CN103886635A - Self-adaption LOD model establishing method based on face clustering - Google Patents

Abstract

The invention discloses an adaptive LOD model construction method based on surface clustering, which belongs to the field of three-dimensional digital cities and geographic information systems. The invention reduces the data volume of the LOD model; the simplified LOD model obtained by using the method is a kind of Progressive loading, the composition of each LOD comes from the partial surface composition of the original model, so it will not cause confusion of vertex attributes; the algorithm of the present invention has high efficiency and can be used for real-time self-adaptive generation of LOD models, effectively avoiding LOD switching The transition is not natural enough; the texture optimization is more standardized according to the texture resolution.

Description

Adaptive LOD model construction method based on face clustering

technical field

The invention belongs to the field of three-dimensional digital cities and geographic information systems, and in particular relates to a method for constructing an LOD model.

Background technique

LOD (Levels of Detail) is a commonly used 3D model loading idea, and its core lies in generating a good LOD model and a loading strategy for the LOD model. A good model optimization result can improve rendering efficiency under the same loading model; and for the same model, a good loading mode can also improve rendering speed.

Most of the existing architectural model simplification methods use free mesh-based simplification methods, which will cause deformation of the building during the process of deleting or merging surfaces, and the effect is not good; secondly, since architectural models are usually pasted with texture However, in most of the model simplification methods, the vertex attributes are not considered enough, and the simplification method will cause confusion in texture map coordinates, resulting in texture map confusion, which greatly reduces the quality of model simplification.

LOD models are not loaded efficiently. When many 3D platforms load massive model data, there are problems such as low model loading efficiency, insufficient switching between LOD models, and slow loading and rendering speeds, which affect the visual experience and further industry applications of the 3D platform. How to improve The dynamic throughput speed of the model and the problems that need to be solved to speed up the 3D rendering speed of large scenes.

Contents of the invention

In view of the above-mentioned defects in the prior art, the technical problem to be solved by the present invention is to provide a method for constructing an LOD model that improves the loading efficiency of a 3D model of a large scene in a digital city.

In order to achieve the above object, the present invention provides a method for building an adaptive LOD model based on surface clustering, which is carried out in the following steps:

Step 1. Generate shape LOD; generate and optimize texture LOD;

Described generation shape LOD is carried out according to the following steps:

A1. Obtain the important factors of each surface;

A2. Determine the segmentation threshold of important factors and divide the model;

A3. Generate a simplified model and build a shape LOD;

Step 2: Use the cumulative method to load the shape LOD, and use the traditional switching method to load the texture LOD.

Preferably, as described in step 1, the important factors of each surface are obtained according to the following steps:

For each face of the building model, define the important factor of its contribution to the appearance of the whole building model, set the sum of the important factors of all faces to 1, set the sum of the important factors of the simplified model to E, 0<E≤1; for The sum E of the important factors of the same size obtains the combination with the least number of faces.

Preferably, the determination of the segmentation threshold of important factors and the division of models described in step 1 are carried out in the following steps:

Arrange all the faces according to the size of the important factors, and divide the building faces by setting the segmentation threshold of more than two important factors, and obtain the result of face clustering, so as to divide the building model into multiple parts; Parts are combined to get different levels of shape LOD.

Preferably, the simplified model is generated as described in step 1, and the shape LOD is constructed according to the following steps:

All the faces within the threshold constitute the simplified model of this level; through different segmentation thresholds, the simplified models of different levels are divided, and the shape LOD is constructed by combining the simplified models of different levels.

Preferably, the LOD model texture optimization described in step 1 is performed in the following steps:

Set the fine texture of the LOD model texture to T1, and set the simplified texture of the LOD model texture to T2;

For fine texture T1, optimize fine model image size:

得到精细模型图片的缩减倍数；将精细模型图片的长度和宽度缩减为原来的1/N(nx,ny)；Set the texture standard resolution used for mapping as R ₁ (r ₁ x, r ₁ y), where r ₁ x and r ₁ y are the resolutions of the texture standard used for mapping in the x and y directions of the coordinate axis, respectively ; Set the resolution of the fine model picture as R ₂ (r ₂ x, r ₂ y), where r ₂ x and r ₂ y are the resolutions of the fine model picture in the coordinate axis x direction and y direction respectively; set the fine model The reduction factor of the picture is N(nx,ny), nx and ny are the reduction factors of the fine texture in the x direction and y direction of the coordinate axis respectively; calculate

Get the reduction factor of the fine model picture; reduce the length and width of the fine model picture to the original 1/N(nx,ny);

Use 3Dmax's rendering-to-texture technology to combine all the simplified textures T2 into one image, and use the resampling method to reduce the resolution of the combined simplified texture T2.

