CN101894370A - Automatic generation method of shape parameter-adaptive oracle-bone inscription contour glyphs - Google Patents

Abstract

The invention discloses an automatic generation method of parameter-adaptive oracle-bone inscription contour glyphs, comprising the following steps: preprocessing an oracle-bone inscription rubbing image; extracting characteristic points on an original contour curve, wherein, the characteristic points comprise angular points and segmentation points; fitting the contour curve with a cubic B-spline curve, and reversely obtaining a control point of the cubic B-spline curve according to the information of the characteristic points; solving the shape parameter value of a cubic B-spline basis function with the minimum total distance deviation from discrete points corresponding to the original contour curve by adopting a least square method; and fitting the cubic B-spline curve according to the control points and the shape parameter value to obtain the needed oracle-bone inscription glyph contour curve. The oracle-bone inscription glyph contour curve generated by the method has the characteristics of small storage content, low distortion rate and high precision and is capable of greatly reducing the process of manual intervention.

Description

Have the adaptive oracle-bone inscription contour font of form parameter and generate method automatically

Technical field

The present invention relates to a kind of graphic processing method, be specifically related to a kind of automatic generation method inscriptions on bones or tortoise shells figure character contour curve.

Background technology

The inscriptions on bones or tortoise shells is the literal of the existing precocity system of China, also is one of the most ancient pictograph of finding so far.Still belong to the drawing stage period that the inscriptions on bones or tortoise shells produces, and the syntax rule that is embodied is the retrospective source the earliest of Chinese grammar.People have obtained certain achievement for the research of the inscriptions on bones or tortoise shells, and making philological studies of ancient texts of literal also made partial interpretation to its meaning.Along with development of modern science, how to use computer technology with inscriptions on bones or tortoise shells digitizing, to inherit this traditional culture better, this rarity of development has crucial historic significance and cultural heritage conservation value on the basis of protection.

The inscriptions on bones or tortoise shells has had general layout symmetry, stable as calligraphy on its structure, formed the system of a uniqueness on the structure and the technique of writing.For the disposal route of Chinese character in ancient times, adopt the artificial method of extracting type matrix mostly, though existing active computer image processing software is as ancillary technique, but efficient is still lower, processing time is long, introduces error easily owing to manual operation is improper, and degree of accuracy does not ensure yet.Simultaneously, if adopt the graphics mode storage inscriptions on bones or tortoise shells, it is big to take storage space, and it is long to open the figure required time.When dwindling, amplification can produce crenellated phenomena.

Though, at present existing font treatment technology can be used for handling modern Chinese character, but because the inscriptions on bones or tortoise shells has the font characteristics of oneself, the modern Chinese character that is different from rule, so inscriptions on bones or tortoise shells font treatment technology is more than the font treatment technology complexity of modern Chinese character, inscriptions on bones or tortoise shells font generates can not apply mechanically modern Chinese character font manufacturing technology simply, need be more suitable for the double bubble technique of the body characteristics of the inscriptions on bones or tortoise shells.Otherwise the font of generation can produce font distortion because the amplification of curve dwindles.

Summary of the invention

The purpose of this invention is to provide a kind of curve fitting generation method,, realize inscriptions on bones or tortoise shells rubbing image transitions is become the form of the contour curve of mathematical description to be applied to the generation of inscriptions on bones or tortoise shells font curved profile with form parameter self-adaptation character.

For achieving the above object, the technical solution used in the present invention is: a kind of have parameter adaptive oracle-bone inscription contour font and generate method automatically, comprises the following steps:

(1) inscriptions on bones or tortoise shells rubbing image is carried out pre-service, described pre-service comprises the original contour curve that image is carried out binary conversion treatment, extraction font;

In concrete enforcement, above-mentioned pre-service can comprise, successively to inscriptions on bones or tortoise shells original image carry out denoising, inscriptions on bones or tortoise shells literal target is obtained in image segmentation operations such as (binaryzations); And then carry out profile and extract and follow the tracks of, obtain contour point information.

(2) the original contour curve that obtains according to step (1) extracts the unique point on the original contour curve, and described unique point comprises angle point and waypoint; The extracting method of described angle point is, calculate each point and the formed angle of line that continues before it at interval between line between k the point of putting and this point and follow-up interval k the point of putting, the angle of this angle is designated as the angle value of this point, begin to search for successively up to the some A of angle value from profile spring of curve S greater than predefined threshold value, in contour segment SA, find the some B of angle value minimum, sign B is an angle point, with this angle point B is that starting point repeats said process to identify next angle point, up to all point of traversal; The starting point of contour curve and terminal point are waypoint; According to experimental result, the value of k is 10, and the threshold value of angle is 160 °.

