CN102855661B - Large-scale forest scene quick generation method based on space similarity - Google Patents

Abstract

The invention discloses a large-scale forest scene quick generation method based on space similarity. The method comprises the steps of 1) setting initialization parameters of a large-scale forest simulation scene, and acquiring scene distribution information of visible data and forest space of the whole scene; 2) partitioning the space data of the large-scale forest scene based on a quadtree, and dividing the scene into blocks with the same size; 3) calculating the space similarity of the forest scene; 4) judging the similarity between the scene blocks to quickly obtain biological amounts of plants; and 5) if the similarity between the scene blocks does not reach a certain proportion, calculating a growth model of the scene blocks, determining basic plants according to initial parameter data transmitted by a user, calculating an influence range and the biological amounts of the basic plants, then introducing a three-dimensional tree model into simulation for realizing the forest scene, and displaying a visible result to the user. The method is high in speed and high in precision.

Description

Based on the large-scale forest scene quick generation method of spatial simlanty

Technical field

The present invention relates to scale Forest Scene generation technique, especially a kind of scale Forest Scene generation method.

Background technology

Due to the complicacy of the general layout of scale Forest Scene, process and mutual relationship, be difficult to directly test it, adopting computing machine to set up forest growth model base is an effective approach.Forest growth model base needs to consider the dynamic growth characteristic sum environmental impact of plant, carrys out pre-measuring plants in the growth tendency in all ages and classes stage and change, and simulates Forest Growth situation.Extensive scale Forest Scene dynamic simulation not only will store the information of each trees, but also the competition that will calculate between trees interphase interaction and all trees and mutually beneficial combined influence, its computation process is complicated, calculated amount is also very huge, and the application and research of extensive Forest Growth emulation receives great restriction.

When scale Forest Scene visual, being used for the technology of visual modeling at present mainly contains level of detail model (LOD) and Image based rendering technology.Level of detail technology is the method for drafting based on geometry, can the geometric model structure of expression tree well.Image based rendering technology is the Real-Time Rendering that the another kind of people's extensive concern in recent years realizes high presence virtual scene, its advantage uses texture to replace real model, thus the complexity of model can be reduced on a large scale, but compared with the model based on geometry, there is distortion in visual effect.Visual in order to accelerate scale Forest Scene further, existing visualization simulation technology also realizes the dynamic dispatching of data in conjunction with the method such as visibility cutting, data pre-fetching, reduces EMS memory occupation, improves and draws efficiency.

Summary of the invention

In order to overcome the deficiency that rapidity is poor, accuracy is poor of existing scale Forest Scene generation method, the invention provides the large-scale forest scene quick generation method based on spatial simlanty that a kind of rapidity is good, accuracy is strong.

The technical solution adopted for the present invention to solve the technical problems is:

Based on a large-scale forest scene quick generation method for spatial simlanty, described rapid generation comprises the following steps:

1) extensive forest simulating scenes carries out initiation parameter setting, and from external memory database, the scene distribution information of whole environment Visualization data and forest space is obtained according to set initiation parameter, described initiation parameter comprises: the spacing of the quantity of trees, the kind of trees, initial age, tree, Growing years and envirment factor; Described envirment factor comprises: sunlight, temperature, moisture and soil; Described environment Visualization data comprise: terrain data, terrain texture, atural object data except trees and Sky Scene data; The scene distribution information of described forest space comprises: the position of trees and initial effects circle size;

2) segmentation based on quaternary tree is carried out to extensive scale Forest Scene spatial data, scene partitioning is become equal-sized piece, and set up the block message concordance list of an out-of-core technique, for recording the status information of all scenario block;

3) spatial simlanty of scale Forest Scene is calculated, and by the similarity between each piecemeal of two-dimensional array record, if similarity reaches certain proportion between scenario block, then the piecemeal calculated that can be used in internal memory replaces to be calculated piece; If the similarity between scenario block does not reach certain proportion, then to pass come in initial parameter and knowledge process and store, and by process after these parameters be used for the growth model calculating plant, obtain the growth result of plant;

4) similarity between world subdivision is judged, if similarity reaches certain proportion between world subdivision, and the growth model of one of them scenario block completes calculating, the scenario block do not calculated so can be replaced by the data calculating scenario block, and do not need to recalculate scenario block, thus obtain the biomass of plant fast;

5) if the similarity between world subdivision does not reach certain proportion, the growth model then carrying out scenario block calculates, the initial parameter data determination base strain of coming in is transmitted by user, calculate influence circle scope and the biomass of base strain, then Three-dimension Tree model is imported the emulation realizing scale Forest Scene, visualization result is presented to user.

