WO2012171312A1

WO2012171312A1 - Ubiquitous terminal-oriented three-dimensional mesh model continuous multiresolution coding method

Info

Publication number: WO2012171312A1
Application number: PCT/CN2011/084594
Authority: WO
Inventors: 朱为鹏; 罗笑南
Original assignee: 中山大学
Priority date: 2011-06-15
Filing date: 2011-12-24
Publication date: 2012-12-20
Also published as: CN102324107A; CN102324107B

Abstract

Disclosed is a ubiquitous terminal-oriented three-dimensional mesh model continuous multiresolution coding method. The method comprises: a server end using a feature-enhanced quadric error metric method to simplify a mesh model; the server end constructing a resolution continuously-adjustable multiresolution mesh structure on the basis of the simplification process; the server end transmitting the resolution continuously-adjustable multiresolution mesh structure to a ubiquitous terminal; the ubiquitous terminal determining a resolution size corresponding to the received multiresolution mesh structure on the basis of the resolution size of a terminal screen and of resolution information carried by the mesh structure. Implementation of embodiments of the present invention prevents the problem of the three-dimensional mesh model having excessive loss in mesh feature when the resolution is relatively low, thus preventing the problem found in a multiresolution construction method of severe distortion when at relatively low resolution, and additionally, improving cases of mismatch between the model resolution and the screen resolution.

Description

A three-dimensional mesh model continuous multi-resolution coding method for universal terminals

The invention relates to the technical field of multi-resolution modeling and reconstruction of a three-dimensional mesh model, in particular to a multi-resolution modeling and reconstruction technology in the field of mobile graphics, in particular to a continuous multi-resolution of a three-dimensional mesh model for a universal terminal. Rate coding method. Background technique

With the continuous development of wireless networks and mobile computing technologies, people are increasingly demanding real-time and interactive operations for mobile 3D graphics applications. However, the computing power and memory capacity of the mobile computing terminal are relatively low, the display screen is small, the size is not standardized, and the battery life is limited. These factors greatly restrict the application of interactive 3D graphics in mobile computing terminals.

Multi-resolution modeling techniques are a better solution. Multi-resolution representation techniques use a uniform data structure to characterize model representations of varying degrees of sophistication. The specific fineness can be controlled by a variable. By assigning a value to the variable, a simplified model result of a certain degree of fineness can be obtained, thereby dynamically and randomly generating a three-dimensional mesh model of different fineness according to the user's needs.

When drawing a multi-resolution representation, according to the rendering performance of the hardware rendering platform, select the appropriate level detail model to achieve a better balance between drawing speed and drawing quality. From the perspective of controlling the rendering error, the commonly used model selection strategy is as follows:

The cone judgment strategy, also known as the viewpoint-based hierarchical detail model selection strategy, is mainly to roughen the hierarchical detail model outside the cone;

The backside judgment strategy, which coarsens the hierarchical detail model facing away from the line of sight;

The area judgment strategy refines the hierarchical detail model with a large projected area of the screen space;

A contour retention strategy that refines the hierarchical detail model located near the contour;

The curvature judgment strategy refines the area where the curvature changes greatly.

However, the biggest problem caused by the above-mentioned viewpoint-based hierarchical detail model selection strategy is that the amount of calculation is too large to be applicable to a pervasive computing environment with limited hardware computing power. In addition, the hardware performance of the rendering platform, such as the screen size of the mobile terminal, is not fully considered from the perspective of controlling the error of the model. Not the same as the actual situation.

When adjusting and controlling the resolution of the model, most of the current algorithms are controlled by the number of triangles of the 3D mesh model, the number of vertices, or the simplified error of the approximation model. Since the model's own size and screen size are not considered for the resolution of the model, it is usually necessary to manually intervene and make multiple adjustments to obtain a model that satisfies the accuracy requirements without wasting resources due to excessive redundant details. . As a result, on mobile terminals where the screen size cannot be unified, display roughness is sometimes caused by using a mesh that is too fine, and even system resources are wasted and it is difficult to operate; sometimes, a mesh model that is too rough or has too much error is used. Can not achieve satisfactory visual effects. Summary of the invention

The object of the present invention is to overcome the deficiencies of the prior art. The present invention provides a three-dimensional mesh model continuous multi-resolution coding method for a universal terminal, and constructs a multi-resolution mesh model by folding the mesh, and Record the area of the two triangular patches removed by the edge folding operation to achieve a mesh model that can be applied to any topological structure, and to provide an error-controlled continuous resolution mesh.