In fact these unique points are exactly character contour raw information, and different is the feature that these unique points can be represented inscriptions on bones or tortoise shells font, inscriptions on bones or tortoise shells font can be expressed again by certain technology.

(3) with B-spline Curve match contour curve, according to the counter reference mark of asking B-spline Curve of characteristic point information of step (2) acquisition;

(4) utilize least square method to find the solution and the corresponding discrete point of original contour curve form parameter value apart from the cubic B-spline basis function of total departure minimum;

(5) carry out the B-spline Curve match according to the reference mark of step (3) acquisition and the form parameter value of step (4) acquisition, obtain required inscriptions on bones or tortoise shells character contour curve.

In the technique scheme, in the described step (3), at first go out system of equations (1) by the B-spline Curve section of unique point according to the latent structure of B-spline Curve,

In the system of equations (1), Q _i(i=1,2 ..., n) be unique point on the original contour curve, p _i(i=0,1 ..., n n+1) is the reference mark of B-spline Curve, the numerical value of substitution unique point is obtained the value at reference mark.

In the described step (4), the acquisition process of form parameter value is:

$\mathrm{\φ}\left({\mathrm{\λ}}_i\right)=\underset{k=1}{\overset{m}{\mathrm{\Σ}}}{|\underset{j=0}{\overset{3}{\mathrm{\Σ}}}{p}_{i+j-1}{b}_j\left(t_k\right)-d_k|}^2---\left(2\right)$

In the formula (2), φ (λ _i) be matched curve section and discrete point sequence d _kBetween square distance and, the number of m discrete point, t _k(k=1,2,3 ... m) be step-length, adopt the chord length that on average adds up to estimate parametric t here _k, t _k∈ [0,1], initial t ₀=0, t _m=1, t _K+1-t _k=c, c are constant;

Formula (3) is a band parameter lambda in the formula (2) _iThe basis function b of cubic B-spline _j(t):

$\left\{\begin{array}{c}{b}_0\left(t\right)=\frac{1}{6}(1-{\mathrm{\λ}}_{i}t){(1-t)}^2\\ b_1\left(t\right)=\frac{1}{6}[4-(7-{\mathrm{\λ}}_i)t^2+(4-{\mathrm{\λ}}_i)t^3]\\ b_2\left(t\right)=\frac{1}{6}[1+(2+{\mathrm{\λ}}_i)t+(5-2{\mathrm{\λ}}_i)t^2-(4-{\mathrm{\λ}}_i)t^3]\\ b_3\left(t\right)=\frac{1}{6}[1-{\mathrm{\λ}}_i(1-t)]t^2\end{array}\right.---\left(3\right)$

According to

Here force definition

Find the solution form parameter λ _i, promptly obtain form parameter λ _iValue.

Four reference mark supposing the B-spline Curve section of band local shape parameter are respectively p _I-1, p _i, p _I+1, p _I+2, come match unique point Q with these four reference mark _i, Q _I+1Between curve, and the data point on the curve is d between two unique points _k(k=1,2,3 ... m).

Utilize least square method to list equation

Obtain the local shape parameter lambda then _iValue, t wherein _k(k=1,2,3 ... m) be step-length, adopt the chord length that on average adds up to estimate parametric t here _k, t _k∈ [0,1], initial t ₀=0, t _m=1, t _K+1-t _k=c (c is a constant).According to Here forcing the definition divisor is that 0 equation result is 0, can obtain proper λ _iSo just, obtained making the matched curve profile to have the form parameter value of least error, this also is an automatic calculation form parameter, makes parameter reach adaptive process.

Described step (5) is, according to the fitting formula (4) of B-spline curves,

$C_{i,3}({\mathrm{\λ}}_i,t)=\underset{i=1}{\overset{n-2}{\mathrm{\Σ}}}\underset{j=0}{\overset{3}{\mathrm{\Σ}}}{p}_{i+j-1}{b}_j\left(t\right)---\left(4\right)$

In the formula, C _{I, 3}(λ _i, t) be the belt shape parameter lambda _iThe segmentation B-spline Curve, carry out the B-spline Curve match in conjunction with control point information that obtains in the above-mentioned steps and self-adaptation form parameter value, generate the contour curve of inscriptions on bones or tortoise shells font.