Technical conceive of the present invention is: existing method accelerate forest growth model base calculate and visual, but these methods do not make full use of the data calculated in internal memory.If by the similarity between scene, make full use of the data calculated in internal memory and substitute current calculative data, so simulation velocity will improve greatly.The scene of the zones of different of Forest Growth emulation may have similar growing environment, certain similarity can be there is between scene, so when Forest Growth simulation accuracy is less demanding, this similarity can be utilized to accelerate the process of plant-growth model calculating and visual drafting, and take full advantage of the data calculated in internal memory, improve memory usage, thus reach the object of forest emulation acceleration.

Beneficial effect of the present invention is mainly manifested in: analysis and the judgement that 1, can carry out spatial simlanty based on the large-scale forest scene quick generation method of spatial simlanty respectively from index of similarities such as plant growth density, terrain feature and envirment factors.When the space similarity of two scenario block reaches certain value, and wherein the growth model of a piece completes calculating, so can replace the scenario block do not calculated by the data calculating scenario block, and not need to recalculate scenario block, thus reach the object of forest emulation acceleration.

2, the method can select different similarity thresholds according to viewpoint to the distance of scenario block automatically, realizes adaptive similarity threshold and determines, ensures that scenario block replacement has higher accuracy.From viewpoint more close to scenario block, the details required for it will be abundanter, and the sense of reality is stronger, and other block so can replacing this block piecemeal must need the similarity-rough set with this block high; Otherwise from viewpoint more away from block, user does not need to know details in sufficient detail, and only needing has a sensory effects true to nature a little to scene, and other piecemeal so can replacing this block piecemeal follows the similarity of this block piecemeal to reduce.

Accompanying drawing explanation

Fig. 1 is the scale Forest Scene simulation contact surface based on spatial simlanty.

Fig. 2 is the schematic diagram of the scene cut process based on quaternary tree.

Embodiment

Below in conjunction with accompanying drawing, the invention will be further described.

See figures.1.and.2, a kind of large-scale forest scene quick generation method based on spatial simlanty, comprises the following steps:

1), extensive forest simulating scenes carries out initiation parameter setting, and from external memory database, obtains the scene distribution information of whole environment Visualization data and forest space according to set initiation parameter.Described initiation parameter comprises: the spacing of the quantity of trees, the kind of trees, initial age, tree, Growing years and envirment factor.Described envirment factor comprises: sunlight, temperature, moisture and soil; Described environment Visualization data comprise: terrain data, terrain texture, atural object data except trees and Sky Scene data; The scene distribution information of described forest space comprises: the position of trees and initial effects circle size.

2), to extensive scale Forest Scene spatial data carry out the segmentation based on quaternary tree, scene partitioning is become equal-sized piece, and set up the block message concordance list of an out-of-core technique, for recording the status information of all scenario block.

World subdivision is the basis judging similarity between each piecemeal.As shown in Figure 2, first user sets the breadth extreme of block to scale Forest Scene dividing method, and this value is the basis for estimation of the recurrence of landform being carried out to four points; Then, use top-down mode, using whole scale Forest Scene as root node, judge whether the width of colony area exceedes the breadth extreme of user's setting from root node, if do not meet, not split and as leafy node, and the relevant information of node is preserved; Otherwise 4 the sub-nodal regions equal to the continuous recursive subdivision of root node, if this node has sibling, same recursive subdivision, until no longer meet segmentation condition.Finally all leafy nodes are all kept in external memory.

In Forest Growth simulation process, setting up the block message concordance list of an out-of-core technique, for recording the status information of all scenario block, and dynamically updating according to the change of current view point region parameter.Carry out the memory address of store data block with one dimension array of pointers B [L], as B [i]=0, represent that the data of i-th piece are not loaded into internal memory; As B [i] ≠ 0, the value of B [i] represents can be expressed as the first address of memory field, data block place:

And 0≤i≤L-1, n ∈ N ⁺.