In order to solve the above problems, the present invention provides a continuous multi-resolution coding method for a three-dimensional mesh model of a universal terminal, the method comprising:

The server side uses a feature-enhanced quadratic error measure method to simplify the mesh model; the server side constructs a multi-resolution mesh structure with continuously adjustable resolution based on the above simplified process;

Transmitting, by the server end, the continuously adjustable multi-resolution grid structure to the universal terminal, the universal terminal determines, according to the resolution of the terminal screen and the resolution information carried by the grid structure, The resolution size of the resolution grid structure.

Preferably, the step of simplifying the mesh model by using the feature-enhanced quadratic error measurement method on the server side comprises:

Calculating a corresponding discrete mean curvature value and a quadratic error measure value at each vertex of the mesh; calculating corresponding features of each mesh edge according to the corresponding discrete mean curvature value and the quadratic error measure value at each vertex of the mesh Enhanced quadratic error measure;

All mesh edges are sorted according to the size of the quadratic error measure value corresponding to each feature of the mesh edge, and the edge with the smallest quadratic error measure value is deleted and a new network model is obtained; It is judged whether the detailed information of the new network model is simplified, and if so, the simplification process is ended and the most basic base network ^{M is obtained} . If not, return to continue to calculate the corresponding discrete mean curvature value and quadratic error measure value at the vertices of each mesh.

Preferably, the step of constructing, by the server side, the continuously adjustable multi-resolution grid structure based on the simplified process described above comprises:

The most simplified base mesh for recording simplification is ^M . ;

The progressive mesh model is established according to the quadratic error measure simplification process based on feature enhancement: MultiR _M = {M ₀ , {split ₀ , split, split _n } } ^ where W holds the information including: the vertex pair corresponding to the contracted edge ( ^V 1, ^V 2), two vertices ^V ' and ^ν '· adjacent to ^', ¹ ^, the type corresponding to the edge contraction, and the resolution information ^R i corresponding to the mesh model.

Preferably, the step of determining, by the ubiquitous terminal, the resolution size corresponding to the received multi-resolution grid structure according to the resolution size of the terminal screen and the resolution information carried by the grid structure comprises:

The ubiquitous terminal calculates an area corresponding to each pixel according to the envelope box size and the display resolution of the terminal screen according to the envelope box size of the most simplified base network calculation model received from the server end;

Decoding reconstruction according to the received data of the three-dimensional mesh model and the area corresponding to each pixel, and obtaining the highest resolution model that the current universal terminal screen can display.

Preferably, the multi-resolution grid structure with continuously adjustable resolution is transmitted to the ubiquitous terminal by a progressive transmission method over a wireless network.

By implementing the embodiments of the present invention, it is possible to avoid the excessive loss of the mesh features that occurs when the three-dimensional mesh model has a small resolution, thereby avoiding the severe distortion that occurs when the multi-resolution construction method encounters at a small resolution. In addition, according to the correspondence between the grid resolution and the display resolution of the ubiquitous terminal, the resolution of the grid model is recorded by the average area of the deleted triangle corresponding to the edge folding operation, and then adaptive according to the screen resolution information of the terminal. The mesh model resolution is selected, so that the model resolution does not match the screen resolution.

DRAWINGS

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the embodiments or the description of the prior art will be briefly described below. Obviously, the drawings in the following description are only Some embodiments of the present invention, for those of ordinary skill in the art, do not pay Other drawings can be obtained from these drawings on the premise of creative labor.

1 is a schematic flow chart of a three-dimensional mesh model continuous multi-resolution encoding method according to an embodiment of the present invention; FIG. 2 is a schematic flowchart of a specific embodiment of the method of the present invention;

3a, 3b, 3c, and 3d are diagrams showing the effect of binarization and resolution selection using the three-dimensional mesh model continuous multi-resolution encoding method of the embodiment of the present invention. detailed description

BRIEF DESCRIPTION OF THE DRAWINGS The technical solutions in the embodiments of the present invention will be described in detail with reference to the accompanying drawings. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments of the present invention without creative work are within the scope of the present invention.