Because the technique scheme utilization, the present invention compared with prior art has following advantage:

1. the present invention is by the feature point extraction to inscriptions on bones or tortoise shells figure, utilize the B-spline Curve match, realize the automatic generation of inscriptions on bones or tortoise shells font, solved artificial extraction inscriptions on bones or tortoise shells type matrix a large amount of labours of process consumption and time, and the not high problem of precision of human factor influence.

2. the inscriptions on bones or tortoise shells character contour curve that utilizes the present invention to obtain greatly reduces the shared space of storage in the process of preserving the inscriptions on bones or tortoise shells, accelerated the reading speed of font file.

3. this method is applied to the generation of inscriptions on bones or tortoise shells font, more helps reducing curve and dwindle the font distortion that produces because of amplification.

Description of drawings

Fig. 1 is the method flow diagram of the embodiment of the invention one;

Fig. 2 is the former word graph of the inscriptions on bones or tortoise shells among the embodiment one;

Fig. 3 carries out pretreated synoptic diagram to Fig. 2;

Fig. 4 is the synoptic diagram after the binaryzation of " dragon " word among the embodiment one;

Fig. 4 is the original contour curve synoptic diagram of " dragon " word among the embodiment one;

Fig. 5 is the synoptic diagram after the feature point extraction of " dragon " word among the embodiment one;

Fig. 6 is among the embodiment one, the unique point synoptic diagram on the curved profile;

Fig. 7 is a reference mark synoptic diagram of crossing the cubic B-spline matched curve of the unique point on the virgin curve among the embodiment one;

Fig. 8 is an inscriptions on bones or tortoise shells curved profile fitting result synoptic diagram among the embodiment one.

Embodiment

Below in conjunction with drawings and Examples the present invention is further described:

Embodiment one: at first to the original rubbing picture of the inscriptions on bones or tortoise shells carry out denoising, image segmentation, image pretreatment operation such as level and smooth is handled inscriptions on bones or tortoise shells image, obtains the bianry image of inscriptions on bones or tortoise shells literal; And then carry out profile and extract and follow the tracks of, obtain the point discrete data; Choose needed characteristic point information in the curve fitting process by the decision rule of setting at last, these unique points have reflected the feature of the wide curve of inscriptions on bones or tortoise shells character wheel, are used for the initial fragment of curve fitting.Entire method mainly can be divided into following step, and process flow diagram as shown in Figure 1.

1. image segmentation

Employing is based on the image partition method of iteration threshold method.And carry out binaryzation according to the threshold value that self-adaptation produces, and it is 0 or 255 that the literal target that is about to split is converted into pixel value, and wherein the stain of 0 expression constitutes inscriptions on bones or tortoise shells word target, and the white points of 255 expressions constitute backgrounds.After image segmentation and binaryzation, inscriptions on bones or tortoise shells character image has been removed noise spot effectively, is one of indispensable step in the subsequent treatment.

The iteration threshold method is the method by iteration, selects optimal threshold adaptively.The specific algorithm step is described below:

(1) obtains the initial value T of the mean value of minimum and maximum gray-scale value in the image as threshold value ₀, T ₀Image is divided into background and target two big zones.

(2) according to T _iImage is divided into two big zones of background and target, respectively two parts is obtained average gray value, average once more as new threshold value T according to two average gray values _I+1, specific implementation is formula as follows:

$T_{i+1}=\frac{1}{2}(\frac{\underset{k=0}{\overset{T_i}{\mathrm{\Σ}}}{h}_k\×k}{\underset{k=0}{\overset{T_i}{\mathrm{\Σ}}}{h}_k}+\frac{\underset{k=T_i+1}{\overset{L-1}{\mathrm{\Σ}}}{h}_k\×k}{\underset{k=T_i+1}{\overset{L-1}{\mathrm{\Σ}}}{h}_k})---\left(5\right)$

L is as number of greyscale levels in the formula (5), h _kBe that gray-scale value is the pixel number of k.

(3) if T _I+1=T _i, then iteration finishes, otherwise i ← i+1 forwards (2) step continuation iteration to, asks for optimal threshold.

