JP4656633B2 - Volume data rendering system and volume data rendering processing program - Google Patents

Description

  The present invention relates to a volume data rendering system and volume data rendering processing method for visualizing volume data defined in a three-dimensional space.

Various representation formats for three-dimensional computer graphics have been proposed.
In general, data of a three-dimensional object is often given in units of planes as a combination of polygon data and texture data. If polygon data is used, the outer surface shape of the three-dimensional object can be defined by a mesh-like wire frame. At the time of drawing (rendering), polygon data is arranged while adjusting the position and angle of the three-dimensional object according to the display scene, and drawing is performed by pasting texture data on the surface. It can be said that the expression by the polygon data is excellent as the expression format of the three-dimensional object.

  However, polygon data is not suitable as a representation format for objects and physical phenomena represented by the density and density of particles distributed in a three-dimensional space. For example, it is difficult to express “clouds” that are the concentration distribution of water vapor in the atmosphere as polygon data. Clouds absorb and scatter sunlight depending on the concentration distribution of water vapor, and are expressed as white or gray shading, but complex and light uneven patterns continuously spread in the space, and the outer surface shape is It is difficult to express with polygon data defined by a mesh-like wire frame. Similarly, it is difficult to express “smoke” as polygon data.

  Therefore, an expression format based on volume data has been studied as a data expression format suitable for expressing the density and density of particles distributed in a three-dimensional space. The volume data is three-dimensional array data in which each voxel has a feature amount and opacity such as density and density of particles distributed in a three-dimensional space. Here, the voxel is a lattice-like small cube obtained by finely dividing a three-dimensional space. By filling the three-dimensional space with voxels and giving the feature quantity to each voxel, it is possible to appropriately define the feature quantity such as density and shading of the particles distributed in the three-dimensional space. Here, various features are assumed and may be given as high-dimensional data. In the above-described cloud example, brightness and hue are assumed as feature amounts.

  On the other hand, opacity is introduced to give a “blur” or “watermark” effect to a three-dimensional object. By introducing opacity into each voxel, it is possible to give a three-dimensional object a translucent state expression or a so-called “blurring” effect like fog.

  By introducing volume data as described above, it is possible to appropriately define a three-dimensional object such as a cloud or smoke distributed with a spread in a three-dimensional space.

  Next, a rendering process for a three-dimensional object defined using volume data will be described.

FIG. 10 is a diagram showing a basic flow of a rendering processing procedure when volume data is used.
First, volume data is acquired (S1). The volume data may be prepared in advance or may be created in real time by observing a three-dimensional object. It is preferable to perform preprocessing such as noise removal and image enhancement necessary based on the acquired volume data.

After this pre-processing, the luminance value and opacity are determined for each voxel (S2), and a voxel data set volume is formed.
Next, a final luminance value for each pixel is obtained by ray casting processing. The ray casting process includes a sampling process (S3) and an integral calculation process (S4) based on the luminance value and opacity of the sampled voxel.

FIG. 11 is a diagram schematically illustrating the concept of ray casting processing.
In the ray casting, a line-of-sight (ray) 1202 is given in each direction from the viewpoint, and the voxels 1201 distributed in the volume voxel data set 1200 in the three-dimensional space are sampled along the ray 1202 and the sampling is performed. This is a method of adding the feature amount (luminance value) of each voxel 1201. If opacity is given to each voxel, the inside is visualized by a semi-transparent display by ray casting. In ray casting, the product of the brightness value and opacity of each voxel is added sequentially, and when the sum of opacity becomes 1 or the ray penetrates the 3D object, the processing for that pixel is finished. The addition result is displayed as the pixel value 1301 of the drawing surface 1300.

If the luminance value and opacity at the current sample position are i and α, respectively, the luminance value I _out of the pixel by ray casting is expressed by the following equation 1 when the luminance value at the time of line-of-sight input is I _in. It is represented by

  The pixel value 1301 to be finally drawn is obtained by integrating all the sampled voxels according to Equation 1 above.

In the prior art, volume data rendering processing is performed according to the above procedure, and drawing processing (S5) is performed on each pixel on the monitor. However, the following problems have been pointed out.
The first is a high calculation cost. Since the volume data is calculated based on each voxel spreading in a three-dimensional space, complicated and enormous calculation is required.
Second, the amount of memory consumed is large. Since it is necessary to handle the luminance value and opacity of each voxel spreading over the entire three-dimensional space as data, a large amount of memory is consumed.

  In order to solve these two problems and perform volume data rendering at high speed, it was necessary to develop a dedicated processor equipped with special resources such as a dedicated accelerator and a large amount of memory. .

  In particular, when rendering volume data with movement like smoke in the virtual reality space instead of displaying a stationary object like a cloud in a stationary state, every time smoke moves or an observer Each time the moves, the complexity increases in real time, increasing the difficulty.

  JP-A-5-266216 and JP-A-2003-263651 are cited as conventional techniques aiming to solve this problem. Both of these are techniques for introducing a slice-like surface model representation format into the volume data representation format and simplifying the rendering process inside the volume data.

