CN104166748B - Forest stand growth modeling method based on relation model - Google Patents

Abstract

The invention relates to a forest stand growth modeling method based on a relation model. The method includes the following steps that (1) tree trunks are established, and the trunks are considered as models respectively formed by a single cone frustum or multiple cone frustums; (2) bending of the trunks is controlled, the bending degree of the trunks is controlled by adjusting the relative positions of the cone frustums, the shapes of the tree trunks are fit into a parabola set, and every two adjacent cone frustums are processed in a smooth splicing mode; (3) the angles of elevation of branches are controlled, the branches are set on the trunks, the angles of elevation of the branches are set to obey the normal distribution law, and the initial angles of elevation of the branches are determined through a probability density function and a random function; (4) the azimuth angles of the branches are controlled, and the azimuth angles of the branches are controlled through a chaos function; (5) leaves are drawn, the leaves are set on the branches, and image vectorization and triangularization subdivision or leave photo transparency efficient processing are adopted; (6) a tree growth model is established; (7) a forest stand growth model is established. Compared with the prior art, the forest stand growth modeling method has the advantages that modeling is performed based on a small amount of actually measured data of trees, trees and forest stands can be vividly simulated while efficient and fast modeling is achieved.

Description

A kind of Stand Growth modeling method based on relational model

Technical field

The present invention relates to a kind of Stand Growth modeling method based on relational model, can be used to simulate the software of Stand Growth Realize.

Background technology

Since 21 century, with the high speed development of the subjects such as computer, network, remote sensing technology and 3D information technologys, these Technology also progressively becomes the study hotspot of forest-science and application.The proposition of " Digital Forestry " and " wisdom forestry " construction strategy, The step of Forestry informationlization is even more greatly accelerated, Management offorestry is sent out towards business decision direction from simple information management Exhibition.The dynamic analog of Stand Growth three dimensional structure can describe three-dimensional visualization figure during standing forest dynamic change, three-dimensionally instead Spatial framework, the forest land utilization obstacle of standing forest are reflected, more intuitively the dynamic change in simulating reality stand growth process.Can be more In detail, more accurately, more simple and clearly reflect stand information, scientifically to evaluate orest management effect, forest is made artificial Control is lower to play maximum economic, ecological, social and cultural benefit, and people implement operation and management to forest, improve orest management Intensive degree, give full play to the benefit of forest.Stand Growth three dimensional structure dynamic analog can help people's research Forest Growth The scientific law of evolution, so as to customize correct, rational forest management plan, with important scientific meaning and wide application Prospect.

Research of the recent domestic to Digital Forestry correlation technique, with the maturation of each correlation technique, has obtained quick Development, and occur in that a series of valuable theoretical research result.Stand simulation and visualization are mainly from mathematical model at present Start with terms of graphics, standing forest is explained.Such as, in terms of mathematical model, lot of domestic and international silviculturist is from mathematical model Angle angularly launches research to the Spatial Distribution Pattern of standing forest, arboreal growth, summarizes many empirical equations and growth prediction Model；Since the eighties in 20th century, visual software both domestic and external develops into trees from simple graph visualization and standing forest can Depending on changing emulation, but these software major parts are all the displaying standing forest entirety on the basis of graphics.

For the visual research of Stand Growth in digital forest resources, the following bottleneck problem is presently, there are：It is right The traditional method of Tree vigorous degree in standing forest, such as based on the tree modeling method of geometric parameter in, it is the most frequently used have L- systems and Iterated function system (IFS) method etc..The research of these methods achieves some achievements, also demonstrates their feasibility and has Effect property.But, due to the restriction of correlation theory and algorithm, can only be a kind of approximate mould to natural tree using these methods Intend, particularly to trees of the same race, the model of generation is often that general character is had a surplus and individual character is not enough.With the trees based on geometric parameter Modeling method is compared, and employing reduces the dependence to tree structure data message, but the party based on the tree modeling method of image Method is that the two dimensional image based on trees is modeled, and needs to carry out image the analyzing and processing of complexity, branch is partitioned into from image, and Extract effective information to be modeled, this process is very difficult, and the effectiveness of information that obtains and accuracy be difficult to control to, The model accuracy and validity of structure be not high.The trees geometric model that the research of Forest Growth three-dimensional modeling is adopted is excessively simple, Lack the three-dimensional modeling algorithm of specialty, it is impossible to which dynamically simulating forest grows evolutionary process in three-dimensional scenic, and interaction is not By force；Existing research work is the research to Single-issue mostly, does not form the solution of system；It is constructed without into specialty Software platform, the professional software system or functional subsystem based on low level development are very few and immature, it is difficult to form the soft of practicality Part system, instructs forest production and manages；Forestry technology is high not enough with the conjugation of information technology and degrees of fusion, is provided simultaneously with The researcher phoenix feathers and unicorn horns of the deep profession basis of two technologies, or the communication exchange between different majors researcher is limited, hinders The further investigation of correlation technique and development；For the research of prediction of stand growth is concentrated mainly on theoretic, that is, use chart The prediction increment overall to represent standing forest each somatomedin, the growing state rare research to individual plant trees in standing forest, It is not vivid enough so to the prediction of Stand Growth Model.If the growth predicted by individual plant trees, competition must be just studied Situation, but competition factor complexity is various, in addition up to hundreds of, very big difficulty is brought to prediction of stand growth modeling.Towards state The great demand of industry of family's Construction for Forest Information, breaks through above-mentioned bottleneck and is extremely necessary.

The content of the invention

It is an object of the invention to overcome above-mentioned deficiency present in prior art, and a kind of reasonable in design is provided, Step is few, it is easy to accomplish the Stand Growth modeling method based on relational model.

The present invention the adopted technical scheme that solves the above problems is：It is somebody's turn to do the Stand Growth modeling side based on relational model Method, its characteristic are to comprise the steps：

(1) tree limb is built, branch is considered as the model being made up of single or several frustum cone structures；

(2) branch bending is controlled, the degree of crook of branch is controlled by the relative position of several round platforms described in adjustment, By the form fit of tree limb into parabola set, smooth splicing between adjacent round platform, is done；

(3) elevation angle of branch is controlled, branch is set on branch, the elevation angle of branch is set as into that Normal Distribution is advised Rule, and the initial elevation of branch is determined by a probability density function with reference to random function；

(4) azimuth of branch is controlled, and the azimuth of branch is controlled using chaotic function；

(5) leaf is drawn, leaf is set on branch, leaf is set on branch and is cutd open with trigonometric ratio using image vector Divide or leaveves photo transparence efficient process；

(6) Trees growing models are set up；

(7) set up Stand Growth Model.