Iteration threshold method advantage is to ask for the half-tone information of not only having considered pixel in the process in the continuous iteration of threshold value, and image is divided into background and target two zones according to threshold value, like this with indirect the including of image-region gray feature to ask for the threshold value of optimal segmentation effect.Image segmentation result as shown in Figure 3.

To " dragon " word carry out after the binary conversion treatment the result as shown in Figure 4.

2. extraction profile

It mainly is the discrete point that obtains on the inscriptions on bones or tortoise shells font curve wheel profile that profile extracts: the pixel of scan image successively, if the value of pixel is 0, and its 8 consecutive point all are 0, and then the value that will put makes 255 into; Otherwise the pixel value that keeps this point.Continue retrieval, up to having retrieved all literal target areas.Sheng Xia pixel point value is that 0 point is the point on the outline line at last.

It mainly is in order to obtain through outline data and serializing after the profile extraction that profile is followed the tracks of.Pixel after scanning oracle-bone inscription contour successively and extracting, first pixel value of mark are that 0 point is p ₀If, p ₀Sign for very then stopping search, otherwise it is true that being masked as of this point is set.Find p then ₀Adjacent stain is designated as p ₁And it is set is labeled as very, continue search, to the last Sou Suo some p from this point _N+1Equal p ₀Till, { p so ₀, p ₁, p ₂... p _nIt is exactly an occluding contour.And then the search starting point move on to other places, proceed the search of next font graphics field profile, till all point all go on record.What Fig. 5 showed is that profile extracts the result who follows the tracks of, i.e. original contour curve of Huo Deing.

3. curved profile feature extraction

The curved profile feature mainly is to be showed by the unique point on the profile, as shown in Figure 6.The unique point here comprises angle point and waypoint.According to the font characteristics of the inscriptions on bones or tortoise shells own, this method employing angle Calculation Method is promptly calculated each point and continue before it point and follow-up interval k the angle that the point of putting is formed of k point at interval, and the angle of this angle is designated as the angle value of this point.Begin up to 1 A from outline line starting point S, the angle value of A is greater than predefined threshold value, and finding the some B of angle value minimum then in contour segment SA is local minimum, and so just identifying B is angle point, is exactly a ready-portioned contour segment like this from the starting point to B.Be that starting point is set out and continued to survey with B, up to all point of traversal.The starting point of profile and terminal point are designated as the waypoint of profile.

4. be with the B-spline curves approximating method of auto-adaptive parameter

This method is according to the character of band local shape parametric cubic B-spline curves, the at first counter reference mark of obtaining three Uniform B-spline curves of unique point, utilize least square method to find the solution and the local shape parameter of discrete point then apart from the total departure minimum, at last carry out the B-spline Curve match, thereby generate inscriptions on bones or tortoise shells font curved profile according to reference mark of being asked and form parameter.

Form parameter obtains by detecting the B-spline Curve fitting result in this algorithmic procedure, and this process is to detect the process of coupling automatically, so this form parameter is adaptive form parameter, and obtains good curve fitting effect.

Three Uniform B-spline of belt shape parameter:

For t ∈ [0,1], the definition following formula is the band parameter lambda _iThe allotment function of cubic B-spline:

$\left\{\begin{array}{c}{b}_0\left(t\right)=\frac{1}{6}(1-{\mathrm{\λ}}_{i}t){(1-t)}^2\\ b_1\left(t\right)=\frac{1}{6}[4-(7-{\mathrm{\λ}}_i)t^2+(4-{\mathrm{\λ}}_i)t^3]\\ b_2\left(t\right)=\frac{1}{6}[1+(2+{\mathrm{\λ}}_i)t+(5-2{\mathrm{\λ}}_i)t^2-(4-{\mathrm{\λ}}_i)t^3]\\ b_3\left(t\right)=\frac{1}{6}[1-{\mathrm{\λ}}_i(1-t)]t^2\end{array}\right.$

If given n+1 reference mark p _i(i=0,1 ..., n), also being referred to as the polygonal summit of feature, the expression formula of the B-spline Curve of belt shape parameter is:

$C_{i,3}({\mathrm{\λ}}_i,t)=\underset{i=1}{\overset{n-2}{\mathrm{\Σ}}}\underset{j=0}{\overset{3}{\mathrm{\Σ}}}{p}_{i+j-1}{b}_j\left(t\right)---\left(4\right)$

B in the formula (4) _i(t) be the allotment function of belt shape parametric cubic Uniform B-spline, t is a step-length, C _{I, 3}(λ _i, t) be the belt shape parameter lambda _iThe segmentation B-spline Curve.