The beneficial effects of the present invention are:

1. The present invention reduces the data volume of the LOD model and improves the loading efficiency of the model. The general LOD model includes at least the original fine model and the generated simplified model. If there are multiple levels of LOD, the amount of the model will be more. However, the simplified model obtained by using this method is composed of parts of the original fine model, and all the total model quantities are still the original fine model. In terms of texture, the simplified texture generated by the method in this paper is the same texture for the multi-level LOD model. However, due to the deformation of the usually generated LOD model, more sets of textures and multiple sets of texture coordinates may be required to meet the requirements.

2. The present invention will not cause confusion of vertex attributes; the simplified LOD model obtained by using this method is a progressive loading, and the composition of each LOD is derived from the partial face composition of the original model, so it will not cause vertex property confusion. For example, the texture coordinates and normal information contained in the vertices will not be confused after the LOD is generated.

3. The algorithm of the present invention has high efficiency and can be used for real-time self-adaptive generation of LOD models; the texture of the LOD model can be fixedly set as two layers of simplified texture and fine texture, and in the generation of shape LOD, the simplified model is composed of partial faces of the fine model. It only needs to increase or decrease part of the face in real time according to the set fineness and the size of the important factors of the face to obtain a new cluster division, thereby obtaining a new shape LOD, and forming an LOD model with the LOD model texture.

4. The present invention effectively avoids excessive and unnatural LOD switching; for ordinary LOD switching, since the simplified model and the fine model are different, it is inevitable that there will be inconsistencies between different models, the better the simplification effect , the transition is more natural. When using this method to switch, new surfaces are continuously added on the basis of the LOD model in the previous part to form a new and more refined LOD model, and finally achieve the display of the fine model.

5. The texture optimization is more standardized according to the texture resolution; for the three-dimensional digital city model, not only should there be model accuracy, but also the accuracy of the texture attached to the model should be considered uniformly. The present invention uses the resolution to automatically optimize the texture, which is more Standardizes the mapping standard for textures.

Description of drawings

Fig. 1 is a schematic flow chart of a specific embodiment of the present invention.

Detailed ways

Below in conjunction with accompanying drawing and embodiment the present invention will be further described:

As shown in Figure 1, an adaptive LOD model construction method based on surface clustering is carried out in the following steps:

Step 1. Generate shape LOD, generate and optimize texture LOD;

Described generation shape LOD is carried out according to the following steps:

A1. Obtain the important factors of each surface;

A2. Determine the division threshold of important factors and divide the model;

A3. Generate a simplified model and build a shape LOD;

Step 2: Use the cumulative method to load the shape LOD, and use the traditional switching method to load the texture LOD.

Obtaining the important factors of each surface described in step 1 is carried out in the following steps:

For each face of the building model, define the important factors of its contribution to the appearance of the whole building model, and define the important factors of the face according to the model shape itself and the actual meaning of the model face. In this embodiment, the important factors defining the surface according to the actual meaning of the model shape itself are: the important factor of the large surface area is greater than the important factor of the small surface; the important factor of the surface defined according to the actual meaning of the model surface is: the building The important factor of the main structural surface is greater than the important factor of the architectural detail modeling surface. Define the sum of the important factors of all surfaces to be 1, set the sum of the important factors of the simplified model to be E, 0<E≤1; the simplified model is for the sum of the important factors E of the same size, and obtain the combination of the least number of surfaces used. The larger E is, the finer the simplified model is.

As described in step 1, determine the segmentation threshold of important factors and divide the model according to the following steps:

Arrange all the faces according to the size of the important factors, and cluster the different faces of the building model by setting the segmentation threshold of more than two important factors, so as to divide them into multiple parts, and then combine these divided parts to obtain different levels The shape of the LOD. In this embodiment, the segmentation threshold is obtained by an adaptive method, and is set according to the requirement for the level of refinement of the simplified model at the time of loading and displaying.