The building process that step (6) of the present invention sets up Trees growing models is, by the diameter of a cross-section of a tree trunk 1.3 meters above the ground respectively by different growing stages Branch bottom radius, branch top radius, branch be long, sub- branch number as space vector, obtain a series of space vectors, using the diameter of a cross-section of a tree trunk 1.3 meters above the ground as ginseng Number, is fitted with high order Bezier curve, build between each part of trees and each part and time curvilinear equation.

The step of step (7) of the present invention sets up Stand Growth Model, first obtains such as mean DBH increment, mean stand height etc. and puts down Equal data, then and press " powerhouse is stronger, and weak person is weaker " competitive Principle, the average increment of standing forest is assigned on each tree wood.

Increment distribution principle of the present invention is to be distributed increment in each tree by competitive index and competitiveness.

The elevation angle of step (3) branch growth of the present invention refers to the angle with parent branch, is generally adapted to normal distribution law, Concentrate on 0-90 ° it is interval, with normal distyribution function as branch elevation angle probability density：

By taking Lignum seu Ramulus Cunninghamiae Lanceolatae as an example, θ₀=35 represent that branch elevations angle probability of occurrence near 35 ° is higher, and the appearance for deviateing the angle is general Rate is gradually reduced；θ ∈ [0,75] represent branch elevation limits, and δ and a is the characteristic parameter at the different tree species branch elevation angle, this In Lignum seu Ramulus Cunninghamiae Lanceolatae take δ=1 and a=0.5 because rule is compared in the distribution of the Lignum seu Ramulus Cunninghamiae Lanceolatae branch elevation angle, and concentrate near 35 °.

If θ step-lengths θ_i+1-θ_i=5,15 numbers can be extracted in [0,75], (I) formula is substituted into respectively and is calculated each θ_iIt is corresponding ε_i, then s is gone out by following formula computer_i

Wherein (i=1,2,3 ... 15)

As θ_iThe number of times of appearance, s_iθ is represented for 0_iOccur without, s_iθ is represented for 1_iOccur 1 time, s_i>=2 expression θ_iOccur Repeatedly, the θ that then will appear from_iOrder lists an one-dimension array, such as L in_ij, then just there are continuous 5 θ in the array₅。

The method not only can calculate the elevation angle of branch growth, can also calculate the probability of branch appearance.

Step (5) of the present invention draws leaf, arranges leaf and can select one of following two methods on branch：Base Leaveves method for drafting in the leaveves method for drafting of image vector or based on images transparentization process.

Step (4) of the present invention, controls the azimuth of branch using Logistic functions.

When step (6) of the present invention sets up Trees growing models, Bezier curve is fitted method and is, five times Bezier curve equation is

Wherein,t∈[0,1]；V_kFor control point set, wherein k=0,1,2,3,4,5.One Individual three dimensions point set P={ p₀, p₁..., p_n, it is fitted with five Bezier curves, then spatial data points and curve should be made Distance it is minimum.From the property of Bezier curve, curve negotiating vector p₀With vectorial p_n, have：V₀=p₀,V₅=p_n。

Spatial point is defined as follows to the side-play amount of curve：

d_i=[Q (t_i,V)-P_i]^T·[Q(t_i,V)-P_i] (IV)

Space point set can be expressed as to the total drift amount of one section of curve

When one space point set is fitted to five Bezier curve Q (t, V), to make the distance of point set to curve minimum, To D (t, V) minimizing, the control point V1 of curve, V2, V3, V4 can be solved.Solution procedure is as follows：

V is asked respectively to D (t, V)₁,V₂,V₃,V₄Partial derivative, and make which be equal to 0.

(VI) formula is designated as into BV=P, control point set V=B is solved^-1P substitutes into (III) formula, obtains five Bezier being fitted Curvilinear equation.

Step (7) of the present invention sets up Stand Growth Model, can obtain same diameter of a cross-section of a tree trunk 1.3 meters above the ground Lignum seu Ramulus Cunninghamiae Lanceolatae not according to following methods Same morphosiss, the local coefficient of (1) adjusting parameter equation as needed obtain new tree modelling, and (2) are by reconfiguring The mode of branch obtains new tree modelling.

The present invention compared with prior art, with advantages below and effect：Based on the modeling of trees measured data, can be most Vivid simulation trees.How a small amount of measured data is utilized, it is present invention research ancestor that efficiently can accurately build tree modelling again Purport.The present invention is directed to concrete tree structure attribute, based on the trees data surveyed, using innovation or improved branch and tree Leaf modeling algorithm, by the method for control parameter, variable and attribute, realizes the various form knots of flexible, convenient, fast structure trees Structure threedimensional model.Experiment shows that method proposed by the present invention is different from traditional method, and " general character has to solve the tree modelling of structure Remaining individual character is not enough " problem, modeling algorithm complexity is greatly reduced, modeling efficiency and sense of reality are significantly improved.

The present invention is predicted using the Stand Growth equation and algorithm of innovation for the prediction of Stand Growth, and root According to forestry correlation priori, simplify and optimize standing forest data model, novelty proposes that a kind of standing forest increment is assigned to standing forest In every plant of trees increment algorithm, both avoided between trees complicated competitive model research, caused standing forest again by growing Rule grows, meanwhile, the research to prediction of stand growth is no longer focused only in theoretic, but lifts actual modeling layer Face.The present invention is by innovating and improving standing forest and arboreal growth modeling algorithm so that the structure of Stand Growth three-dimensional scenic with it is dynamic State simulation Stand Growth evolutionary process can efficiently, smoothness on common computer realize, the present invention research and development Stand visualization Modeling, complexity is low, efficiency high, effective strong, and the trees of establishment and standing forest model sense of reality are higher, and system has primary Intelligent and good interaction modeling ability.

Description of the drawings

Fig. 1 a are that the curved aspect graph of parabolic concave of branch after structure bends is shown in figure.

Fig. 1 b are that parabola head and the tail line obtains branch male bend aspect graph as mirror transformation in Fig. 1 a.

