is represented by the connection of divided intercepts,
From [1] a deformation method by changing the inclination without changing the length of each section, [2] a deformation method by changing the length without changing the inclination of each section, and [3] a synthesis of both, This is a synthesis method that can generate the desired deformation curve uniformly. The present invention is characterized in that these means are realized as one method.

【0008】[0008]

【発明の実施の形態】決められた相関係数を持つ変形曲
線を生成する場合、手段[1]では、切片の傾きを変え
得る範囲を与えたこと、さらに手段[1]および[2]
では、変形の基本形を与えたこと、手段[3]では、手
段[1]および[2]で生成された曲線群から一意に目
的とする変形曲線を生成する手順を与えたことにより、
本発明を実現した。
DESCRIPTION OF THE PREFERRED EMBODIMENTS When a deformation curve having a predetermined correlation coefficient is generated, the means [1] provides a range in which the slope of the intercept can be changed, and furthermore, the means [1] and [2].
Then, by giving the basic form of the deformation, and by means [3], by giving the procedure of uniquely generating the desired deformation curve from the group of curves generated by means [1] and [2],
The present invention has been realized.

【0009】[0009]

【実施例】曲線は、曲線上の飛び飛びの点列を折れ線で
結んだ切片列の集まりとする。標準の曲線を次式で表
す。
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A curve is a set of intercepts in which discrete points on the curve are connected by broken lines. The standard curve is represented by the following equation.

【0010】[0010]

【数1】 (Equation 1)

【0011】上式において、wk(k=1,
2,...,n)は、与えられた標準の曲線をn個の切
片に分割し、それぞれの切片をxとy座標の複素数で表
した量である。wkは次式で表される。
In the above equation, wk (k = 1,
2,. . . , N) is a quantity obtained by dividing the given standard curve into n intercepts and expressing each intercept by a complex number of x and y coordinates. wk is represented by the following equation.

【0012】[0012]

【数2】 (Equation 2)

【0013】変形される曲線の切片を次式で表す。The intercept of the curve to be deformed is expressed by the following equation.

【0014】[0014]

【数3】 (Equation 3)

【0015】切片の長さを変えずに、切片の傾きを変え
る変形では、標準の曲線と変形される曲線との相関係数
は次式で与えられる。
In a modification in which the slope of the intercept is changed without changing the length of the intercept, the correlation coefficient between the standard curve and the deformed curve is given by the following equation.

【0016】[0016]

【数4】 (Equation 4)

【0017】上式でcos(Δθ)は指定した相関係数
の値である。Δθkは変形曲線の第k切片が、標準の曲
線の第k切片となす角である。akは次式で示される。
In the above equation, cos (Δθ) is the value of the specified correlation coefficient. Δθk is the angle formed by the k-th intercept of the deformation curve with the k-th intercept of the standard curve. ak is represented by the following equation.

【0018】[0018]

【数5】 (Equation 5)