Generate a simplified model as described in step 1, and construct the shape LOD according to the following steps:

Each segmentation threshold can divide all the faces into two parts, and all the faces within the threshold constitute the simplified model of this level; through different segmentation thresholds, different levels of simplified models can be divided, and the combination of simplified models of different levels Construct a shape LOD; adjacent two levels of LOD, the finer simplified model is the sum of all faces of the rougher model and all faces of the incremental part (including faces between the two division thresholds).

In this embodiment, a three-level LOD model is taken as an example, and more levels can be deduced by analogy. As shown in the following formula (1-4):

f ₁ >= f ₂ ... >= f _k1 ... >= f _k2 ... >= f _n ; (1)

F _all ={F ₁ ,F ₂ ,...,F _k1 ,...,F _k2 ,...,F _n }; (2)

$\left\{\begin{array}{c}\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; \{</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; \}</span>\\ \mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; \{</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; \}</span>\\ \mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; 3</span>}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; \{</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; no</span>}}<span\; class="notranslate">\; \}</span>\end{array}\right.<span\; class="notranslate">\; ;</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 3</span>}<span\; class="notranslate">\; )</span>$

Formula (1) is the important factor of the surface arranged from large to small. In this embodiment, the area size of each surface is used to give the value of the important factor. Formula (2) is the arrangement of the surface corresponding to each important factor. Equation (3) is the division result of area clustering. According to the division result, Equation (4) is obtained, which is the LOD model, including LOD0, LOD1 and LOD2.

In formula (1), f _i is the important factor ranked i in the ranking, i is a positive integer, f _k1 and f _k2 are the minimum value of the important factor value f of the surfaces of M1 and M2, that is, the segmentation threshold, The subscripts k1 and k2 represent their positions in the ranking of important factors; in formula (2), F _i represents the surface corresponding to f _i , F _all refers to the set of all surfaces, and the fine model includes all surfaces, so F _all is Detailed model. M1 , M2 , and M3 represent three-part face collections respectively. For example, M1 is composed of faces F ₁ , F ₂ , . . . , F _k1 . In formula (4), the fine model LOD0 is composed of {M1, M2, M3}, LOD1 and LOD2 represent the generated simplified models of different levels, which are respectively composed of {M1, M2} and {M1}, and the fine model of LOD1 The degree is greater than LOD2.

The determination of the partition minimum f _k1 and f _k2 includes two constraints:

1) Constraint 1:

It is set from the fact that the simplified model can maintain the appearance ratio of the original fine model, and only needs to meet the clarity of the LOD model in its display range. For example, LOD1 is displayed between 500-1000 meters away from the model, as long as LOD1 is guaranteed The appearance at a distance of 500 meters meets the requirements, and the preliminary division of the LOD model surface can be adaptively obtained through a given definition.

Take the example that all the faces of the model are triangular faces, assume that the area of the triangular faces is an important factor of the face, and take the example of generating a two-layer LOD model, including LOD0 and LOD1. Then the threshold can first be determined as follows:

V _screen = V _world * MVPW; (5)

The above formula indicates the transformation relationship from the model coordinates in the world coordinate system to the screen coordinate system, V _screen is the screen coordinates, V _world is the coordinates of the point in the world coordinate system, and MVPW is the transformation matrix. Set the minimum triangle size that is allowed to be ignored on the screen, and calculate the area of the triangle at the switching distance between the simplified model and the fine model in the actual coordinate system, which is the area of the smallest face in the simplified model set, that is, the area corresponding to the smallest face Important factor, in the sequence of faces represented by formula (2), obtain the corresponding P1th triangular face as the division interval, and P1 is a positive integer.