The diameter of a cross-section of a tree trunk 1.3 meters above the ground of Fig. 2 Even-aged pure stands changes over scattergram.

Specific embodiment

The present invention is described in further detail below in conjunction with the accompanying drawings and by embodiment, and following examples are to this Bright explanation and the invention is not limited in following examples.

Embodiment 1.

First, Stand Growth Model is set up, first has to set up the model of individual plant trees.Set up the model step of individual plant trees such as Under.

1st, assume known to a certain bar branch a length of d of branch, ramose root radius be R, branch top radius be r, the branch in not considering In the case of bending, branch is made up of 1 round platform.

2nd, the degree of crook of branch is adjusted, branch need to be divided into n contour round platform when process is curved, by calculating The bending shape and degree of method control branch, and smooth n round platform of splicing constitutes branch bending die.Tree limb integrally bending Form is similar to parabola, if the central coordinate of circle initial value for constructing branch round platform is (x_i,y_i,z_i), it is efficient and compared with accurate simulation branch Dry degree of crook, describes x with the following parabolic equation for simplifying_iAnd y_iRelation, i.e. the bending die of branch：

y_i=x_i ²/ a (i=0,1,2,3....n) is (VII)

The z in the round platform center of circle_iCoordinate can be arbitrary value, according to x_i、y_iObtain with branch azimuthal angle calculation.

If a length of L of certain branch, ramose root radius is R, and branch top radius is r, is made up of n contour round platform, then each height For d=L/n, the center of circle x coordinate of round platform is x_i=id, it is assumed that take d=1, n=8, a=20, has：

X=[0 1234567 8]

(VII) formula of substitution is obtained:

Y=[0 0.05 0.2 0.45 0.8 1.25 1.8 2.45 3.2]

The curved aspect graph of parabolic concave that branch after bending is built by x, y is shown in Fig. 1 a, and parabola head and the tail line in Fig. 1 a is made Mirror transformation obtains branch male bend aspect graph and sees Fig. 1 b.Fig. 1 shows two kinds of grown form figures of trees bending process, two kinds of bendings Form occurs or cross occurrence in different trees or respectively, according to different tree species, controlled with percentage ratio two kinds it is differently curved The probability of occurrence of form, and the degree of the controllable branch bendings of parameter a.By taking Youxian County's Huang Feng Qiao state-owned forest farms Lignum seu Ramulus Cunninghamiae Lanceolataes as an example, branch Male bend and recessed bending state probability of occurrence are 75% and 25%, a value random number, wherein h=d in [hn, 4hn] respectively.

Radius of the round platform i-th stitching portion is obtained by following formula approximate calculation：

After bending process per the height of pitch circle platform it is：

For the drafting and splicing of round platform, all the time using in circle center line connecting plane upper drawing circle of going to the bottom on round platform, use Triangular plate connects upper and lower bottom garden and generates round platform, and realizes the smooth splicing of round platform.Compare tetragon to connect to form round platform, many The methods such as side shape mesh algorithm, fan-shaped filling breach, this method not only efficiently solves round platform splicing gap problem, and makes structure The branch model built has more multiformity, controllability and motility, obtains preferable visual effect and sense of reality.

3rd, the branch elevation angle is controlled, the elevation angle of branch growth refers to the angle with parent branch, is generally adapted to normal distribution law, collection In it is interval at 0-90 °, with normal distyribution function as branch elevation angle probability density：

By taking Lignum seu Ramulus Cunninghamiae Lanceolatae as an example, θ₀=35 represent that branch elevations angle probability of occurrence near 35 ° is higher, and the appearance for deviateing the angle is general Rate is gradually reduced；θ ∈ [0,75] represent branch elevation limits, and δ and a is the characteristic parameter at the different tree species branch elevation angle, this In Lignum seu Ramulus Cunninghamiae Lanceolatae take δ=1 and a=0.5 because rule is compared in the distribution of the Lignum seu Ramulus Cunninghamiae Lanceolatae branch elevation angle, and concentrate near 35 °.

If θ step-lengths θ_i+1-θ_i=5,15 numbers can be extracted in [0,75], (I) formula is substituted into respectively and is calculated each θ_iIt is corresponding ε_i, then s is gone out by following formula computer_i

As θ_iThe number of times of appearance, s_iθ is represented for 0_iOccur without, s_iθ is represented for 1_iOccur 1 time, s_i>=2 expression θ_iOccur Repeatedly, the θ that then will appear from_iOrder lists an one-dimension array, such as s in_i=5, then just have continuous 5 θ in the array₅.Most Element is taken out at random as the elevation angle of branch from this array afterwards, and the method had both met the control of the branch elevation angle and suited trees knot Structure reality, is easy to the quick realization of modeling again, compares Inverse Transform Sampling Method and Box The methods such as Muller, are greatly lowered the complexity of branch modeling.

4th, branch azimuth control, the azimuth coverage that branch grows are 0 ° to 360 °, by taking Lignum seu Ramulus Cunninghamiae Lanceolatae as an example, same position Branch's number that (i.e. away from identical under branch) grows is few, is presented random uniform symmetrical.If using common random Function process, the azimuth of generation is not exclusively controllable.In order to avoid such case, the present invention is mixed using Logistic mappings Ignorant function replaces common random function.Azimuth azimuth=360*logistic (μ₀, x₀), μ₀, x₀For initial value, Logistic mappings, in chaos state, are that aperiodicity does not restrain, and to initial value μ and X₀Extremely sensitive, different is first Even if initial value difference is very little, the random sequences for obtaining also can be entirely different, as long as and initial value is identical, the random sequences for obtaining Also it is just identical.After obtaining random sequences, as long as random sequences to be all multiplied by 360 azimuths that can just obtain randomly generating.

5th, leaf is set on branch, the upper corresponding leaveves of tree limb model addition of structure could more vertical cut ground void Intend tree modelling.There is homoplasy with the leaveves of seeds, although we can set up sense of reality very by trial curved surface modeling method Strong various leaveves body Models, but due on trees leaveves data volume it is huge, this method will be set up and dynamic virtual in real time Trees and standing forest application model are unpractical.The present invention proposes the method that two kinds efficient and more accurate leaveves are drawn, using with Insertion azimuth and the elevation angle of the machine Variable Control leaveves in branch relevant position, generate tree modelling.

