WO2016125754A1

WO2016125754A1 - Particle rendering processing device, particle rendering method, and computer program

Info

Publication number: WO2016125754A1
Application number: PCT/JP2016/052954
Authority: WO
Inventors: 耕二小山田; 尚久坂本
Original assignee: 国立大学法人京都大学
Priority date: 2015-02-03
Filing date: 2016-02-01
Publication date: 2016-08-11
Also published as: JP6719774B2; JPWO2016125754A1

Abstract

The purpose of the present invention is to suppress an increase in the processing load for particle rendering. Provided is a particle rendering processing device that comprises: a storage unit 20 that stores target data 21 configured by including matrix data in which numerical data is defined for the nodes of a matrix; and a processing unit 10 that performs particle rendering processing 100 on the target data 21. The particle rendering processing 100 includes: processing 101 for estimating the particle density ρ_i of a node i in a case where one particle P_i is provided to the node i, such processing performed on the basis of the total content V_i of one or a plurality of matrices E which have the node i in common; processing 103 for determining, from numerical data S_i defined for the node i, the opacity αi of the node i; processing 104 for generating, at the node i, the one particle P_i that has a size determined on the basis of the particle density ρ_i and the opacity α_i of the node i; and processing 105 for generating image data 401, which is the generated particle projected onto an image surface.

Description

Particle rendering processing apparatus, particle rendering method, and computer program

The present invention relates to particle rendering.

Particle rendering is known as a technique for visualizing large-scale volume data (see Non-Patent Document 1). Particle rendering is a technique for rendering a given volume data with opaque luminous particles.

The particle rendering process includes two processes: particle generation and particle projection. Prior to particle generation, it is necessary to estimate the particle density for each grid constituting the volume data. The particle density is the number of particles per unit volume. The particle density for each lattice is obtained from numerical data defined for the lattice using opacity, particle radius, and the like. In Non-Patent Document 1, the opacity is calculated from numerical data defined in the grid using a transfer function specified by the user.

In the particle generation process, the number of particles according to the particle density is generated, and the generated particles are arranged in the lattice. In the particle projection process, the generated particles are projected onto the image plane. The particles are projected onto the image plane and a pixel value is calculated at each pixel of the image.

Since conventional particle generation requires that a number of particles according to the particle density be arranged in the lattice, the processing load may become very large depending on the particle generation method. Visualization understands the distribution and characteristics of data while interactively changing the relationship between numerical data and opacity, so a particle generation process is required for each change. Therefore, when a very large amount of data is targeted, the processing time may be long.

Therefore, it is desirable to suppress an increase in the processing load of particle rendering.

An aspect of the present invention is a particle rendering processing apparatus, the storage unit storing target data configured to have grid data in which numerical data is defined at nodes of the grid, and a particle rendering process for the target data And a particle rendering process that determines the particle density of the node when one particle is arranged at the node based on a volume sum of one or more grids sharing the node. A first process to be estimated; a second process for determining opacity at the node from numerical data defined for the node; and a size determined based on the particle density and the opacity of the node. And a third process for generating one particle having the node at the node and a fourth process for generating image data in which the generated particle is projected on the image plane. Another aspect of the present invention is a particle rendering method including performing a particle rendering process. Another aspect of the present invention is a computer program that causes a computer to execute part or all of a particle rendering process. The computer program is recorded on a computer-readable recording medium.

According to the present invention, an increase in the processing load of particle rendering can be suppressed.

It is a block diagram of a particle rendering processing apparatus. It is explanatory drawing of a particle rendering process. It is a graph which shows the processing time of particle | grain rendering.

Hereinafter, preferred embodiments of the present invention will be described with reference to the accompanying drawings.
[1. Particle rendering processor]
As illustrated in FIG. 1, the particle rendering processing apparatus 1 according to the embodiment includes a processing unit 10 and a storage unit 20. The particle rendering processing apparatus 1 is configured by a computer having a processing unit 10 and a storage unit 20. The processing unit 10 executes the computer program stored in the storage unit 20 to cause the computer to perform particle rendering processing, and causes the computer to function as the particle rendering processing device 1. The processing unit 10 includes a CPU (Central Processing Unit). The processing unit 10 may include a GPU (Graphics Processing Unit) for image processing in addition to the CPU. Of the processing performed by the particle rendering processing device 1, the

particle projection processing

105 and 205 is preferably performed by the GPU instead of the CPU in order to increase the processing speed.