FIG. 1 is a schematic flowchart of a three-dimensional mesh model continuous multi-resolution encoding method according to an embodiment of the present invention. As shown in FIG. 1, the method includes:

5101, the server side uses a feature-enhanced quadratic error measure method to simplify the mesh model;

5102, the server side constructs a multi-resolution mesh structure with continuously adjustable resolution based on the above simplified process;

5103, the server end transmits the continuously adjustable multi-resolution grid structure to the universal terminal;

5104. The ubiquitous terminal determines, according to the resolution of the terminal screen and the resolution information carried by the grid structure, the resolution size corresponding to the received multi-resolution grid structure.

Further, S101 may further include:

Calculating the corresponding discrete mean curvature value and the quadratic error measure value at the vertices of each mesh; calculating the feature enhancement corresponding to each mesh edge according to the corresponding discrete mean curvature value and the quadratic error measure value at each mesh vertex Quadratic error measure

According to the size of the quadratic error measure value corresponding to each feature of the mesh edge, all the grid edges are sorted, and the edge of the quadratic error measure value is deleted, and a new network model is obtained;

It is judged whether the details of the new network model are simplified, and if so, the simplification process is ended and the most simplified base network ^{M is obtained} . If not, return to continue to calculate the corresponding discrete mean curvature value and quadratic error measure value at the vertices of each mesh.

S102 includes:

The most simplified base mesh for recording simplification is ^M . ; A progressive mesh model is established according to the simplified process of quadratic error measurement based on feature enhancement: M _U ltiR— M = \M _Q , , _S plit. , _S plU , _S plU„,,,, wherein the saved information includes: a vertex pair corresponding to the contracted edge ( ^ν ι ' ^ν 2), and two vertices ^v ' adjacent to ( ^ν ι, ^ν 2) and ^ν ' , the type corresponding to the edge contraction, and the resolution information corresponding to the mesh model.

The S103 can be implemented by: transmitting a continuously adjustable multi-resolution grid structure to a ubiquitous terminal by a progressive transmission method over a wireless network.

S104 includes:

The most simplified universal terminal receives from a network-based server ^Μ. Calculating the envelope size of the model, and calculating the area corresponding to each pixel according to the size of the envelope box and the display resolution of the terminal screen;

According to the received data of the three-dimensional mesh model and the corresponding reconstruction area of each pixel, the highest resolution model that can be displayed on the current universal terminal screen is obtained.

In a specific implementation, the feature-enhanced edge folding simplification process used in the embodiment of the present invention is as follows: Step 1: Calculate a corresponding discrete mean curvature value of each mesh vertex;

Step 2: Calculate the corresponding quadratic error measure value of each mesh vertex;

Step 3: Calculate a quadratic error measure value based on the feature enhancement corresponding to each mesh edge according to the discrete mean curvature value and the quadratic error measure value corresponding to each vertex;

Step 4: Sort the mesh edges according to the quadratic error measure value corresponding to the feature enhancement corresponding to the mesh edge, and perform the edge folding operation on the edge with the smallest quadratic error measure value to obtain a new mesh model, and repeat step 1 4, until the base of the most simplified mesh ^Μ. ; (continue to simplify the larger features or contour information that will lose the mesh model);

Step 5: Record the minimum information required to restore each edge folding simplification operation and the average area of the triangular patches removed by the edge folding operation;

Step 6: According to the grid-based process corresponding to the base mesh and step 5 and the corresponding resolution information, construct a multi-resolution grid structure with continuously varying resolutions " ^{3⁄4 ?} ^ , ^, ^^..., ^^ .

In order to correctly restore the original topology structure during mesh reconstruction and to clarify the current resolution information, the following information needs to be saved: The vertices corresponding to the contracted edges are adjacent to (A) Vertex ^v ' and ^ν '·, the type corresponding to the edge contraction (whether it is the boundary edge), and the resolution information corresponding to the mesh model obtained after the current operation. The resolution information corresponding to the mesh model is defined as follows:

^R i = , where 4 is the average area of the triangular patches deleted by the current edge folding operation.