(1) the leaveves method for drafting based on image vector

The method had previously carried out vectorized process to every kind of leaveves image, generated the seeds leaveves vector array, supplied Tree vigorous degree system is called, and draws leaveves with reference to parameter control.

The high-definition image for obtaining the leaveves of certain seeds is easy, need to only adopt uniform in color and two to leaveves original image Value is processed；By being scanned extraction leaveves profile to bianry image in length and breadth；Based on mathematical morphology, carry out crack bridge joint, Expansion, refinement and deburring are processed, obtain pixel it is wide can characterize region shape feature contour skeleton.

Vectorized process process to leaveves, each key point that leaveves profile is found out (i.e. flex point), takes up an official post from profile Meaning is a little set out, and calculates the slope k 1 of the point and the 5th pixel line, and the slope k 2 of the point and the 9th pixel line, If abs (k1-k2)>D, then be flex point at the 5th pixel, retain the point, and otherwise the point does not retain, then with the 5th picture Vegetarian refreshments is starting point, repeats said process, until finding all flex points.It is available not by the spacing and d values that control capture element With the leaveves vectogram of flex point number precision.It is after obtaining building polygonal summit (flex point) array of leaveves, point-to-points based on pushing up Side shape carries out the algorithmic procedure of triangulation, the leaveves tri patch of generation and coloring effect, the tree modelling effect of structure.

(2) the leaveves method for drafting based on images transparentization process：For further improve leaveves drafting efficiency, using base In the leaveves string method for drafting of image, leaveves string photo Vitrification management is added on branch model, tree can be greatly improved Wooden modeling efficiency.By taking Lignum seu Ramulus Cunninghamiae Lanceolatae leaf as an example, concrete grammar is as follows：

1. the photo of a Lignum seu Ramulus Cunninghamiae Lanceolatae leaf is chosen, is that photo adds Alpha passages using Photoshop instruments, and is saved as .dds form.

2. add planar rectangular in the branch orientation for needing to add leaf, and transparent processing is carried out to planar rectangular.

3. using the leaf photo handled well as texture mapping to planar rectangular.

Add the fir type that leaf builds with the method.The leaveves drawn using this succinct method are not only significantly reduced Modeling complexity, moreover it is possible to obtain preferable visual effect.Experiment shows that most trees can all adopt leaveves string picture processing Method is building the leaveves of its model.

2nd, after individual plant tree modelling is set up, it is necessary to set up arboreal growth relational model.

By taking the relational model in structure Lignum seu Ramulus Cunninghamiae Lanceolatae branch growth course as an example, by the diameter of a cross-section of a tree trunk 1.3 meters above the ground respectively by the branch bottom of different growing stages half Footpath, branch top radius, branch are long, sub- branch number as space vector, obtain a series of space vectors, using diameter of a cross-section of a tree trunk 1.3 meters above the ground t as parameter, use Three times or high order Bezier curve are fitted, and build its curvilinear equation.

By taking the relational model of five Bezier curve fitting branch numbers, the diameter of a cross-section of a tree trunk 1.3 meters above the ground and Tree height growths as an example, method is as follows：

Five times Bezier curve equation is

Wherein,t∈[0,1]；V_kFor control point set, wherein k=0,1,2,3,4,5.One Individual three dimensions point set P={ p₀, p₁..., p_n, it is fitted with five Bezier curves, then spatial data points and curve should be made Distance it is minimum.From the property of Bezier curve, curve negotiating vector p₀With vectorial p_n, have：V₀=p₀,V₅=p_n。

Spatial point is defined as follows to the side-play amount of curve：

d_i=[Q (t_i,V)-P_i]^T·[Q(t_i,V)-P_i] (IV)

Space point set can be expressed as to the total drift amount of one section of curve

When one space point set is fitted to five Bezier curve Q (t, V), to make the distance of point set to curve minimum, To D (t, V) minimizing, the control point V1 of curve, V2, V3, V4 can be solved.Solution procedure is as follows：

V is asked respectively to D (t, V)₁,V₂,V₃,V₄Partial derivative, and make which be equal to 0.

(VI) formula is designated as into BV=P, control point set V=B is solved^-1P substitutes into (III) formula, obtains five Bezier being fitted Curvilinear equation.

As a example by building Growth of Chinese Fir relational model, by the age of tree using the diameter of a cross-section of a tree trunk 1.3 meters above the ground, sub- branch number and the height of tree as space right-angle Three-dimensional coordinate point set (X, Y, Z), obtains a series of spatial spreading coordinate points, and parameter t ∈ [0,1] is considered as linear pass with age of tree T System has：T=at+b, a, b are that coefficient is determined with the age of tree is terminated by the initial of Lignum seu Ramulus Cunninghamiae Lanceolatae.Growth relationship model is built as stated above, Both the relation of the Lignum seu Ramulus Cunninghamiae Lanceolatae growth process diameter of a cross-section of a tree trunk 1.3 meters above the ground, sub- branch number and height of tree life had been embodied, the growth relationship of they and the age of tree had also been obtained.

Similar method can build the growth relationship model of secondary branch number, the branch diameter of a cross-section of a tree trunk 1.3 meters above the ground and length, can also build tree The single morphological parameters of wood and time dependent model.Used as Timber stands trees, its value is mainly reflected in the trunk diameter of a cross-section of a tree trunk 1.3 meters above the ground to Lignum seu Ramulus Cunninghamiae Lanceolatae In the height of tree, the diameter of a cross-section of a tree trunk 1.3 meters above the ground and the height of tree determine sub- branch sum of series quantity and sub- branch size, from the reality of Growth of Chinese Fir visual modeling Considered with property and economy point, the details without the need for excessively focusing on sub- branch structure change, the Lignum seu Ramulus Cunninghamiae Lanceolatae of the different trunk diameters of a cross-section of a tree trunk 1.3 meters above the ground generally have The sub- branch structured data of certain rule, the morphosiss of its sub- branch completely can by empirical data combine the control of parameter come Generate.For concrete tree kind is in modeling process, on the premise of the diameter of a cross-section of a tree trunk 1.3 meters above the ground, long and total base of branch this determination, branch bottom and branch top Away from, sub- branch number under radius, branch, and the amplitude of variation such as the elevation angle, azimuth, flexibility also can be tied by the empirical parameter of the seeds Close parameter to automatically generate, the such as series of Lignum seu Ramulus Cunninghamiae Lanceolatae its sub- branch of a certain diameter of a cross-section of a tree trunk 1.3 meters above the ground, and the quantity per Their First Branch has corresponding than row Relation and excursion.Thus the configuration and symbol of more diversified change can under data cases as few as possible, be constructed Close the tree modelling of the seeds growth rhythm.