【0019】図1は、本発明方法の1実施例の図であっ
て、手段[1]において切片の傾きを変えることができ
る範囲を説明する図である。1は線分0Aを示し、座標
原点0と点Aとの長さは1である。線分0Aは、長さa
k(k=1,2,...,n)の区間に分割されてい
る。図1の線分0Aと横軸xeとのなす角はΔθであ
る。線分0Aでは、全ての区間は横軸に対して同じ角度
Δθの傾きである。点Aから横軸上への垂線の位置がC
である。0Cの長さはcos(Δθ)で、指定された相
関係数の値である。垂線AC上の任意の点Bをとる。区
間kの長さakを保って、原点0から点Bへ到る繋がっ
た区間列は変形のひとつである。これは図1の2で示さ
れる。それぞれの区間kの横軸に対する傾きΔθkの定
め方は、点Bを指定したとき、多くの定め方がある。あ
たかも直線鎖0AのA端をB端へ動かしたとき、たるむ
鎖の形態にはいろいろとりうるように。鎖の置かれる範
囲は、点0と点Bを焦点として、長径の2倍が1である
楕円内である。これは図1の3で示される。どれかの区
間kの一端が楕円の周上に乗るときは、その点と点0、
点Bとは直線で結ばれる置き方になる。点0に一端が乗
っている区間1の傾きΔθ1を決めて、つぎに区間2の
Δθ2を決めて、という順に進む場合を考える。区間1
のΔθ1が定まると、鎖(2,3,...,n)の置き
方は、区間1の端(点0ではない)と点Bとを焦点とし
て長径の2倍が1−a1である楕円内になる。この手続
きは順次に再帰的である。最後のn−1,n区間が定め
るべき残りとなったとき、両者の共通端点は楕円周上に
置かねばならない。図1を上下に反転したものも可能な
変形を与える。
FIG. 1 is a diagram of an embodiment of the method of the present invention, and is a diagram for explaining a range in which the inclination of the intercept can be changed by means [1]. 1 indicates a line segment 0A, and the length between the coordinate origin 0 and the point A is 1. The line segment 0A has a length a
k (k = 1, 2,..., n). The angle between the line segment 0A in FIG. 1 and the horizontal axis xe is Δθ. In the line segment 0A, all the sections have inclinations of the same angle Δθ with respect to the horizontal axis. The position of the perpendicular from point A to the horizontal axis is C
It is. The length of 0C is cos (Δθ), which is the value of the specified correlation coefficient. An arbitrary point B on the vertical line AC is taken. The connected section sequence from the origin 0 to the point B while maintaining the length ak of the section k is one of the modifications. This is indicated by 2 in FIG. There are many ways to determine the slope Δθk of each section k with respect to the horizontal axis when the point B is specified. As if moving the A end of the straight chain 0A to the B end, the form of the slack chain could be various. The range where the chain is placed is within an ellipse whose focal point is point 0 and point B and whose major axis is twice as long. This is indicated by 3 in FIG. When one end of any section k rides on the circumference of the ellipse, that point and point 0,
The point B is placed in a straight line. Consider a case where the slope Δθ1 of the section 1 having one end on the point 0 is determined, then the Δθ2 of the section 2 is determined, and the process proceeds in the order of. Section 1
Is determined, the chain (2, 3,..., N) is placed with the end of section 1 (not point 0) and point B as the focal point and twice the major axis is 1-a1. Within the ellipse. This procedure is sequentially recursive. When the last n-1 and n sections are left to be determined, their common endpoints must be located on the circumference of the ellipse. The upside down of FIG. 1 also gives possible variations.

【0020】図2に標準の曲線の例を、図3に切片の長
さを変えずに、傾きを変える変形の例をしめす。
FIG. 2 shows an example of a standard curve, and FIG. 3 shows an example of a modification in which the slope is changed without changing the length of the intercept.

【0021】図4は、図1の鎖の置き方のもう一つの実
施例である。点0と点BをTchebycheffの3
次多項式の曲線で繋いである。ある程度低次の多項式を
用いると、ギクシャクした変形を生じない変形の基本形
を準備することができる。
FIG. 4 is another embodiment of the arrangement of the chains of FIG. Point 0 and point B are 3 of Tchebycheff
They are connected by a curve of degree polynomial. By using a low-order polynomial to some extent, it is possible to prepare a basic form of deformation that does not cause jerky deformation.

【0022】つぎに、切片の傾きは変えずに、長さを変
える場合は、標準の曲線と変形曲線との相関係数は次式
となる。
Next, when changing the length without changing the inclination of the intercept, the correlation coefficient between the standard curve and the deformation curve is as follows.

【0023】[0023]

【数6】 (Equation 6)

【0024】ここで、ak〜は次式で与えられる。Here, ak〜 is given by the following equation.

【0025】[0025]

【数7】 (Equation 7)

【0026】数3に示す変形曲線の切片wk〜を、数2
に示す標準の曲線のwkから次式に従って変える。これ
は変形の基本形を与える。
The intercept wk ~ of the deformation curve shown in Equation 3 is expressed by Equation 2
Is changed according to the following equation from wk of the standard curve shown in FIG. This gives the basic form of the deformation.