2) Constraint 2:

Only the restriction of P1 is not enough. If a model is composed of small faces, all of them will be ignored, and the face set of the resulting simplified model is empty, so P2 needs to be added. The setting of P2 is as follows:

The important factors of the surface are gradually accumulated from large to small, and the sum E of the first P2 items in formula (1) is calculated (the sum of all items is 1), and E satisfies:

$\mathrm{<span\; class="notranslate">\; E.</span>}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; P</span>}\mathrm{<span\; class="notranslate">\; 2</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; \&Greater\; Equal;</span>{\mathrm{<span\; class="notranslate">\; E.</span>}}_{\mathrm{<span\; class="notranslate">\; min</span>}}<span\; class="notranslate">\; ;</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 6</span>}<span\; class="notranslate">\; )</span>$

f _i is an important factor of the i-th triangle, E _min is the allowable minimum value of E, and the value of P2 can be obtained by solving formula (6).

Combining the size of P1 obtained by constraint 1 and the value of P2 obtained by constraint 2, there are:

$\left\{\begin{array}{c}m1=\{f_1,f_2,......,f_k\}\\ m2=\{f_k,......,f_{\mathrm{no}}\}\end{array}\right.,$ where k=max(P1,P2); (7)

Then the composition of the shape LOD is obtained as follows:

$\left\{\begin{array}{c}\mathrm{<span\; class="notranslate">\; LOD</span>}\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; \{</span>\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; \}</span>\\ \mathrm{<span\; class="notranslate">\; LOD</span>}\mathrm{<span\; class="notranslate">\; 0</span>}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; \{</span>\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; \}</span>\end{array}\right.<span\; class="notranslate">\; ;</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 8</span>}<span\; class="notranslate">\; )</span>$

Generate and optimize the texture LOD as described in step 1 and proceed as follows:

Set the fine texture of the LOD model texture to T1, and set the simplified texture of the LOD model texture to T2;

For fine texture T1, optimize fine model image size:

得到精细模型图片的缩减倍数；将精细模型图片的长度和宽度缩减为原来的1/N(nx,ny)。Set the texture standard resolution used for mapping as R ₁ (r ₁ x, r ₁ y), where r ₁ x and r ₁ y are the resolutions of the texture standard used for mapping in the x and y directions of the coordinate axis, respectively ; Set the resolution of the fine model picture as R ₂ (r ₂ x, r ₂ y), where r ₂ x and r ₂ y are the resolutions of the fine model picture in the coordinate axis x direction and y direction respectively; set the fine model The reduction factor of the picture is N(nx,ny), nx and ny are the reduction factors of the fine texture in the x direction and y direction of the coordinate axis respectively; calculate

Get the reduction factor of the fine model picture; reduce the length and width of the fine model picture to the original 1/N(nx,ny).

Use 3Dmax's rendering-to-texture technology to merge all the simplified textures T2 into one image, and use the resampling method to reduce the combined resolution of the simplified textures T2 by 3/4 or 15/16. In this embodiment, the sampling method will The merged simplified texture T2 is reduced in resolution by 3/4.

In step 2, the shape LOD and the LOD model texture are loaded in an accumulative manner. In this embodiment, a three-layer LOD model and a two-layer LOD texture are used as examples. The loading distance ranges of different LODs of the model are as follows:

$\left\{\begin{array}{cc}\mathrm{<span\; class="notranslate">\; LOD</span>}\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; \{</span>\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; T</span>}\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; \}</span>& \mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; \&Greater\; Equal;</span>\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; 1</span>}\\ \mathrm{<span\; class="notranslate">\; LOD</span>}\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; \{</span>\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; T</span>}\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; \}</span>& \mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ></span>\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; \&Greater\; Equal;</span>\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; 2</span>}\\ \mathrm{<span\; class="notranslate">\; LOD</span>}\mathrm{<span\; class="notranslate">\; 0</span>}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; \{</span>\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; 3</span>}<span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; T</span>}\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; \}</span>& \mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; ></span>\mathrm{<span\; class="notranslate">\; the\; s</span>}\end{array}\right.<span\; class="notranslate">\; ;</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 9</span>}<span\; class="notranslate">\; )</span>$

s is the distance from the model position to the current viewpoint, s1 and s2 are the L0D switching distances, combined with the parameters when the LOD model is generated, that is, the switching distance set when the P1 value is acquired is used as the values of s1 and s2. Similarly, after the switching distance is given, the cluster division can also be adjusted in real time to obtain a new real-time adaptive LOD model.