3rd, set up Stand Growth relational model.

One standing forest model is huge comprising quantity of information, and this also determines the complexity of standing forest modeling.Prediction of stand growth mould Type can only obtain the general characteristic of standing forest, such as mean DBH increment, mean stand height etc., it is impossible to directly obtain in each tree growth course The structural informations such as the diameter of a cross-section of a tree trunk 1.3 meters above the ground, the height of tree, thus need the average increment of standing forest is assigned on each tree wood to Stand Growth modeling.

The similar method of arboreal growth relational model is built with section, it is average by standing forest sample ground Lignum seu Ramulus Cunninghamiae Lanceolatae difference age of tree periods The diameter of a cross-section of a tree trunk 1.3 meters above the ground, the height of tree, hat width, as space right-angle three-dimensional coordinate point set, are fitted its five BEZIER curve linear relationship equations, are closed by setting up It is that equation combines standing forest sample ground attribute and a small amount of constraint parameter can just reconstruct all trees data messages of standing forest.This is for woods The Dummy modeling of mitogenetic length is significant, and one standing forest can be expressed with parametric equation, and its two changes parameter is obtained difference The standing forest and tree modelling of age different shape structure.

The general characteristic of the simply standing forest that prediction of stand growth is obtained, such as mean DBH increment, mean stand height etc., can not but obtain The factor informations such as the diameter of a cross-section of a tree trunk 1.3 meters above the ground of every plant of trees, the height of tree.In modeling process, if all giving full standing forest to every plant of trees in standing forest Mean DBH increment and the height of tree, then have bigger difference with the real conditions of forest.With full standing forest-Individual tree model, using constructed Growth equation predicts certain Graph One factor average magnitude of standing forest, such as mean DBH increment, mean stand height etc., and then amount is by certain principle or rule Then it is assigned on every plant of trees, carries out error minimize matching and Parameter fusion, obtains the optimum ginseng of Stand Growth residing for trees Exponential model, the discussion middle in a separate paper of the structure of the rule-based algorithm in author.

According to the measured data that standing forest grows year after year, data and Variable Control, the tree breast-height diameter of structure, tree are incorporated experience into High order BEZIER curvilinear equation model of the high, hat width with time parameter growth change, and the relational model between them, to spy Determine certain trees (the such as Lignum seu Ramulus Cunninghamiae Lanceolatae of hilly regions in south China or Pinus massoniana Lamb etc.) under the conditions of standing forest, with general applicability, so as to can intelligence The model of any trophophase of standing forest trees can be inferred.

The vector set data algorithm that parametric equation is obtained according to five Bezier curves and whole tree is obtained is described as follows：

Read each rank branch information (trunk information, 1 grade of sub- branch, 2 grades of sub- branches ...) of different growing stages trees, trunk The radius sequence of branch bottom is pressed in part, and branching section sorts by sub- branch number.The parameter of each rank branch is tried to achieve with method of least square Equation.The Lignum seu Ramulus Cunninghamiae Lanceolatae morphosiss data that the diameter of a cross-section of a tree trunk 1.3 meters above the ground is d are asked by parametric equation, according to data above drafting tree modelling method with more than Method is similar to.

Template of the tree modelling drawn out as same diameter of a cross-section of a tree trunk 1.3 meters above the ground Lignum seu Ramulus Cunninghamiae Lanceolatae, can obtain same diameter of a cross-section of a tree trunk 1.3 meters above the ground China fir according to following methods The different shape structure of wood：

(1) local data of adjusting parameter equation obtains new tree modelling as needed

(2) new tree modelling is obtained by way of reconfiguring branch

Wherein, only have the general characteristic of standing forest, such as standing forest area, strain number in Stand investigation element factor；What prediction was obtained And the general characteristic of standing forest, such as mean DBH increment, mean stand height etc., can not but obtain the factors such as the diameter of a cross-section of a tree trunk 1.3 meters above the ground, the height of tree of every plant of trees Information.If all giving the mean DBH increment and the height of tree of full standing forest to every plant of trees in standing forest, have with the real conditions of forest Bigger difference.The diameter structure regularity of Even-aged pure stands is generally the unimodal mountain shape similar normal state with standing forest average diameter as peak value point Cloth curve.For many years, silviculturist is fitted using normal distyribution function, describe Even-aged pure stands diameter Distribution achieve it is good Effect.According to Stand Structure Laws, either artificial forest or wildwood, the factor information such as the diameter of a cross-section of a tree trunk 1.3 meters above the ground, the height of tree in standing forest are universal Follow certain Stand Structure Laws.The diameter of a cross-section of a tree trunk 1.3 meters above the ground-the strain number of Even-aged pure stands into normal distribution, the not Thoracic sympathicectomy function of even-aged stand Also there is specific Changing Pattern.

In Fig. 2, highest normal distribution curve is that the time is most front, and it is minimum for the time it is last.Can from figure Go out, trees strain number difference of the Even-aged pure stands Thoracic sympathicectomy when starting in each footpath rank is larger, and peak value is higher, illustrates that Thoracic sympathicectomy exists It is more near average.As time goes on, Thoracic sympathicectomy normal curve integrally moves right, and trees strain number is poor in each footpath rank Do not reduce, peak value is reduced, illustrate between trees as competition is so that Thoracic sympathicectomy gap increases gradually.