When the multi-resolution mesh structure is built, it can be applied to the mesh transmission in the ubiquitous environment. Grid applications for transmission and ubiquitous terminals. The application process can be divided into the following steps:

Step 1: First, transmit the base grid over the wireless network on the server side;

Step 2: When the mobile terminal receives from the server side through the network, according to the above algorithm, the progressive mesh model is compiled ^tiR - ^Μ = { ^{Μ s} p ^!it . '-ψ^ · · · , } }, first based on the received minimum base mesh. Calculate the envelope size of the model χ and then display it according to the envelope box size and the terminal screen

2 fraction "" counts the area of each pixel, ph w' ^! , since at least three pixels are required to render each triangle, so when continuing to receive ^ζΥ ' information, the current split operation recorded in it The resulting new triangle has an average area of 4 compared to

^{When 4 ≤ 3 χ Α} ", it indicates that the new triangular patch produced by this splitting operation will form a rough point due to repeated rendering.

Since ^s P ^lit i is not sorted according to the average area of the new triangle generated by the split operation, the average area of new triangles generated from the split operation to ^S P ^Ui '', ² ,..., 4 may be greater than

^{3 x} A. However, since the ^s P ^Ut w to the corresponding edge contraction weights are smaller than the corresponding edge contraction weights, that is, from the 3⁄4^„ each edge contraction operation has less influence on the visual effect of the model. The effect of the effect is already small enough to distinguish beyond the screen resolution.

Step 3: When the ^ubiquitous terminal receives the detailed information ^z K ^Q ^≤ ^z ' ^≤ W, judges 4 and ³ according to the resolution information of the ^ record (that is, the average area of the triangular patches deleted by the corresponding edge folding operation 4) <The relationship between A, if 4 > ^3x A, indicates that for the current mobile terminal, the recorded model details are necessary; if 4· ^{≤ 3χ} , then for the current mobile terminal, The recorded model details have exceeded the resolution of the screen and stopped receiving data.

2 is a schematic flow chart of a specific embodiment of the method of the present invention. As shown in FIG. 2, the method includes:

5201, input a three-dimensional mesh model;

5202, calculating a quadratic error measure value of each vertex of the model based on feature enhancement;

5203, performing edge folding simplification according to the quadratic error measure value based on feature enhancement;

5204, obtaining an average area corresponding to the triangle piece that is simplified by the side folding;

5205, constructing a multi-resolution grid according to a simplified process;

5206, transmitting a multi-resolution grid;

5207, calculating, by the base grid, a mesh resolution matching the terminal;

5208, continue to receive the detailed information until the detailed information exceeds the resolution resolution of the terminal resolution;

5209, generate the final model.

The feature enhancement value corresponding to the edge folding simplification method in the method of the invention is applied to the grid The way the feature is described. In order to calculate the quadratic error measure based on feature enhancement corresponding to the mesh edge, in order to enhance the error measure corresponding to the feature region in the original model, the algorithm adds a vertex feature value based on the quadratic error measure to represent the vertex. The importance of the feature. The vertex feature value is given by the curvature of the vertex region and the sum of all its neighboring edge lengths. The feature area of the model surface usually corresponds to a large normal vector variation value, that is, the curvature is large. Therefore, the larger the curvature, the more the position of the vertex can represent the physical features of the model. The longer the side length adjacent to the vertex, the area affected by the vertex on the surface of the model: I or larger.

Know all with vertices! The normal vector direction of adjacent triangle patches, then the normal vector of vertex V can be calculated as follows:

«ν = ∑ ⁿ p

P e P(v)

Where ^P ( ^v ) represents the set of all triangle patches adjacent to vertex v; is the normal vector of the adjacent triangle patch. ⁼ " 1, normalize the normal vector of the adjacent triangle patch and average it to get the corresponding normal vector at the vertex. Then normalize the processing: = /| |.

According to , the relative curvature at the vertex V can be calculated as follows:

Arccos(n _p '·η _ν ')

C =

I corpse (v)

Where l ^p(v )l represents the number of contiguous triangles of the vertex v.

It is known that the mesh edges ( ^V ', ^V 2) are composed of vertices ^V ', ^V 2 , and the feature weights ^ ^V corresponding to this edge can be calculated by the formula (2.6) as follows:

(v _l2 ) = / ₁₂ x +-( _Cl +c ₂ ) where ^ is the edge of the mesh, ^V 2).