The anti-B-spline Curve reference mark of asking:

Group of feature point Q on the known curve _i(i=1,2 ..., n), look for a B-spline Curve C _{I, 3}(t) cross unique point Q _i(i=1,2 ..., n), promptly look for one group of B batten control polygon vertex p corresponding with unique point _i(i=0,1 ... n+1), the so-called anti-B-spline Curve reference mark problem of asking that Here it is.

For B-spline Curve, the relation between the data point on it and the position vector at reference mark:

C _i，3(0)＝(p _i-1+4p _i+p _i+1)/6＝Q _i，i＝1，2，3，…，n-1，n

Work as p ₀=p _n, p _N+1=p ₁The time, the matched curve of cubic B-spline is sealed, then by reference mark p _i(i=0,1 ..., n, n+1) system of linear equations of the cubic B-spline closed curve of match generation is:

Calculating can be appeared the unique point Q on the virgin curve _i(i=1,2 ..., the reference mark p of cubic B-spline matched curve n) _i(i=0,1 ..., n, n+1), as shown in Figure 7.

In curve fitting process, tend to produce some deviations, and the approximation ratio of curve fitting and deviation close ties.Weighing the most frequently used a kind of method of approximation ratio is to get approximating curve C (t _k) and virgin curve on data point d _kBetween sum of square of deviations, promptly total departure is expressed as:

$\mathrm{\φ}=\underset{k=1}{\overset{m}{\mathrm{\Σ}}}{|C\left(t_k\right)-d_k|}^2$

Least square method is exactly the function will obtain to make total departure φ reach minimal value the time, as the optimal approximation function.

Here three Uniform B-spline matched curve sections with the belt shape parameter are updated in the formula of least square method, by finding the solution partial derivative

Be obtained from suitable form parameter.

$\mathrm{\φ}\left({\mathrm{\λ}}_i\right)=\underset{k=1}{\overset{m}{\mathrm{\Σ}}}{|\underset{j=0}{\overset{3}{\mathrm{\Σ}}}{p}_{i+j-1}{b}_j\left(t_k\right)-d_k|}^2$

T in the formula _k(k=1,2,3 ... m) be step-length, adopt the chord length that on average adds up to estimate parametric t here _k, t _k∈ [0,1], initial t ₀=0, t _m=1, t _K+1-t _k=c (c is a constant) forces definition here

The concrete generative process of inscriptions on bones or tortoise shells font:

The B-spline Curve approximating method of band self-adaptation form parameter is applied to the generation of inscriptions on bones or tortoise shells font curved profile, and its process is as follows:

AL1. extract the unique point Q on the inscriptions on bones or tortoise shells virgin curve profile _iSet.

AL2. according to the character of three Uniform B-spline curves, with characteristic point information substitution formula (2), the anti-reference mark p that asked three B sample curves of unique point _iSet.

AL3. set form parameter λ _iBe unknown number, utilize least square method to find the solution each parameter of B-spline Curve section.

AL4. with form parameter λ _iWith control point information substitution formula (4), match generates inscriptions on bones or tortoise shells font curved profile section.

If AL5. the whole matches of inscriptions on bones or tortoise shells font primary curve profile line segment are finished, then finish, generate other curved profile section otherwise forward AL1 continuation match to.

Inscriptions on bones or tortoise shells curved profile fitting result is referring to shown in the accompanying drawing 8.

Information such as the unique point of preservation inscriptions on bones or tortoise shells font, point so that need not carry out above-mentioned image processing operations next time, directly generate inscriptions on bones or tortoise shells font by the information in these texts to text.

The inscriptions on bones or tortoise shells font storage space that generates with the present invention takies less, and opening speed is fast.The storage space that takies as accompanying drawing 4 is 179KB, and the storage space that the contour point information of accompanying drawing 5 takies is 34KB, and the shared storage space in the reference mark of accompanying drawing 6 only is 8KB still less.