The present invention proposes a kind of standing forest increment of the competitive Principle based on " powerhouse is stronger, and weak person is weaker " newly side of distribution Method.Lignum seu Ramulus Cunninghamiae Lanceolatae in the standing forest of 10 years is grown, its diameter of a cross-section of a tree trunk 1.3 meters above the ground can not possibly reach unification, but it is according to Natural Survival rule that is, competing Strive, always in normal distribution.The competitive Principle of " powerhouse is stronger, and weak person is weaker " can hold the competition shape of trees on the whole Condition, obtains overall trees increment, the impact without studying each factor comprehensively obtains increment again, it is to avoid complexity Competitive model research.For the power of trees competitiveness, the present invention is represented with the competitive index of every plant of trees.Competition refers to Number expresses the competitive pressure that forest is born.It is generally divided into two kinds:A kind of and distance dependent, it is a kind of with apart from unrelated.The present invention The simple competitive index of the diameter of a cross-section of a tree trunk 1.3 meters above the ground with distance dependent is chosen, expression formula is as follows：

Wherein, CI_iIt is the competitive index of object wood i, D_jIt is the diameter of a cross-section of a tree trunk 1.3 meters above the ground of neighboring trees j, D_iIt is the diameter of a cross-section of a tree trunk 1.3 meters above the ground of object wood i, L_ijIt is right As the distance between wooden i and neighboring trees j, n is that wooden strain number is competed around object wood i.The competitive index is that neighboring trees compete wood Competitive index to object wood, its value are bigger, and the interference suffered by object wood is bigger, then object wood is in a disadvantageous position in competition；Its Value is less, and suffered annoyance level is less, and object wood is had the advantage in competition.Therefore, the competitive index can not be represented completely Competition of the trees in standing forest is strong and weak, and the competitive index adds a certain somatomedin, as above a moment tree breast-height diameter size Its competitive index is deducted, using competitiveness of this difference as this plant of trees in whole standing forest, difference is bigger, and competitiveness is got over By force, conversely, weaker.Then the competitiveness of individual plant trees is represented by following form：

CP=y-CI Ⅺ

By taking the diameter of a cross-section of a tree trunk 1.3 meters above the ground as an example, main thought is：Diameter of a cross-section of a tree trunk 1.3 meters above the ground meansigma methodss y at required moment were that diameter of a cross-section of a tree trunk 1.3 meters above the ground meansigma methodss Y at a upper moment add Upper increment increment, i.e. Y=y+increment.Concrete assigning process is as follows：

Obtain the stand average breast diameter △ of subsequent time first according to growth equation_avg, with the average breast of the standing forest of previous moment Subtract each other the diameter of a cross-section of a tree trunk 1.3 meters above the ground average increasing amount for obtaining standing forest in footpath.By △_avgN increment is generated at random, respectively as n strains trees in standing forest Increment.As the increment of strong competitiveness trees can be higher than meansigma methodss, the increment of weak competitiveness trees can be less than average Value, therefore, it is possible to be (0, (1+slope) △ the range set of random increment_avg), slope is the required standing forest moment Growth curve slope, i.e. rate of increase.

Random increment to generating will ensure that summation will be equal to overall growth amount.Then the competitiveness of every plant of trees is calculated, The maximum increment of the trees distribution maximum for competitiveness, the trees second largest to competitiveness distribute second largest increment, By that analogy.

Embodiment 2.

The Tree vigorous degree system that thinking and method based on embodiment 1 builds, there is provided flexible and changeable characteristic parameter and change Amount control mode, can be directed to concrete seeds modeling object, can easily build the seeds each age group, variously-shaped structure Tree modelling.

(1) build trees basic model to change with model

System read need modeling seeds one group of measured data (mainly including trunk, branch series, branch number, Away from geometry and characteristic attribute informations such as, the elevation angle, azimuths under dry branch length, dry branch diameter, branch), system automatically creates the seeds Trees basis threedimensional model.By taking one group of 4 years raw Lignum seu Ramulus Cunninghamiae Lanceolatae measured data as an example.

The editor of model.

The height of system adjustable section trunk, root radius, top radius, can select bending towards and flexibility for trunk Number.

Editor's attribute of branch is more, mainly include bending, the elevation angle, azimuth, long branch, ramose root radius, branch top radius, with And under branch away from etc., and the branch of different series separately can adjust, and so increase adjustment dynamics and the motility of branch.Except Outside the morphosiss of adjustment branch, with the interactively deletion of mouse or branch can also be added.

Model it is derivative.

The tree modelling that the various morphosiss of the age bracket are derived by basic tree modelling be it is necessary, so can be significantly Reduce modeling complexity and improve modeling efficiency, system changes random parameter and variable by interactive.

The growth of model.

The growth of trees and standing forest is a complicated problem, and author build by corresponding growth of the proposition in some documents Modulo n arithmetic.The system in the actual measurement parameter of 4,6,8,10,15,20 years, carries out high order curve fitting, obtains using to a Lignum seu Ramulus Cunninghamiae Lanceolatae Relation equation and their relation equations with the time between the diameter of a cross-section of a tree trunk 1.3 meters above the ground, the height of tree, branch number, hat width etc..

Addition leaveves.

To building tree limb model, addition leaveves are that virtual reality trees are requisite, based on the seeds leaveves The leaveves macrograph handled well is adopted the mode of complete intelligence to add for branch by distribution characteristicss and position construction parameter, system Leaveves, operator can be adjusted in rational scope to leaf size and density parameter.

(2) build Stand Growth Model

The present invention by the raw Lignum seu Ramulus Cunninghamiae Lanceolatae sample in Hunan Youxian County Huang Feng Qiao state-owned forest farms Buddhist monks ridge 10 years as a example by measured data, the sample Coordinate is 27 degree 19 points of north latitude 27 seconds, 113 degree 45 points of east longitude 14.60 seconds, and size is 40x80m, sample ground totally 955 plants of Lignum seu Ramulus Cunninghamiae Lanceolataes.Experiment Using Matlab 7.0 and Visual Studio C#2008 as development environment, standing forest virtual growth model construction is carried out, realized Three dimension realistic entity Lignum seu Ramulus Cunninghamiae Lanceolatae and standing forest modeling rendering.Five Bezier to Lignum seu Ramulus Cunninghamiae Lanceolatae measured data are completed in Matlab 7.0 Curve matching builds relation equation, obtains standing forest and Lignum seu Ramulus Cunninghamiae Lanceolatae model forms structure excel by calculating, optimizing, control and derive Tables of data.The drafting of Lignum seu Ramulus Cunninghamiae Lanceolatae and standing forest model, the modeling built with reference to Direct3D by 2008 C# languages of Visual Studio System is completed, and single tree-model of foundation is converted into .X files, and system imports .X file generated standing forests according to standing forest ground sample attribute list Model.