The particle rendering process performed by the processing unit 10 is a process of generating visualized

image data

401 and 402 from the target data 21 stored in the storage unit 20. The generated

image data

401 and 402 are stored in the storage unit 20. The particle rendering apparatus 1 has a monitor for displaying an image, and displays the generated

image data

401 and 402 on the monitor.

The target data 21 to be subjected to the particle rendering process is, for example, volume data. The volume data is composed of a plurality of grid data. The grid data is data in which numerical data is defined at the nodes (vertices) of the grid. The lattice may be a regular lattice, a structured lattice, or an irregular lattice. The shape of the lattice is, for example, a tetrahedron or a hexahedron, and may be another three-dimensional shape. The numerical data is, for example, scalar data. The scalar data is, for example, values indicating temperature, pressure, density, and energy.

An example of the volume data 21 is shown in FIG. The three-dimensional space in which the volume data 21 is defined is divided into small three-dimensional solids called a lattice E. Grid E has a plurality of nodes. The lattice E in FIG. 2 is a tetrahedron. Here, i is an identifier of a node (vertex) of the grid E. The grid E in FIG. 2 has four nodes indicated by node identifiers i = 0-3. For each node i = 0 to 3, numerical data such as scalar data S _i is defined. That is, the volume data 21 is configured to have the positions (coordinates) of the plurality of nodes i = 0 to 3 and the scalar data S _i associated with the nodes i = 0 to 3 respectively. As is apparent from the volume data 21 in FIG. 2, a node of a certain grid E is also a node of another grid. That is, a node may be shared by multiple grids.

The target data 21 is not limited to volume data that is three-dimensional data, and may be two-dimensional data (surface data) or one-dimensional data (line data). When the target data 21 is two-dimensional data, the lattice constituting the two-dimensional data is a planar two-dimensional lattice, and strictly speaking, the lattice thickness is zero. In this case, particle generation is performed by regarding the two-dimensional lattice as a plate-like three-dimensional lattice having a thickness equal to the diameter of the generated particles. Similarly, when the target data 21 is one-dimensional data, particle generation is performed by regarding a linear one-dimensional lattice as a rod-shaped three-dimensional lattice having a diameter equal to the diameter of the generated particles.

The particle rendering process performed by the processing unit 10 includes a first particle rendering process 100 and a second particle rendering process 200. As will be described later, the first particle rendering process 100 generates particles P only at the positions of the nodes V. For this reason, in the first particle rendering processing 100, an increase in processing load when particle generation is repeated is suppressed. Further, an increase in the memory area for holding the generated particle P data is also suppressed.

On the other hand, the second particle rendering process 200 generates a number of particles according to the particle density and places the generated particles P in the lattice E, as in the conventional particle rendering process. For this reason, in the second particle rendering process 200, the processing load tends to increase when particle generation is repeated. In addition, a large amount of memory area is required to hold the generated particle data. However, in the second particle rendering process 200, the distribution of the particles P corresponding to the particle density is obtained in the lattice E, so that the image quality of the generated image data 402 is better than that of the first particle rendering process 100. There is expected.

In the particle rendering process performed by the processing unit 10, a process selection 300 is performed to determine which of the first particle rendering process 100 and the second particle rendering process 100 is to be executed. In the process selection 300, whether to execute the first particle rendering process 100 or the second particle rendering process 100 on the target data 21 is determined based on a selection input by the user.

The first particle rendering process 100 with a low processing load can be used as a preview process for generating image data 401 for preview. When the user wants to obtain the image data 401 visualizing the target data 21 as a preview image, the user can obtain the preview image 401 efficiently in a relatively short time by selecting the first particle rendering process (preview process). be able to. After confirming the preview image 401, the user can efficiently perform the visualization work by obtaining the image data 401 with good image quality by the second particle rendering processing 200.
Note that the particle rendering processing apparatus 1 may not have the function of executing the second particle rendering process 200.

[2. First particle rendering process and second particle rendering process]
Hereinafter, the first particle rendering process 100 and the second particle rendering process 200 will be described. In the following, a single grid will be described, but the same processing is performed for other grids constituting the target data 21.