When s≥s1, only M1 is loaded, and the texture uses T2, (T2 is a simplified texture, T1 is a fine texture, two layers of LOD textures are obtained from the texture LOD generation and optimization in the previous section), when s1>s≥s2, increase M2 , at this time, LOD1 is formed, and T2 is still used for the texture; when s2>s, M3 is further added to the scene to form a fine model LOD0, and the fine texture T1 is used at the same time. In this way, a progressive model LOD loading system is formed.

The present invention utilizes the model LOD technology to obtain a group of model nodes with different levels of detail. The LOD model consists of two parts, shape LOD and texture LOD. The LOD model texture involved in this embodiment mainly includes two layers, a fine texture and a simplified texture with one level of resolution reduction. The shape LOD also includes a fine model and a multi-level simplified model, which together constitute the LOD model. For example, the level 2 LOD model includes the fine model and the first-level simplified model, and the third-level LOD model includes the fine model and the second-level simplified model. The detailed model is the obtained 3D model, which can be obtained through 3D manual modeling or automatic modeling.

The faces of the model involved in this embodiment include triangular faces after all the models are converted into triangular meshes, and sets including polygonal faces and triangular faces.

The preferred specific embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make many modifications and changes according to the concept of the present invention without creative efforts. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning or limited experiments on the basis of the prior art shall be within the scope of protection defined by the claims.

Claims (5)

1. a method for building an adaptive LOD model based on face clustering, characterized in that it proceeds in the following steps: Step 1. Generate shape LOD; generate and optimize texture LOD; Described generation shape LOD is carried out according to the following steps: A1. Obtain the important factors of each surface; A2. Determine the segmentation threshold of important factors and divide the model; A3. Generate a simplified model and build a shape LOD; Step 2: Use the cumulative method to load the shape LOD, and use the traditional switching method to load the texture LOD. 2. the self-adaptive LOD model construction method based on face clustering as claimed in claim 1, is characterized in that: the important factor that obtains each face described in step 1 is carried out by the following steps: For each face of the building model, define the important factor of its contribution to the appearance of the whole building model, set the sum of the important factors of all faces to 1, set the sum of the important factors of the simplified model to E, 0<E≤1; for The sum E of the important factors of the same size obtains the combination with the least number of faces. 3. the self-adaptive LOD model construction method based on surface clustering as claimed in claim 1, is characterized in that: the segmentation threshold of determining important factor described in step 1 and dividing model are carried out by the following steps: Arrange all the faces according to the size of the important factors, and divide the faces of the building model by setting the segmentation threshold of more than two important factors to obtain the result of face clustering, thereby dividing the model into multiple parts; and then divide these parts The parts are combined to get different levels of shape LOD. 4. the self-adaptive LOD model construction method based on face clustering as claimed in claim 1, is characterized in that: generate simplified model described in step 1, build shape LOD and carry out by the following steps: All the faces within the threshold constitute the simplified model of this level; through different segmentation thresholds, the simplified models of different levels are divided, and the shape LOD is constructed by combining the simplified models of different levels. 5. the self-adaptive LOD model construction method based on face clustering as claimed in claim 1, is characterized in that: carry out LOD model texture optimization as described in step 1 and carry out by the following steps: Set the fine texture of the LOD model texture to T1, and set the simplified texture of the LOD model texture to T2; For fine texture T1, optimize fine model image size: Set the texture standard resolution used for mapping as R ₁ (r ₁ x, r ₁ y), where r ₁ x and r ₁ y are the resolutions of the texture standard used for mapping in the x and y directions of the coordinate axis, respectively ; Set the resolution of the fine model picture as R ₂ (r ₂ x, r ₂ y), where r ₂ x and r ₂ y are the resolutions of the fine model picture in the coordinate axis x direction and y direction respectively; set the fine model The reduction factor of the picture is N(nx,ny), nx and ny are the reduction factors of the fine texture in the x direction and y direction of the coordinate axis respectively; calculate

Get the reduction factor of the fine model picture; reduce the length and width of the fine model picture to the original 1/N(nx,ny); Use 3Dmax's rendering-to-texture technology to combine all the simplified textures T2 into one image, and use the resampling method to reduce the combined resolution of the simplified texture T2 according to actual needs.

Images