3), to the spatial simlanty of scale Forest Scene calculate, and by the similarity between each piecemeal of two-dimensional array record, if similarity reaches certain proportion between scenario block, then the piecemeal calculated that can be used in internal memory replaces to be calculated piece; If the similarity between scenario block does not reach certain proportion, then to pass come in initial parameter and knowledge process and store, and by process after these parameters be used for the growth model calculating plant, obtain the growth result of plant.The index of similarity from the viewpoint of plant growth density, terrain feature and envirment factor these.

(a) density similarity.By the area of distribution space with count than the density regarding point group as, and utilize the some population density of the trees in different distributions space to judge its similarity.In pure forest plantation situation, the trees quantity of object space 1 is counted N ₁, the area shared by distributed areas is designated as S ₁; The trees quantity of object space 2 counts N ₂, the area shared by distributed areas is designated as S ₂, density similarity is defined as:

$\mathrm{Sim}\left(\mathrm{den}\right)=\min \{\frac{S_1/N_1}{S_2/N_2},\frac{S_2/N_2}{S_1/N_1}\}---\left(2\right)$

Under mongrel plants situation, the seeds 1 trees quantity of object space 1 is counted N ₁₁, seeds 2 trees quantity counts N ₁₂, the area shared by distributed areas is designated as S ₁, the seeds 1 trees quantity of object space 2 counts N ₂₁, seeds 2 trees quantity counts N ₂₂, the area shared by distributed areas is designated as S ₂, then density similarity can be defined as:

Sim(den)=max{Sim(den1),Sim(den2),Sim(den12)} （3）

Wherein, $\mathrm{Sim}\left(\mathrm{den}12\right)=\min \{\frac{S_1/(N_{11}+N_{12})}{S_2/(N_{21}+N_{22})},\frac{S_2/(N_{21}+N_{22})}{S_1/(N_{11}+N_{12})}\},$

$\mathrm{Sim}\left(\mathrm{den}1\right)=\min \{\frac{S_1/N_{11}}{S_2/N_{21}},\frac{S_2/N_{21}}{S_1/N_{11}}\},$ $\mathrm{Sim}\left(\mathrm{den}2\right)=\min \{\frac{S_1/N_{12}}{S_2/N_{22}},\frac{S_2/N_{22}}{S_1/N_{12}}\}.$

(b) direction similarity.The direction of the feature landform on slope is expressed, and the leg-of-mutton normal vector according to forming scenario block landform determines its direction, and concrete grammar obtains all normal vectors forming world subdivision, and be averaging it.If the topographical direction angle of distribution space 1 is angle ₁, the topographical direction angle of distribution space 2 is angle ₂, direction similarity is defined as:

$\mathrm{Sim}\left(\mathrm{dir}\right)=\min \{\frac{{\mathrm{angle}}_1}{{\mathrm{angle}}_2},\frac{{\mathrm{angle}}_2}{{\mathrm{angle}}_1}\}---\left(4\right)$

(c) area similarity.The height fluctuating situation of landform is considered by area similarity, the landform of scale Forest Scene is generally represented by triangular mesh, realize the expression to space group targeted graphical by the triangulation network setting up landform, the surface area of the terrain mesh formation of scenario block can be calculated according to the computing formula of triangle area.S ₁for the surface area in target distribution space 1, S ₂for the surface area in target distribution space 2, the area ratio of similitude in two target distribution spaces is:

$\mathrm{Sim}\left(\mathrm{area}\right)=\min \{\frac{s_1}{s_2},\frac{s_2}{s_1}\}---\left(5\right)$

(d) environment similarity.Analyzed by the mean sea level of scenario block, average soil thickness two factors, if two scenario block P comparing similarity ₁, P ₂mean sea level be respectively H ₁and H ₂, average soil thickness is respectively T ₁and T ₂, spatial simlanty is expressed as:

$\mathrm{Sim}\left(\mathrm{env}\right)=\left\{\begin{array}{cc}0& \mathrm{if}|H_1-H_2|>\mathrm{MaxH\; or}|T_1-T_2|>\mathrm{MaxT}\\ 1& \mathrm{otherwise}\end{array}\right.---\left(6\right)$

Wherein, MaxH is that to compare the sea level elevation allowed between the scenario block of similarity poor, and represent in the scope of this sea level elevation difference, the growing state of two scenario block is more similar.MaxT is the difference comparing the thickness of soil allowed between the scenario block of similarity, represents that, in the scope of this thickness of soil difference, the impact of thickness of soil on the growing state of two scenario block is smaller.When Sim (env) value is 0, represent scenario block P ₁and P ₂dissimilar; When its value is 1, represent two scenario block P ₁and P ₂spatial similarity.

(e) comprehensive similarity.Integrally, these several respects such as the comprehensive density of trees, terrain feature, envirment factor consider the growing space target of forest community.Comprehensive similarity computing method are:

Sim(space)=k ₁*Sim(den)+k ₂*Sim(area)+k ₃*Sim(dir)+k ₄*Sim(env) （7）

Wherein, Sim (den) is density similarity, and Sim (area) is area similarity, and Sim (dir) is direction similarity, the similarity that Sim (env) is envirment factor.K ₁, k ₂, k ₃, k ₄be respectively the weight coefficient of density, area, direction, envirment factor, the value of weight coefficient can be set according to the simulating scenes of reality, and k ₁+ k ₂+ k ₃+ k ₄=1.

Employing two-dimensional array deposits the similarity between piecemeal, if the scenario block quantity of the scene tree bottom layer node obtained after segmentation is L, each A [i] [j] (0≤i < L in table 1 in array, 0≤j < L) represent similarity between i-th piece and jth block, the value of each of i-th in table-1 row show respectively the similarity between the i-th blocks of data block and other block.The calculating of similarity carries out in pre-service, so do not need to expend extra time.

The storage organization of table 1 similarity

4) similarity between world subdivision, is judged, if similarity reaches certain proportion between world subdivision, and the growth model of one of them scenario block completes calculating, the scenario block do not calculated so can be replaced by the data calculating scenario block, and do not need to recalculate scenario block, thus obtain the biomass of plant fast.

Suppose that similarity threshold between the piecemeal in scene from small to large, is designated as SIM respectively ₁, SIM ₂..., SIM _n, by the distance decision similarity threshold of viewpoint to piecemeal, be formulated as:

$\mathrm{SIM}=\left\{\begin{array}{c}{\mathrm{SIM}}_1,d_1\&le;d\\ {\mathrm{SIM}}_2,d_2\&le;d<d_1\\ ......\\ \mathrm{SIMn},d_n\&le;d<d_{n-1}\end{array}\right.---\left(8\right)$

In formula, 0≤d _n< ... <d ₂<d ₁, d _i(i=1,2 ..., n) represent the distance threshold being chunked into viewpoint, d represents that viewpoint arrives the actual range of piecemeal.In order to describe this information, by the relation of two-dimensional array C [2] [n] recording distance and similarity threshold, above-mentioned relation is converted into array as shown in table 2, the first row of array represents distance threshold, and the second row represents similarity threshold.

d 1	d 2	d 3	d 4	......	d n
						SIM 1	SIM 2	SIM 3	SIM 4	......	SIM n

The relation of table 2 distance and similarity threshold

The algorithm steps utilizing the spatial simlanty of scene to emulate certain scenario block is as follows:

Then Step1: suppose that current scene block is block i, if B [i] is not 0, then jumps to step 6 by B [i] assignment to tmp pointer variable, otherwise enter step 2.

Step2: the distance d of computing block i and viewpoint.

Step3: the first row of search two-dimensional array C finds first its value to be less than or equal to item C [0] [n] of d, and corresponding C [1] [n] is the minimum Similarity value that can replace.

Step4: i-th row of traversal array A, searching meet value be more than or equal to C [1] [n] and the B of correspondence [k] be not 0 maximal term A [i] [k], if find this Xiang Ze that B [k] is assigned to tmp, then jump to step 6, otherwise enter step 5.

Step5: read the data of i-th piece to internal memory from external memory, and the address of depositing this blocks of data is assigned to B [i] and tmp.