After obtaining the feature weight ^(V '2) of the mesh edge, it is superimposed on the quadratic error measure method, and the shrinkage weight corresponding to the operation of the mesh edge ( ^V ', contracted to the new vertex V) can be calculated. The values are as follows:

Δ '((v, , v ₂ )→v) = v ^T (Q _l + Q ₂ )v + A (v, ₂ ) where A is a feature enhancement weight coefficient corresponding to the feature weight ^F(V i2). When =0, it corresponds to the standard quadratic error measure simplification algorithm. The larger the value is, the more obvious the local feature enhancement effect is. However, when the value is too large, it may have an adverse effect on the overall corridor information retention of the mesh model.

The quadratic error measure edge folding simplification process using this feature enhancement is performed in two steps: First, initial processing is required to calculate the quadratic error measure value and feature weight corresponding to each mesh vertex. And further derive the folding cost of each mesh edge and the position of the newly generated mesh vertex after folding. Then iteratively, the grid edges are sorted according to the folding cost, and a small top stack is built accordingly.

In each simplification process, the edge with the lowest cost of stacking is taken out for folding operation, and then the quadratic error measure and feature weight of the neighborhood vertices affected by this operation are updated, and the corresponding edge folding cost is recalculated. . The order of the edges to be folded in the heap is adjusted according to the new edge folding cost, and a small top stack is formed again. The side folding operation is repeated until the approximation model satisfying the simplified condition is obtained.

After each folding operation, all the mesh edges need to be reordered according to the shrinkage weight value. The efficiency of the sorting algorithm will affect the efficiency of the whole binning algorithm to a large extent. Features of edge sorting in the analysis algorithm: In addition to the first edge contraction, each sorting operation essentially inserts several records into an ordered sequence and maintains the order of the resulting sequence. Therefore, in the algorithm of this paper, after all the meshing edge weightings are sorted for the first time, each sorting is no longer performed on all edges, except that all the ^V2 connected in the ordered edge table generated by the last sorting is deleted. Edge, then insert the newly generated edge orderly according to the weight value.

When performing the edge folding operation, it is necessary to first calculate the geometric position of the new vertex generated after the edge is contracted. The determination method of the new vertex will directly lead to the difference of the simplified error corresponding to the edge folding, thereby affecting the order of the edge folding during the model simplification process. And affect the accuracy of the final simplified model. The main principle for determining the position of a new vertex is to make the simplified model as close as possible to the original model.

When performing the edge folding operation ( ^V 1, ^V 2 ) ^{→ V} , a simple way to select the vertex position: The position of the new vertex is selected from the two vertices and the midpoint of the folded edge, ie in ^^, ^ Among the three +^^, the corresponding minimum of the grid error will be selected as the new vertex position. The main advantage of this method is that the amount of calculation is small, which can greatly improve the operation speed. In addition, since the vertices of the simplified model are mostly a subset of the original model vertex set, the contour and geometric features of the original model can be better preserved.

For non-closed triangular mesh models, consideration must be given to maintaining the original boundary as much as possible when dealing with boundary edges. Therefore, when the boundary edge is contracted, the new vertex position needs to satisfy the condition of maintaining the original boundary. If both ends of the contracted edge are on the boundary, the edge must be maintained to maintain the original boundary, that is, the edge with both endpoints on the boundary does not have a contractible condition, and the weight cannot be calculated. The table is sorted. If the contracted edge has one and only one end point is on the boundary, the edge shrinks to the end point on this boundary.

The structure of the progressive mesh model based on the above mesh simplification method is as follows:

MultiR—M = ₀ , {split _Q , split _{ split _n }] Where ^M. In order to perform the "+ 1 edge folding operation on the original mesh, the simplified mesh model is obtained, and the information required for each step split operation and the corresponding resolution information are recorded and stored in an ordered linked list. + +··· + · You can get a cylindrical mesh after performing the "- secondary edge shrink operation on the original mesh.

In order to correctly restore the original topology structure during mesh reconstruction and to clarify the current resolution information, the following information needs to be saved: the vertex pair corresponding to the contracted edge ( ^V 1, ^V 2 );

Two vertices ^ and ^V '' adjacent to ^, ^);

The type corresponding to the edge contraction (whether it is a boundary edge);

The resolution information corresponding to the mesh model obtained after the current operation is performed.