Based on the rule that high order Bezier curve equation model trees key structural feature data change with the age of tree, by the age of tree The basic threedimensional model of the seeds trees is set up, then according to the seeds detailed structure feature and Changing Pattern, by adjusting control The variables such as the size of tree limb processed and leaf, quantity, direction, angle, spacing, the acquisition of these variables and excursion meet The growth rhythm of environmental factors residing for the seeds, sets up various morphosiss tree modellings.Test shows that the method is not only significantly The data needed for Tree vigorous degree are reduced, the model of structure is similar to natural tree, and structure change has more multiformity, sense of reality It is higher.

The present invention builds growth relationship model, has both embodied the relation of the diameter of a cross-section of a tree trunk 1.3 meters above the ground, sub- branch number and the height of tree in Growth of Chinese Fir, Also the growth relationship of they and the age of tree has been obtained.The method had both been adapted to the growth prediction modeling that single wood also is adapted for standing forest simultaneously.With Traditional simulation method, pinup picture method and fractal method etc. are compared, it is to avoid need to repeat the loaded down with trivial details of modeling for different relations, overall Algorithm complex is greatly lowered.The more easy data precision for flexibly, deriving and predicting of modeling process is higher, give birth to closer to trees Long reality.

Above content described in this specification is only illustration made for the present invention.Technology belonging to of the invention The technical staff in field can make various modifications or supplement or adopt similar mode to described specific embodiment Substitute, content without departing from description of the invention or surmount scope defined in the claims, this all should be belonged to The protection domain of invention.

Claims (8)

1. a kind of Stand Growth modeling method based on relational model, its characteristic is to comprise the steps：

(1) tree limb of Cunninghamia lanceolata (Lamb.) Hook. is built, branch is considered as the model being made up of single or several frustum cone structures；

(2) bending of Cunninghamia lanceolata (Lamb.) Hook. branch is controlled, the degree of crook of branch is controlled by the relative position of several round platforms described in adjustment, By the form fit of tree limb into parabola set, smooth splicing between adjacent round platform, is done；

(3) elevation angle of Cunninghamia lanceolata (Lamb.) Hook. branch is controlled, branch is set on branch, the elevation angle of branch is set as into that Normal Distribution is advised Rule, and the initial elevation of branch is determined by a probability density function with reference to random function；Wherein, the elevation angle of branch growth refers to With the angle of parent branch, generally be adapted to normal distribution law, concentrate on 0-90 ° it is interval, faced upward as branch with normal distyribution function Angle probability density：

$\mathrm{\ϵ}=e^{-\frac{{(\mathrm{\θ}-{\mathrm{\θ}}_0)}^2\mathrm{\δ}}{a^2}}---\left(I\right)$

Wherein, θ₀=35 represent that the branch elevation angle is 35 °, and the probability of occurrence on 35 ° of directions are deviateed is gradually reduced；θ∈[0,75] Represent branch elevation limits, characteristic parameters of the δ and a for the different tree species branch elevation angle；Here Lignum seu Ramulus Cunninghamiae Lanceolatae takes δ=1 and a=0.5, Because rule is compared in the distribution of the Lignum seu Ramulus Cunninghamiae Lanceolatae branch elevation angle, and concentrates on 35 ° of position；

(4) azimuth of Cunninghamia lanceolata (Lamb.) Hook. branch is controlled, and the azimuth of branch is controlled using chaotic function；

(5) Cunninghamia lanceolata (Lamb.) Hook. leaf is drawn, leaf is set on branch, leaf is set on branch and is cutd open with trigonometric ratio using image vector Divide or leaveves photo transparence efficient process；

(6) set up Cunninghamia lanceolata (Lamb.) Hook. growth model：Building process is, by the diameter of a cross-section of a tree trunk 1.3 meters above the ground respectively by the branch bottom radius of different growing stages, branch top radius, Long, the sub- branch number of branch obtains a series of space vectors, using the diameter of a cross-section of a tree trunk 1.3 meters above the ground as parameter, with three times or five times as space vector Bezier curve is fitted, build between each part of trees and each part and time curvilinear equation；

(7) set up Stand Growth Model：Cunninghamia lanceolata (Lamb.) Hook. growth model according to setting up sets up Stand Growth Model.

2. the Stand Growth modeling method based on relational model according to claim 1, it is characterised in that:The step (7) the step of setting up Stand Growth Model, first obtains the average data of mean DBH increment, mean stand height, then and press " powerhouse is stronger, Weak person is weaker " competitive Principle, the average increment of standing forest is assigned on each tree wood.

3. the Stand Growth modeling method based on relational model according to claim 2, it is characterised in that:Increment is distributed Principle is to be distributed increment in each tree by competitive index and competitiveness.

4. the Stand Growth modeling method based on relational model according to claim 1, it is characterised in that:The step (3) in, if θ step-lengths θ_i+1-θ_i=5,15 numbers are extracted in [0,75], (I) formula is substituted into respectively and is calculated each θ_iCorresponding ε_i, then S is calculated by following formula_i：

$s_i=\mathrm{int}\left(\frac{15{\mathrm{\ϵ}}_i}{\underset{i=1}{\overset{15}{\Σ}}{\mathrm{\ϵ}}_i}\right)---\left(II\right)$

Wherein (i=1,2,3 ... 15)；

As θ_iThe number of times of appearance, s_iθ is represented for 0_iOccur without, s_iθ is represented for 1_iOccur 1 time, s_i>=2 represent θ_iOccur repeatedly, so The θ that will appear from afterwards_iOrder arrangement is used as one-dimension array L_i；

The method can not only calculate the elevation angle of branch growth, additionally it is possible to calculate the probability of branch appearance.

5. the Stand Growth modeling method based on relational model according to claim 1, it is characterised in that:The step (5) leaf is drawn, leaf is set on branch and selects one of following two methods:Leaveves method for drafting based on image vector Or the leaveves method for drafting based on images transparentization process.

6. the Stand Growth modeling method based on relational model according to claim 1, it is characterised in that:The step (4) azimuth of branch, is controlled using Logistic functions.