Step6: the memory field sense data pointed to from tmp also emulates.

5) if the similarity between world subdivision does not reach certain proportion, the growth model then carrying out scenario block calculates, the initial parameter data determination base strain of coming in is transmitted by user, calculate influence circle scope and the biomass of base strain, then Three-dimension Tree model is imported the emulation realizing scale Forest Scene, visualization result is presented to user.

When user's viewpoint changes, draw thread and calculate current visibility region according to current view point, contextual data in internal memory is judged simultaneously, confirm that scene repaints required data whether in internal memory, if contextual data is in internal memory, then directly carries out the drafting of scene and send viewpoint updating message to data prefetching thread.If need the scene of drawing not in internal memory current, drawing thread then needs to carry out LOD calculating according to view information such as the position of current view point, the directions of current view point to the trees in visibility region.

When selecting the trees LOD model of extensive scale Forest Scene, in extremely near-sighted point range, adopt three-dimensional refined model rendering technique.Due to close viewpoint, in roam procedure, viewpoint changes its drafting impact very large.Meanwhile, near-sighted point range limits by field range, and the tree of simultaneously drawing can not be too many, generally at about 2,3, so by the comparatively meticulous model of computer drawing.Short range, adopts the geometry LOD model simplified.Simultaneously because distance viewpoint has certain distance, the change of viewpoint can not require frequently to change drafting.From viewpoint more away from trees, its LOD model accuracy is more coarse.Far range, not obvious owing to drawing individual demand, adopt billboard Texture Mapping Technology.

Claims (5)

1. based on a large-scale forest scene quick generation method for spatial simlanty, it is characterized in that: described rapid generation comprises the following steps:

1) extensive forest simulating scenes carries out initiation parameter setting, and from external memory database, the scene distribution information of whole environment Visualization data and forest space is obtained according to set initiation parameter, described initiation parameter comprises: the spacing of the quantity of trees, the kind of trees, initial age, tree, Growing years and envirment factor; Described envirment factor comprises: sunlight, temperature, moisture and soil; Described environment Visualization data comprise: terrain data, terrain texture, atural object data except trees and Sky Scene data; The scene distribution information of described forest space comprises: the position of trees and initial effects circle size;

2) segmentation based on quaternary tree is carried out to extensive scale Forest Scene spatial data, scene partitioning is become equal-sized piece, and set up the block message concordance list of an out-of-core technique, for recording the status information of all scenario block;

3) spatial simlanty of scale Forest Scene is calculated, and by the similarity between each piecemeal of two-dimensional array record, if similarity reaches certain proportion between scenario block, then the piecemeal calculated that can be used in internal memory replaces to be calculated piece; If the similarity between scenario block does not reach certain proportion, then to pass come in initial parameter and knowledge process and store, and by process after these parameters be used for the growth model calculating plant, obtain the growth result of plant;

4) similarity between world subdivision is judged, if similarity reaches certain proportion between world subdivision, and the growth model of one of them scenario block completes calculating, the scenario block do not calculated so can be replaced by the data calculating scenario block, and do not need to recalculate scenario block, thus obtain the biomass of plant fast;

5) if the similarity between world subdivision does not reach certain proportion, the growth model then carrying out scenario block calculates, the initial parameter data determination base strain of coming in is transmitted by user, calculate influence circle scope and the biomass of base strain, then Three-dimension Tree model is imported the emulation realizing scale Forest Scene, visualization result is presented to user.

2. as claimed in claim 1 based on the large-scale forest scene quick generation method of spatial simlanty, it is characterized in that: described step 2) in, world subdivision process is as follows: first user sets the breadth extreme of block, and this value is the basis for estimation of the recurrence of landform being carried out to four points; Then, use top-down mode, using whole scale Forest Scene as root node, judge whether the width of colony area exceedes the breadth extreme of user's setting from root node, if do not meet, not split and as leafy node, and the relevant information of node is preserved; Otherwise, 4 the sub-nodal regions equal to the continuous recursive subdivision of root node, if this node has sibling, same recursive subdivision, until no longer meet segmentation condition, is finally all kept at all leafy nodes in external memory;

In Forest Growth simulation process, setting up the block message concordance list of an out-of-core technique, for recording the status information of all scenario block, and dynamically updating according to the change of current view point region parameter; Carry out the memory address of store data block with one dimension array of pointers B [L], as B [i]=0, represent that the data of i-th piece are not loaded into internal memory; As B [i] ≠ 0, the value of B [i] represents can be expressed as the first address of memory field, data block place:

And 0≤i≤L-1, n ∈ N ⁺.