[2.1 First particle rendering process]
First particle rendering process 100, rather than the number of particles in accordance with the particle density as in the conventional particle rendering process of placing in the lattice, to produce particles P _i to the position of the node i.

As shown in FIG. 1, in the first particle rendering process 100, a particle density estimation process 101 is performed as the first process. The particle density estimation process 101 is a process for estimating the particle density ρ _i of the node i when one particle P _i is arranged at the node i. The particle density ρ _i estimated in the first particle rendering process 100 is estimated based on the volume sum V _i of one or more grids E sharing each node i = 1 to 3. In the first particle rendering process 100, one particle P _i is arranged at each node i, so that the particle density ρ _i is a volume around the node i, that is, one or more grids E sharing the node i. Calculated based on the volume sum V _i . More specifically, the reciprocal of the volume sum V _i of one or more grids E sharing the node i is the particle density ρ _{i of the} node i when one particle P _i is arranged at the node _i. It becomes. For example, when there are four lattices E sharing a certain node i, 1 / V _i which is the reciprocal of the volume sum V _i of the four lattices E becomes the particle density ρ _{i of the} node i.

In the first particle rendering process 100, an opacity update process (opacity determination process) 103 is further performed as a second process. The opacity update process 103 is a process for determining (updating) the opacity α _i of each node i = 0 to 3. The opacity α _i is calculated from numerical data (scalar data S _i ) defined for each node i using a transfer function specified by the user. The calculation of opacity is the same as in the conventional particle rendering process.

In the first particle rendering process 100, a particle radius calculation process 104 is further performed as a third process. The particle radius calculation process 104 is a process for determining the size of the particle arranged at the node i and generating a particle having the determined size. The size of the particle is determined by the radius r _i of the particle P _i , for example. The particle radius r _i of a node _i is determined based on the particle density ρ _i and opacity α _i of the node _i . More specifically, the particle radius r _i is calculated based on the following formula (1).

The above formula (1) is obtained by modifying the conventional particle density estimation formula described in Patent Document 1. The conventional particle density estimation formula is as the following formula (2).

In the conventional particle density estimation formula shown in Formula (2), when the opacity α and the particle radius r are determined, the particle density ρ can be obtained. Note that Δt is a ray segment length used in ray casting (see Non-Patent Document 1). In the conventional equation (2), the particle radius r is a user-specified value, and the opacity α is a value calculated from numerical data defined at the node using a user-specified transfer function. .

In the conventional equation (2), the particle density ρ is obtained from the opacity α and the particle radius r, whereas in the equation (1), the particle radius r _i of the node i is determined by the particle density estimation processing 101. It is obtained from the obtained particle density ρ _i and the opacity α _i obtained in the opacity update process 103.
As shown in FIG. 2, a particle P _i having the determined particle radius r _i is generated at the position of the node i.

In the first particle rendering process 100, a particle projection process 105 is performed as a fourth process subsequent to the particle radius calculation process 104. The particle projection process 105 is a process for generating image data in which the generated particles are projected on the image plane. The details of the particle projection process 105 are described in Non-Patent Document 1, and the particle projection process 105 of the present embodiment is also performed according to the description of Non-Patent Document 1.
Note that the processing unit 10 can repeatedly execute the processing 102 including the opacity update processing 103 and the particle radius calculation processing 104 and the particle projection processing 105 a plurality of times. When the process 102 is repeated, the generated image data can be changed by changing (updating) the opacity α _i each time the process 102 is repeated.

In the conventional particle rendering process, the particle density ρ is determined according to the opacity α as shown in the equation (2). Therefore, when the opacity α is updated, the particle density ρ is re-estimated to the particle density ρ. It is necessary to perform the process of generating a corresponding number of particles again. Therefore, if the opacity is updated and the process is repeated, the processing load increases accordingly.

On the other hand, in the first particle rendering process 100, since the particles P _i are arranged only at the node i, the particle density ρ _i indicating the number of particles per unit volume is determined regardless of the opacity α _i , Even if the opacity α _i is updated, it does not change. Further, the first particle rendering 100, the particles P _i, for placement only to node i, the process of generating the number of particles corresponding to the particle density is not necessary. Therefore, even if the processing is repeated, an increase in processing load is suppressed.