【0027】[0027]

【数8】 (Equation 8)

【0028】ここで、Tm( )はm次のTcheby
cheffの多項式である。上式右辺のwkにかかる項
は実数であるので、wk〜はwkとは傾きが変わらずに
長さだけ変わる。この項に含まれる未知の実定数cとd
は、数8を数7、数6に代入することにより、二次式の
代数方程式を満たす。cを与えるとdを定めることがで
きる。cの大小は変形曲線のスケールの大きさを変える
だけで、形状は変えない。
Here, Tm () is the m-th order Tcheby.
This is a polynomial of cheff. Since the term related to wk on the right side of the above equation is a real number, wk〜 changes only by length without changing the slope from wk. Unknown real constants c and d contained in this term
Satisfies the quadratic algebraic equation by substituting equation 8 into equations 7 and 6. Given c, d can be determined. The magnitude of c only changes the size of the scale of the deformation curve, but does not change the shape.

【0029】図5の実施例は、このようにして図2の標
準の曲線を変形させた例である。
The embodiment of FIG. 5 is an example in which the standard curve of FIG. 2 is modified in this way.

【0030】最後に、以上に述べた[1]各切片の長さ
を変えずに傾きを変えることによる変形法と、[2]各
切片の傾きを変えずに長さを変えることによる変形法と
を合成する。これらのうちTchebycheffの多
項式を用いる変形法は、変形の基本パターンを与えるも
のであった。万遍なく変形曲線を生成するには、基本パ
ターンの変形曲線同士を合成する必要がある。これは次
式で示される。
Finally, the above-mentioned [1] Deformation method by changing the inclination without changing the length of each section, and [2] Deformation method by changing the length without changing the inclination of each section And are synthesized. Among these, the deformation method using the Tchebycheff polynomial gives a basic pattern of the deformation. In order to generate deformation curves uniformly, it is necessary to combine the deformation curves of the basic patterns. This is shown by the following equation.

【0031】[0031]

【数9】 (Equation 9)

【0032】ここで、vは合成曲線を示し、添え字iは
傾きを変えてできた変形曲線に関する項を、添え字jは
長さを変えてできた変形曲線に関する項を意味する。そ
れぞれM1個とM2個ある。λiとλjは正符号であ
り、かつそれらの総和は1である。λ0は負の値をと
り、標準曲線wを変形曲線同士の合成から差し引く役割
をする。相関係数ρ(w,v)をcos(Δθ)に等し
くするには、λ0について解く必要がある。これは二次
の代数方程式を解くことにより決定される。
Here, v indicates a composite curve, a subscript i indicates a term relating to a deformation curve formed by changing the inclination, and a subscript j indicates a term relating to a deformation curve formed by changing the length. There are M1 and M2 respectively. λi and λj are positive signs, and their sum is 1. λ0 takes a negative value and serves to subtract the standard curve w from the combination of the deformation curves. To make the correlation coefficient ρ (w, v) equal to cos (Δθ), it is necessary to solve for λ0. This is determined by solving a quadratic algebraic equation.

【0033】このようにして求められたλ0とλi、λ
jの総和は1でない。そこで、これらの総和の絶対値を
用いてλ0、λi、λjを割ることにより、その結果の
総和は1になる。このようにして、指定された相関係数
の値の合成曲線を得ることができる。
Λ0 and λi, λ thus obtained
The sum of j is not 1. Therefore, by dividing λ0, λi, and λj using the absolute value of these sums, the sum of the results becomes 1. In this way, it is possible to obtain a composite curve of the values of the specified correlation coefficient.

【0034】以上のようにして、相関係数cos(Δ
θ)の多くの合成曲線vを生成し、これらに正の定数係
数をかけて和をとる。定数係数の総和は1にする。この
ような和からできる曲線の集合には、相関係数がcos
(Δθ)に等しいか、より大きいすべての変形曲線が含
まれる。
As described above, the correlation coefficient cos (Δ
θ) are generated and multiplied by a positive constant coefficient to obtain a sum. The sum of the constant coefficients is set to 1. The set of curves formed from such sums has a correlation coefficient of cos
All deformation curves equal to or greater than (Δθ) are included.