Claims (4)

1. one kind has the adaptive oracle-bone inscription contour font of form parameter and generates method automatically, it is characterized in that, comprises the following steps:

(1) inscriptions on bones or tortoise shells rubbing image is carried out pre-service, described pre-service comprises the original contour curve that image is carried out binary conversion treatment, extraction font;

(2) the original contour curve that obtains according to step (1) extracts the unique point on the original contour curve, and described unique point comprises angle point and waypoint; The extracting method of described angle point is, calculate each point and the formed angle of line that continues before it at interval between line between k the point of putting and this point and follow-up interval k the point of putting, the angle of this angle is designated as the angle value of this point, begin to search for successively up to the some A of angle value from profile spring of curve S greater than predefined threshold value, in contour segment SA, find the some B of angle value minimum, sign B is an angle point, with this angle point B is that starting point repeats said process to identify next angle point, up to all point of traversal; The starting point of contour curve and terminal point are waypoint; Wherein, the k value is 10, and the threshold value of angle is 160 °.

(3) with B-spline Curve match contour curve, according to the counter reference mark of asking B-spline Curve of characteristic point information of step (2) acquisition;

(4) utilize least square method to find the solution and the corresponding discrete point of original contour curve form parameter value apart from the cubic B-spline basis function of total departure minimum;

(5) carry out the B-spline Curve match according to the reference mark of step (3) acquisition and the form parameter value of step (4) acquisition, obtain required inscriptions on bones or tortoise shells character contour curve.

2. according to claim 1 have the adaptive oracle-bone inscription contour font of form parameter and generate method automatically, it is characterized in that: in the described step (3), at first go out system of equations (1) by the B-spline Curve section of unique point according to the latent structure of B-spline Curve

In the system of equations (1), Q _i(i=1,2 ..., n) be unique point on the original contour curve, p _i(i=0,1 ..., n n+1) is the reference mark of B-spline Curve, the numerical value of substitution unique point is obtained the value at reference mark.

3. according to claim 1 have the adaptive oracle-bone inscription contour font of form parameter and generate method automatically, it is characterized in that: in the described step (4), and form parameter value λ _iAcquisition process be:

$\mathrm{\φ}\left({\mathrm{\λ}}_i\right)=\underset{k=1}{\overset{m}{\mathrm{\Σ}}}{|\underset{j=0}{\overset{3}{\mathrm{\Σ}}}{p}_{i+j-1}{b}_j\left(t_k\right)-d_k|}^2---\left(2\right)$

In the formula (2), φ (λ _i) be matched curve section and discrete point sequence d _kBetween square distance and, the number of m discrete point, t _k(k=1,2,3 ... m) be step-length, adopt the chord length that on average adds up to estimate parametric t here _k, t _k∈ [0,1], initial t ₀=0, t _m=1, t _K+1-t _k=c, c are constant;

Formula (3) is a belt shape parameter lambda in the formula (2) _iThe basis function b of cubic B-spline _j(t):

$\left\{\begin{array}{c}{b}_0\left(t\right)=\frac{1}{6}(1-{\mathrm{\λ}}_{i}t){(1-t)}^2\\ b_1\left(t\right)=\frac{1}{6}[4-(7-{\mathrm{\λ}}_i)t^2+(4-{\mathrm{\λ}}_i)t^3]\\ b_2\left(t\right)=\frac{1}{6}[1+(2+{\mathrm{\λ}}_i)t+(5-2{\mathrm{\λ}}_i)t^2-(4-{\mathrm{\λ}}_i)t^3]\\ b_3\left(t\right)=\frac{1}{6}[1-{\mathrm{\λ}}_i(1-t)]t^2\end{array}\right.---\left(3\right)$

According to

Here force definition

Find the solution form parameter λ _i, promptly obtain form parameter λ _iValue.

4. according to claim 1 have the adaptive oracle-bone inscription contour font of form parameter and generate method automatically, it is characterized in that: described step (5) is, according to the fitting formula (4) of B-spline curves,

$C_{i,3}({\mathrm{\λ}}_i,t)=\underset{i=1}{\overset{n-2}{\mathrm{\Σ}}}\underset{j=0}{\overset{3}{\mathrm{\Σ}}}{p}_{i+j-1}{b}_j\left(t\right)---\left(4\right)$

In the formula, C _{I, 3}(λ _i, t) be the belt shape parameter lambda _iThe segmentation B-spline Curve, carry out the B-spline Curve match in conjunction with control point information that obtains in the above-mentioned steps and self-adaptation form parameter value, generate the contour curve of inscriptions on bones or tortoise shells font.

Images