In the first rendering process, a plurality of pixel values obtained by repeating the process 102 a plurality of times are added, and finally, an ensemble average obtained by dividing by the number of repetitions is used as a final pixel value. Yes (see Non-Patent Document 1).

[2.2 Second particle rendering process]
Similar to the conventional particle rendering process, the second particle rendering process places a number of particles according to the particle density in the lattice.

As shown in FIG. 1, in the second particle rendering process 200, an opacity update process (opacity determination process) 202, a particle density estimation process 203, a particle generation process 204, and a particle projection process 205 are performed. . These processes are the same as the conventional particle rendering process.
That is, the opacity update process 202 is a process for determining (updating) the opacity α of each node i = 0 to 3. The opacity α _i is calculated from numerical data (scalar data S _i ) defined for each node i using a transfer function specified by the user.

In the particle density estimation process 203, the particle density ρ for each lattice E is estimated (see Non-Patent Document 1). The estimation of the particle density ρ is performed based on the opacity α and the user-specified particle radius r using the above-described equation (2).
In the particle generation process 204, the number of particles P according to the particle density ρ is generated, and the generated particles P are arranged in the lattice E. The particle generation method employed in the particle generation processing 204 is, for example, a uniform sampling method or a metropolis sampling method (see Non-Patent Document 1). The uniform sampling method is a method of generating particles on the assumption that the particle density in the lattice E is constant. The metropolis sampling method is a particle generation method that takes into account changes in particle density inside the lattice E. FIG. 2 shows an example of the particles P generated by the uniform sampling method and the metropolis sampling method.

In the second particle rendering process 200, as shown in FIG. 2, since a large number of particles P are arranged in the lattice E according to the particle density ρ, it is expected that the image quality of the image data 402 is improved.

In the second particle rendering process 200, a particle projection process 205 is performed following the particle generation process 204. The particle projection process 205 is a process for generating image data in which the generated particles are projected on the image plane (see Non-Patent Document 1).
The processing unit 10 can repeatedly execute the process 201 including the opacity update process 202, the particle density estimation process 203, and the particle generation process 204, and the particle projection process 205 a plurality of times. Further, the processing unit 10 can add a plurality of pixel values obtained by a plurality of repetitions, and finally set an ensemble average obtained by dividing by the number of repetitions as a final pixel value (non-patent document). Reference 1).

[3. Comparison of processing time]
FIG. 3 shows processing times of the first particle rendering process 100 and the second particle rendering process 200, respectively. In FIG. 3, “initial drawing” indicates the processing time when the series of processing from the particle density estimation processing 101 to the particle projection processing 105 is performed once for the first particle rendering processing 100. The process 200 indicates the processing time when the processes from the opacity update process 202 to the particle projection process 205 are performed once. In FIG. 3, “first iteration” means that the first particle rendering processing 100 is performed when a series of processing from the opacity update 103 to the particle projection processing 105 is performed again after “initial drawing”. The processing time is shown, and the second particle rendering processing 100 shows the processing time when processing from the opacity update processing 202 to the particle projection processing 205 is performed again after “initial drawing”. Further, in FIG. 3, “second iteration” is the first particle rendering processing 100, and after the “first iteration”, a series of processing from the opacity update 103 to the particle projection processing 105 is performed again. In the case of the second particle rendering process 100, the processing time when the processes from the opacity update process 202 to the particle projection process 205 are performed again after the “first iteration” is shown.

As is clear from FIG. 3, the second particle rendering process 200 requires a relatively long processing time for each repetition, whereas the first particle rendering process 100 requires a processing time for the repetition of the process after the initial rendering. Thus, an increase in processing load can be suppressed.