3. as claimed in claim 2 based on the large-scale forest scene quick generation method of spatial simlanty, it is characterized in that: described step 3) in, the index of similarity comprises plant growth density similarity, direction similarity, area similarity and environment similarity, wherein

(a) plant growth density similarity: by the area of distribution space with count than the density regarding point group as, and utilize the some population density of the trees in different distributions space to judge its similarity; In pure forest plantation situation, the trees quantity of object space 1 is counted N ₁, the area shared by distributed areas is designated as S ₁, the trees quantity of object space 2 counts N ₂, the area shared by distributed areas is designated as S ₂, density similarity is defined as:

$\mathrm{Sim}\left(\mathrm{den}\right)=\min \{\frac{S_1/N_1}{S_2/N_2},\frac{S_2/N_2}{S_1/N_1}\}---\left(2\right)$

Under mongrel plants situation, the seeds 1 trees quantity of object space 1 is counted N ₁₁, seeds 2 trees quantity counts N ₁₂, the area shared by distributed areas is designated as S ₁; The seeds 1 trees quantity of object space 2 counts N ₂₁, seeds 2 trees quantity counts N ₂₂, the area shared by distributed areas is designated as S ₂, then density similarity is defined as:

Sim(den)＝max{Sim(den1),Sim(den2),Sim(den12)} (3)

Wherein, $\mathrm{Sim}\left(\mathrm{den}12\right)=\min \{\frac{S_1/(N_{11}+N_{12})}{S_2/(N_{21}+N_{22})},\frac{S_2/(N_{21}+N_{22})}{S_1/(N_{11}+N_{12})}\},$

$\mathrm{Sim}\left(\mathrm{den}1\right)=\min \{\frac{S_1/N_{11}}{S_2/N_2},\frac{S_2/N_{21}}{S_1/N_{11}}\},\mathrm{Sim}\left(\mathrm{den}2\right)=\min \{\frac{S_1/N_{12}}{S_2/N_{22}},\frac{S_2/N_{22}}{S_1/N_{12}}\};$

(b) direction similarity: the direction of the feature landform on slope is expressed, the leg-of-mutton normal vector according to forming scenario block landform determines its direction, and concrete grammar obtains all normal vectors forming world subdivision, and be averaging it; If the topographical direction angle of distribution space 1 is angle ₁, the topographical direction angle of distribution space 2 is angle ₂, direction similarity is defined as:

$\mathrm{Sim}\left(\mathrm{dir}\right)=\min \{\frac{{\mathrm{angle}}_1}{{\mathrm{angle}}_2},\frac{{\mathrm{angle}}_2}{{\mathrm{angle}}_1}\}---\left(4\right)$

(c) area similarity: the height fluctuating situation being considered landform by area similarity, the landform of scale Forest Scene is represented by triangular mesh, realize the expression to space group targeted graphical by the triangulation network setting up landform, the surface area of the terrain mesh formation of scenario block can be calculated according to the computing formula of triangle area; S ₁for the surface area in target distribution space 1, S ₂for the surface area in target distribution space 2, the area ratio of similitude in two target distribution spaces is:

$\mathrm{Sim}\left(\mathrm{area}\right)=\min \{\frac{S_1}{S_2},\frac{S_2}{S_1}\}---\left(5\right)$

D () environment similarity: analyzed by the mean sea level of scenario block, average soil thickness two factors, if two scenario block P comparing similarity ₁, P ₂mean sea level be respectively H ₁and H ₂, average soil thickness is respectively T ₁and T ₂, spatial simlanty is expressed as:

Wherein, MaxH is that to compare the sea level elevation allowed between the scenario block of similarity poor, and represent in the scope of this sea level elevation difference, the growing state of two scenario block is more similar; MaxT is the difference comparing the thickness of soil allowed between the scenario block of similarity, represents that, in the scope of this thickness of soil difference, the impact of thickness of soil on the growing state of two scenario block is smaller; When Sim (env) value is 0, represent scenario block P ₁and P ₂dissimilar; When its value is 1, represent two scenario block P ₁and P ₂spatial similarity;

(e) comprehensive similarity: integrally, comprehensive similarity computing method are the growing space target of forest community:

Sim(space)＝k ₁*Sim(den)+k ₂*Sim(area)+k ₃*Sim(dir)+k ₄*Sim(env) (7)

Wherein, Sim (den) is density similarity, and Sim (area) is area similarity, and Sim (dir) is direction similarity, the similarity that Sim (env) is envirment factor; k ₁, k ₂, k ₃, k ₄be respectively the weight coefficient of density, area, direction, envirment factor, the value of weight coefficient can be set according to the simulating scenes of reality, and k ₁+ k ₂+ k ₃+ k ₄=1;

Employing two-dimensional array deposits the similarity between piecemeal, if the scenario block quantity of the scene tree bottom layer node obtained after segmentation is L, each A [i] [j] (0≤i < L in table 1 in array, 0≤j < L) represent similarity between i-th piece and jth block, the value of each of i-th in table 1-1 row show respectively the similarity between the i-th blocks of data block and other block; Table 1 shows the storage organization of similarity,

4., as claimed in claim 3 based on the large-scale forest scene quick generation method of spatial simlanty, it is characterized in that: described step 4) in, suppose that the similarity threshold between the piecemeal in scene is designated as SIM from small to large respectively ₁, SIM ₂..., SIM _n, by the distance decision similarity threshold of viewpoint to piecemeal, be formulated as:

$\mathrm{SIM}=\left\{\begin{array}{c}{\mathrm{SIM}}_1,d_1\&le;d\\ {\mathrm{SIM}}_2,d_2\&le;d<d_1\\ ......\\ \mathrm{SIMn},d_n\&le;d<d_{n-1}\end{array}\right.---\left(8\right)$

In formula, 0≤d _n<......<d ₂<d ₁, d _i(i=1,2 ..., n) represent the distance threshold being chunked into viewpoint, d represents that viewpoint arrives the actual range of piecemeal; In order to describe this information, by the relation of two-dimensional array C [2] [n] recording distance and similarity threshold, the first row of array represents distance threshold, and the second row represents similarity threshold; Table 2 shows the storage organization middle distance of array C [2] [n] and the relation of similarity threshold,

d ₁ d ₂ d ₃ d ₄ …… d _n SIM ₁ SIM ₂ SIM ₃ SIM ₄ …… SIM _n

；

The algorithm steps utilizing the spatial simlanty of scene to emulate certain scenario block is as follows:

Then Step1: suppose that current scene block is block i, if B [i] is not 0, then jumps to Step 6 by B [i] assignment to tmp pointer variable, otherwise enter Step 2;

Step2: the distance d of computing block i and viewpoint;

Step3: the first row of search two-dimensional array C finds first its value to be less than or equal to item C [0] [n] of d, and corresponding C [1] [n] is the minimum Similarity value that can replace;

Step4: i-th row of traversal array A, searching meet value be more than or equal to C [1] [n] and the B of correspondence [k] be not 0 maximal term A [i] [k], if find this Xiang Ze that B [k] is assigned to tmp, then jump to Step 6, otherwise enter Step5;

Step5: read the data of i-th piece to internal memory from external memory, and the address of depositing this blocks of data is assigned to B [i] and tmp;

Step6: the memory field sense data pointed to from tmp also emulates.

5. as claimed in claim 3 based on the large-scale forest scene quick generation method of spatial simlanty, it is characterized in that: described step 5) in, when user's viewpoint changes, draw thread and calculate current visibility region according to current view point, contextual data in internal memory is judged simultaneously, confirm that scene repaints required data whether in internal memory, if contextual data is in internal memory, then directly carries out the drafting of scene and send viewpoint updating message to data prefetching thread; If need the scene of drawing not in internal memory current, drawing thread then needs to carry out LOD calculating according to view information such as the position of current view point, the directions of current view point to the trees in visibility region.