7. the Stand Growth modeling method based on relational model according to claim 2, it is characterised in that:The step (6) it is that five times Bezier curve equation is that when setting up Trees growing models, Bezier curve is fitted method

$Q(t,V)=\underset{k=0}{\overset{5}{\Σ}}{B}_k\left(t\right)V_k---\left(III\right)$

Wherein,t∈[0,1]；V_kFor control point set, wherein k=0,1,2,3,4,5；One Three dimensions point set P={ P₀, P₁..., P_n, be fitted with five Bezier curves, then should make spatial data points and curve away from From minimum；From the property of Bezier curve, curve negotiating vector P₀With vectorial P_n, have:V₀=P₀, V₅=P_n；

Spatial point is defined as follows to the side-play amount of curve：

d_i=[Q (t_i, V) and-P_i]^T·[Q(t_i, V) and-P_i] (IV)

Space point set is expressed as to the total drift amount of one section of curve

$D(t,V)=\underset{i=1}{\overset{n-1}{\Σ}}{d}_i=\underset{i=1}{\overset{n-1}{\Σ}}{\[Q(t_i,V)-P_i\]}^T\·\[Q(t_i,V)-P_i\]---\left(V\right)$

When one space point set is fitted to five Bezier curve Q (t, V), to make the distance of point set to curve minimum, to D (t, V) minimizing, can solve the control point V of curve₁,V₂,V₃,V₄；Solution procedure is as follows:

V is asked respectively to D (t, V)₁,V₂,V₃,V₄Partial derivative, and make which be equal to 0；

$\left\{\begin{array}{c}\frac{\∂D(t,V)}{\∂V_1}=2\underset{i=1}{\overset{n-1}{\Σ}}\frac{\∂D(t_i,V)}{\∂V_1}\[Q(t_i,V)-P_i\]=0\\ \frac{\∂D(t,V)}{\∂V_2}=2\underset{i=1}{\overset{n-1}{\Σ}}\frac{\∂D(t_i,V)}{\∂V_2}\[Q(t_i,V)-P_i\]=0\\ \frac{\∂D(t,V)}{\∂V_3}=2\underset{i=1}{\overset{n-1}{\Σ}}\frac{\∂D(t_i,V)}{\∂V_3}\[Q(t_i,V)-P_i\]=0\\ \frac{\∂D(t,V)}{\∂V_4}=2\underset{i=1}{\overset{n-1}{\Σ}}\frac{\∂D(t_i,V)}{\∂V_4}\[Q(t_i,V)-P_i\]=0\end{array}\right.\⇒\left\{\begin{array}{c}2\underset{i=1}{\overset{n-1}{\Σ}}{B}_1\left(t_i\right)\[Q(t_i,V)-P_i\]=0\\ 2\underset{i=1}{\overset{n-1}{\Σ}}{B}_2\left(t_i\right)\[Q(t_i,V)-P_i\]=0\\ 2\underset{i=1}{\overset{n-1}{\Σ}}{B}_3\left(t_i\right)\[Q(t_i,V)-P_i\]=0\\ 2\underset{i=1}{\overset{n-1}{\Σ}}{B}_4\left(t_i\right)\[Q(t_i,V)-P_i\]=0\end{array}\right.\⇒$

$\left[\begin{array}{cccc}\underset{i=1}{\overset{n-1}{\Σ}}{\[B_1\left(t_i\right)\]}^2& \underset{i=1}{\overset{n-1}{\Σ}}{B}_1\left(t_i\right)B_2\left(t_i\right)& \underset{i=1}{\overset{n-1}{\Σ}}{B}_1\left(t_i\right)B_3\left(t_i\right)& \underset{i=1}{\overset{n-1}{\Σ}}{B}_1\left(t_i\right)B_4\left(t_i\right)\\ \underset{i=1}{\overset{n-1}{\Σ}}{B}_1\left(t_i\right)B_2\left(t_i\right)& \underset{i=1}{\overset{n-1}{\Σ}}{\[B_2\left(t_i\right)\]}^2& \underset{i=1}{\overset{n-1}{\Σ}}{B}_2\left(t_i\right)B_3\left(t_i\right)& \underset{i=1}{\overset{n-1}{\Σ}}{B}_2\left(t_i\right)B_4\left(t_i\right)\\ \underset{i=1}{\overset{n-1}{\Σ}}{B}_1\left(t_i\right)B_3\left(t_i\right)& \underset{i=1}{\overset{n-1}{\Σ}}{B}_2\left(t_i\right)B_3\left(t_i\right)& \underset{i=1}{\overset{n-1}{\Σ}}{\[B_3\left(t_i\right)\]}^2& \underset{i=1}{\overset{n-1}{\Σ}}{B}_3\left(t_i\right)B_4\left(t_i\right)\\ \underset{i=1}{\overset{n-1}{\Σ}}{B}_1\left(t_i\right)B_4\left(t_i\right)& \underset{i=1}{\overset{n-1}{\Σ}}{B}_2\left(t_i\right)B_4\left(t_i\right)& \underset{i=1}{\overset{n-1}{\Σ}}{B}_3\left(t_i\right)B_4\left(t_i\right)& \underset{i=1}{\overset{n-1}{\Σ}}{\[B_4\left(t_i\right)\]}^2\end{array}\right]\left[\begin{array}{c}{V}_1\\ V_2\\ V_3\\ V_4\end{array}\right]=\left[\begin{array}{c}\underset{i=1}{\overset{n-1}{\Σ}}{B}_1\left(t_i\right)(P_i-B_0(t_i)V_0-B_5(t_i\left)V_5\right)\\ \underset{i=1}{\overset{n-1}{\Σ}}{B}_2\left(t_i\right)(P_i-B_0(t_i)V_0-B_5(t_i\left)V_5\right)\\ \underset{i=1}{\overset{n-1}{\Σ}}{B}_3\left(t_i\right)(P_i-B_0(t_i)V_0-B_5(t_i\left)V_5\right)\\ \underset{i=1}{\overset{n-1}{\Σ}}{B}_4\left(t_i\right)(P_i-B_0(t_i)V_0-B_5(t_i\left)V_5\right)\end{array}\right]---\left(VI\right)$

(VI) formula is designated as into BV=P, control point set V=B is solved^-1P substitutes into (III) formula, obtains five Bezier curves being fitted Equation.

8. the Stand Growth modeling method based on relational model according to claim 1, it is characterised in that:The step (7) Stand Growth Model being set up, the different shape structure of same diameter of a cross-section of a tree trunk 1.3 meters above the ground Lignum seu Ramulus Cunninghamiae Lanceolatae being obtained according to following methods, (1) adjusts as needed The local coefficient of whole parametric equation obtains new tree modelling, and (2) obtain new trees mould by way of reconfiguring branch Type.