【0035】[0035]

【発明の効果】以上説明したように本発明の曲線変形方
法は、識別法の判定基準として古くから使用されている
相関係数を変形度とし、一つの標準の曲線に対して決め
られた相関係数を持つ変形曲線を生成する三つの手段を
与える。すなわち、曲線を分割切片のつながりで表現し
たとき、[1]各切片の長さを変えずに傾きを変えるこ
とによる変形法と、[2]各切片の傾きを変えずに長さ
を変えることによる変形法と、[3]両者の合成から変
形する方法である。本発明はこれらの手段を一つの方法
として実現することにより、目的とする変形曲線を万遍
なく生成できる合成法を実現する。本発明により、認識
の識別度で区別される曲線の集合と直接の関係をつける
ことができるサンプル集合を生成できる利点が生まれ
る。
As described above, according to the curve deformation method of the present invention, the correlation coefficient which has been used for a long time as a criterion for the discrimination method is used as the degree of deformation, and the phase determined for one standard curve is used. We provide three means for generating deformation curves with relational numbers. That is, when a curve is represented by a series of divided sections, [1] a deformation method by changing the slope without changing the length of each section, and [2] changing the length without changing the slope of each section. And [3] a method of transforming from a combination of both. The present invention realizes a synthesizing method capable of uniformly generating a desired deformation curve by realizing these means as one method. According to the present invention, there is an advantage that a sample set can be generated which can be directly related to a set of curves distinguished by recognition discrimination.

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

【図1】曲線の分割切片の長さを変えずに、傾きを変え
る変形方法を説明する図。相関係数の値=0.95。
(実施例1)
FIG. 1 is a view for explaining a deformation method of changing a slope without changing the length of a segment of a curve. Correlation coefficient value = 0.95.
(Example 1)

【図2】標準の曲線の例を示す図。(実施例1)FIG. 2 is a diagram showing an example of a standard curve. (Example 1)

【図3】図1の方法により標準の曲線を変形した例を示
す図。左側と右側の曲線は別々に変形させた。相関係数
の値=0.95。(実施例1)
FIG. 3 is a view showing an example in which a standard curve is modified by the method of FIG. 1; The left and right curves were deformed separately. Correlation coefficient value = 0.95. (Example 1)

【図4】図1の鎖をTchebycheffの3次多項
式の曲線で置いた例。(実施例2)
FIG. 4 is an example in which the chains of FIG. 1 are arranged by a curve of a third-order polynomial of Tchebycheff. (Example 2)

【図5】曲線の分割切片の傾きを変えずに、長さを変え
る変形方法により、図2の標準曲線を変形させた例。左
側と右側の曲線は別々に変形させた。相関係数の値=
0.95。Tchebycheffの3次多項式を使
用。c=0.7。左の曲線のd=0.95。右の曲線の
d=0.86。(実施例3)
FIG. 5 is an example in which the standard curve of FIG. 2 is deformed by a deformation method of changing the length without changing the inclination of the segment of the curve. The left and right curves were deformed separately. Correlation coefficient value =
0.95. Use the third-order polynomial of Tchebycheff. c = 0.7. D = 0.95 for the left curve. D = 0.86 for the right curve. (Example 3)

【符号の説明】[Explanation of symbols]

1 全ての区間は横軸に対して同じ角度Δθの傾きで
あることを示す線分。 2 繋がった区間列は傾きを変える変形のひとつであ
ることを示す鎖。 3 鎖の置かれる範囲を示す楕円。 xe 曲線の分割切片の長さを変えずに、傾きを変える
変形方法を行う平面での横軸座標 ye 曲線の分割切片の長さを変えずに、傾きを変える
変形方法を行う平面での縦軸座標 A (xe,ye)=(0,0)から長さが1で、横軸
に対してΔθの角度をなす線分の端点 B Aから横軸に下ろした垂線上の任意の点 C Aから横軸に下ろした垂線の足 x 曲線を表す平面の横軸座標 y 曲線を表す平面の縦軸座標
1 A line segment indicating that all sections have the same angle of inclination Δθ with respect to the horizontal axis. 2 A chain indicating that the connected section sequence is one of the transformations that change the slope. 3 Ellipse showing the area where the chain is placed. xe The horizontal axis coordinate on the plane on which the deformation method for changing the inclination is performed without changing the length of the divided section of the curve. ye The vertical axis on the plane on which the deformation method for changing the inclination is changed without changing the length of the divided section of the curve. Axial coordinates A (xe, ye) = (0,0), end point of a line segment having a length of 1 from the (0,0) and forming an angle of [Delta] [theta] with respect to the horizontal axis. Foot of a perpendicular drawn from A to the horizontal axis x Horizontal axis coordinate of plane representing curve y Vertical axis coordinate of plane representing curve