Where: "The vertex pair corresponding to the contracted edge ( ^V 1 ' ^V 2 )" corresponds to the geometric position information of the new vertex generated by the point splitting operation, "two vertices adjacent to ( ^V I, ^V 2 ) ^V 'and ^V '" and "type corresponding to edge contraction" are used to restore the connection between the new vertex and the original mesh vertex.

The resolution information corresponding to the mesh model obtained after the current operation is performed is used to quantize the resolution state corresponding to the current hierarchical detail model, and the mobile terminal adaptively selects an appropriate hierarchical detail model according to the resolution precision of the screen. Resolution information '. The definition is as follows:

R _t = 4

4 is the average area of the triangle patch deleted by the current edge folding operation ^.

The three-dimensional mesh model continuous multi-resolution coding method of the embodiment of the present invention is terminal-oriented.

3a, 3b, 3c, 3d are diagrams of the binarization and resolution selection of the 3D mesh model continuous multi-resolution coding method according to the embodiment of the present invention, wherein FIG. 3a is an input initial 3D mesh model; 3b is the constraint posture; Fig. 3c is the deformation texture extracted by the method of the present invention; Fig. 3d is the result obtained after the deformation.

In the embodiment of the present invention, it is also possible to adaptively select a three-dimensional mesh model on a mobile terminal. The resolution can ensure the visual effect of the model as much as possible while avoiding waste of limited hardware resources. In the construction of progressive mesh, the feature-enhanced edge folding simplification algorithm can be applied to the mesh model of any topology structure, and the model can be well maintained when down-sampling to lower resolution, thus ensuring a simplified model. The visual effect; and the adaptive selection of the model resolution can be easily realized, which can avoid the waste of system resources and even the operation difficulty when the 3D mesh model is applied on the mobile terminal, and provide the best possible The visual effect of the model can better guarantee the quality of the multimedia service in the form of a grid model on the mobile terminal.

The above description of the three-dimensional mesh model continuous multi-resolution coding method provided by the embodiment of the present invention is only for helping to understand the method and the core idea of the present invention. Meanwhile, for those skilled in the art, The present invention is not limited by the scope of the present invention.

Claims

A method for encoding a continuous multi-resolution of a three-dimensional mesh model for a universal terminal, the method comprising:

The server end transmits the continuously adjustable multi-resolution grid structure to the universal end. The ubiquitous terminal determines the location according to the resolution of the terminal screen and the resolution information carried by the grid structure. The resolution size of the received multi-resolution grid structure.

2. The method according to claim 1, wherein the server end uses a feature-enhanced quadratic error measurement method to simplify the mesh model. Includes:

Sorting all grid edges according to the magnitude of the quadratic error measure value corresponding to each feature of the mesh edge, and deleting the edge with the smallest quadratic error measure value to obtain a new network model;

It is judged whether the detailed information of the new network model is simplified, and if so, the simplification process is ended and the most simplified base network ^{M is obtained} . If not, return to continue to calculate the corresponding discrete mean curvature value and quadratic error measure value at the vertices of each mesh.

The method for continuous multi-resolution coding of a three-dimensional mesh model for a universal terminal according to claim 1 or 2, wherein the server end constructs a multi-resolution network with continuously adjustable resolution based on the simplified process described above. The steps of the lattice structure include:

The most simplified base mesh for recording simplification is ^M . ;

A progressive mesh model is established according to the simplified process of quadratic error measurement based on feature enhancement: M _U ltiR—M =, M. ,, spUt. , split, sp _n ]] , where the saved information includes: the edge pair being shrunk The resolution information ^R i corresponding to the lattice model.

4. The method according to claim 3, wherein the ubiquitous terminal is based on a resolution of the screen of the terminal and a resolution information carried by the grid structure. The determining the size of the resolution corresponding to the received multi-resolution grid structure comprises: the ubiquitous terminal according to the envelope box size of the most simplified base network calculation model received from the server end, according to the Calculating the area corresponding to each pixel by the envelope size and the display resolution of the terminal screen;

The method of claim 1, wherein the server end transmits the continuously adjustable multi-resolution grid structure to the unidirectional terminal. The steps of the universal terminal are specifically as follows: To the universal terminal.