1 Particle Rendering Processing Device 10 Processing Unit 20 Storage Unit 21 Target Data 101 Particle Density Estimation Processing 103 Opacity Update Processing 104 Particle Radius Calculation Processing 105 Particle Projection Processing 401 Image Data

Claims

A particle rendering processing device,
A storage unit for storing target data configured to have grid data in which numerical data is defined at nodes of the grid;
A processing unit for performing a particle rendering process on the target data;
With
The particle rendering process includes:
A first process for estimating a particle density of the node when one particle is arranged at the node based on a volume sum of one or more grids sharing the node;
A second process for determining opacity at the node from the numerical data defined at the node;
A third process for generating one particle at the node having a size determined based on the particle density and the opacity of the node;
A fourth process for generating image data in which the generated particles are projected on the image plane;
A particle rendering processing apparatus.
The processing unit may repeatedly execute the second process, the third process, and the fourth process after executing the first process, the second process, the third process, and the fourth process. The particle rendering processing device according to claim 1, which is configured.
The particle rendering processing apparatus according to claim 2, wherein the opacity is updated each time the second process is repeated.
The particle rendering process includes:
A process selection for determining which of the first particle rendering process and the second particle rendering process is to be performed on the target data;
The first rendering process includes the first process, the second process, the third process, and the fourth process,
The second rendering process includes
A process for determining opacity at the node from the numerical data defined at the node;
Estimating the particle density of the lattice based on the opacity and a user-specified particle radius;
Generating a number of particles according to the particle density of the lattice, and placing the generated particles in the lattice;
A process of generating image data in which the generated particles are projected on the image plane;
The particle rendering processing apparatus according to any one of claims 1 to 3, further comprising:
The particle rendering processing apparatus according to claim 4, wherein the processing selection is determined based on a selection input by a user.
The particle rendering processing apparatus according to any one of claims 1 to 5, wherein, in the first processing, the particle density is a reciprocal of a volume sum of one or a plurality of lattices sharing the node.
The particle rendering processing apparatus according to any one of claims 1 to 6, wherein in the third process, the size of the particle is determined as a particle radius calculated based on the following expression.

here,
i is the node identifier r i is the particle radius α i of the node i is the opacity of the node i ρ i is the particle density of the node i Δt is the ray segment length used in ray casting
A particle rendering method,
Performing particle rendering processing on target data configured to have grid data in which numerical data is defined at nodes of the grid,
The particle rendering process includes:
A first process for estimating a particle density of the node when one particle is arranged at the node based on a volume sum of one or more grids sharing the node;
A second process for determining opacity at the node from the numerical data defined at the node;
A third process for generating one particle at the node having a size determined based on the particle density and the opacity of the node;
A fourth process for generating image data in which the generated particles are projected on the image plane;
Particle rendering method including:
The particle according to claim 8, wherein the second process, the third process, and the fourth process are repeatedly executed after the first process, the second process, the third process, and the fourth process are executed. Rendering method.
The particle rendering method according to claim 9, wherein the opacity is updated each time the second process is repeated.
Determining whether to perform a first particle rendering process or a second particle rendering process on the target data;
The first particle rendering process includes the first process, the second process, the third process, and the fourth process,
The second rendering process includes
A process for determining opacity at the node from the numerical data defined at the node;
Estimating the particle density of the lattice based on the opacity and a user-specified particle radius;
Generating a number of particles according to the particle density of the lattice, and placing the generated particles in the lattice;
A process of generating image data in which the generated particles are projected on the image plane;
The particle rendering method according to any one of claims 8 to 10.
The particle rendering method according to claim 11, wherein determining which of the first particle rendering process and the second particle rendering process is to be performed is based on a selection input by a user.
The particle rendering method according to any one of claims 8 to 12, wherein, in the first processing, the particle density is a reciprocal of a volume sum of one or a plurality of lattices sharing the node.
The particle rendering processing apparatus according to any one of claims 8 to 13, wherein in the third process, the size of the particle is determined as a particle radius calculated based on the following equation.

here,
i is the node identifier ρ i is the particle density of the node i r i is the particle radius α i of the node i is the opacity of the node i Δt is the ray segment length used in ray casting
A computer program for causing a computer to perform a particle rendering process on target data including a plurality of grid data in which numerical data is defined at nodes of the grid,
The particle rendering process includes:
A first process for estimating a particle density of the node when one particle is arranged at the node based on a volume sum of one or more grids sharing the node;
A second process for determining opacity at the node from the numerical data defined at the node;
A third process for generating one particle at the node having a size determined based on the particle density and the opacity of the node;
A fourth process for generating image data in which the generated particles are projected on the image plane;
A computer program containing