Claims (1)

【特許請求の範囲】[Claims] 【請求項1】 与えられた標準の曲線をその上の飛び飛
びの点列を結ぶ折れ線で表し、それぞれの折れ線すなわ
ち切片について、各切片の長さを変えずに傾きを変えて
その曲線を変形する方法と、各切片の傾きを変えずに長
さを変えてその曲線を変形する方法と、それら両方法に
よる変形曲線群を一つの曲線に合成することにより標準
の曲線を変形する方法から構成され、標準の曲線と指定
された相関係数を持つ変形曲線を生成することを特徴と
する曲線変形法。
1. A given standard curve is represented by a polygonal line connecting a series of discrete points on the standard curve, and the curve is deformed by changing the inclination of each polygonal line, that is, the intercept, without changing the length of the intercept. It consists of a method, a method of deforming the curve by changing the length without changing the slope of each intercept, and a method of deforming a standard curve by synthesizing a group of deformation curves by both methods into one curve. Generating a deformation curve having a standard curve and a specified correlation coefficient.
JP2001130376A 2001-03-23 2001-03-23 Curve linear transformation method Withdrawn JP2002288668A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP2001130376A JP2002288668A (en) 2001-03-23 2001-03-23 Curve linear transformation method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
JP2001130376A JP2002288668A (en) 2001-03-23 2001-03-23 Curve linear transformation method

Publications (1)

Publication Number Publication Date
JP2002288668A true JP2002288668A (en) 2002-10-04

Family

ID=18978763

Family Applications (1)

Application Number Title Priority Date Filing Date
JP2001130376A Withdrawn JP2002288668A (en) 2001-03-23 2001-03-23 Curve linear transformation method

Country Status (1)

Country Link
JP (1) JP2002288668A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103324925A (en) * 2012-03-22 2013-09-25 佳能株式会社 Method and device used for obtaining character data used for handwritten character recognition

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103324925A (en) * 2012-03-22 2013-09-25 佳能株式会社 Method and device used for obtaining character data used for handwritten character recognition

Similar Documents

Publication Publication Date Title
De Graef Introduction to conventional transmission electron microscopy
KR0167618B1 (en) Character generating apparatus and method of the same
Krupa et al. Mixed-mode oscillations in a multiple time scale phantom bursting system
JPS6282486A (en) Recognizing device for online handwritten graphic form
US20130129248A1 (en) Method and Device for Obtaining an Image of a Crumpled Document From an Image of Said Document When Same is Crumpled
Nita et al. Dressing the coronal magnetic extrapolations of active regions with a parameterized thermal structure
US20220005212A1 (en) Shape supplementation device, shape supplementation learning device, method, and program
JPH04243487A (en) Three-dimensional contour processing method
CN105469398A (en) Deformation speckle generation method based on reverse mapping method
Dimitrovska et al. Indirect power-system contingency screening for real-time applications based on PCA
CN103679764B (en) A kind of image generating method and device
JPH02278290A (en) Method for applying curve to shape shown by linear short vector train
Schulz-Mirbach On the existence of complete invariant feature spaces in pattern recognition
JP2002288668A (en) Curve linear transformation method
CN112686817A (en) Image completion method based on uncertainty estimation
US6640201B1 (en) Method for comparing points sets
Hori et al. Quantitative measurement of the performance of raster-to-vector conversion algorithms
JP4175470B2 (en) CAD system, curved surface analyzing apparatus, curved surface reproducing apparatus, method and program thereof
WO2003051034A1 (en) Electronic information burying method and extracting method, electronic information burying apparatus and extracting apparatus, and programs therefor
Wu et al. Depth acquisition from dual-frequency fringes based on end-to-end learning
Ruiz Fourier coefficients and moments of piecewise-circular curves
Benney Nonlinear wave packets in flows with critical layers
He et al. Lattice identification and separation: Theory and algorithm
Nechi A relaxation and gradient combination applied to the computer simulation of a plane four-bar chain
JP2004093166A (en) Target distinguishing device

Legal Events

Date Code Title Description
A300 Application deemed to be withdrawn because no request for examination was validly filed

Free format text: JAPANESE INTERMEDIATE CODE: A300

Effective date: 20080603