FIG. 12 is a diagram schematically showing a concept of introducing a sliced surface model expression format into the volume data expression format.
In the prior art shown in FIG. 11, the sampling processing and luminance value product-sum calculation processing performed for each ray are performed independently. In the prior art shown in FIG. 12, volume data is sliced and divided. A surface is described as polygonal data as a surface model. By preparing volume data as a stack of surface models for each slice plane in this manner, sampling processing and luminance value processing for each ray can be performed collectively for each slice plane. For example, if the opacity is given to the surface model of each slice surface, the final luminance value of each pixel can be determined as if a plurality of translucent sheets are viewed through. Further, when the observer moves in the volume, it is possible to perform pseudo-drawing inside the volume by performing high-speed processing of two-dimensional drawing of the slice surface corresponding to the viewpoint with polygon data.

Japanese Patent Laid-Open No. 5-266216 JP 2003-263651 A

  As described above, introducing a slice-like surface model representation format into the volume data representation format can reduce the calculation cost, and even if the observer moves in a virtual reality space, In addition, a three-dimensional drawing process inside the volume can be executed.

However, the conventional rendering processing technique has the following problems.
The first problem is that naturalness is impaired. It is preferable to render the acquired volume data as close to the physical phenomenon as possible. However, if rendering is performed with the above-described conventional technology, a slice-like surface model is introduced into a three-dimensional space that is essentially free space. Therefore, a planar restraint relationship is generated between the voxels, and the naturalness is impaired. That is, smoke and the like are spread in a three-dimensional space, but since a slice-like surface model is introduced, voxels are described as triangular polygon data on the slice plane (Japanese Patent Laid-Open No. 5-266216). The feature of spreading in a three-dimensional space cannot be expressed.

  The second problem is a problem of fluctuation expression. Fluctuation data is not given to the acquired volume data itself, but smoke and the like flow in the three-dimensional space with fluctuations not only in the vertical and horizontal directions but also in the forward and backward directions. Can be drawn so as to represent a state closer to the physical phenomenon of the entity. However, if rendering processing is performed according to the above-described conventional technology, it is impossible to express such fluctuations of an entity such as smoke at a low calculation cost.

  The third problem is the problem of expressing the velocity of particles. The acquired volume data includes macro volume data movement. For example, in the case of smoke, it seems that the lump of mass is moving if it can not be expressed that smoke particles flow and move. It will be expressed. If the flow velocity of smoke particles can be expressed appropriately, it can be visualized to represent a state closer to the physical phenomenon of the substance, and the viewer can intuitively feel the flow velocity of the drawing object such as smoke. It becomes easy.

  In view of the above problems, the present invention can flexibly add fluctuation expression of three-dimensional volume data while suppressing an increase in calculation cost, and also enables expression of particle flow velocity that is easy for a viewer to grasp intuitively. An object of the present invention is to provide a volume data rendering system and a volume data rendering method.

  In order to achieve the above object, a volume data rendering system according to the present invention includes a voxel data set generation module that generates a voxel data set to which a luminance value and opacity are given based on volume data, and a voxel data set generated from the voxel data set. A ray casting module that generates pixel data to be drawn based on a luminance value and opacity of the voxel applied to the sampling, and a drawing module that receives the pixel data and draws volume data. And the ray casting module varies the coordinate position of the voxel for the sampling based on a fluctuation model.

In the above configuration, it is preferable that the fluctuation model is a systematic fluctuation model that reflects fluctuation of a physical phenomenon expressed by the volume data. For example, if the volume data to be drawn is smoke at the time of fire, a systematic fluctuation model that reflects how the smoke rises while fluctuating with smoke is used. In general, a turbulent phenomenon such as smoke is characterized by a statistical quantity. Therefore, it is preferable to prepare a fluctuation volume that reflects the frequency characteristics of the spatial concentration distribution in a phenomenon that actually occurs.
Further, in the above configuration, it is preferable that the fluctuation model also includes a description of fluctuations regarding the voxel color data. Here, the hue data is data related to pixel color representation, and examples thereof include additive color mixing RGB data, subtractive color mixing CMYK data, color development Munsell color system data, HIS color system data, and the like. .

  In addition, a fluctuation model database storing a plurality of fluctuation models is provided, and the ray casting module selects and equips a systematic fluctuation model that reflects fluctuations of a physical phenomenon expressed by the volume data from the fluctuation model database. It is also preferable to do.

  In addition, the ray casting module reflects the movement speed of the physical phenomenon expressed by the volume data with respect to the magnitude and direction of the fluctuation in the change in the coordinate position of the voxel for the sampling based on the fluctuation model. It is also preferable to add adjustments based on field models.

  A velocity field model database storing a plurality of velocity field models, wherein the ray casting module selects a velocity field model reflecting a moving speed of a physical phenomenon expressed by the volume data from the velocity field model database; It is preferable that it can be equipped.

  Further, the fluctuation model is defined in the form of a fluctuation volume in which a fluctuation function is arranged in each voxel, and the velocity field model is defined in the form of a velocity field volume in which a function representing a velocity is arranged in each voxel. An index volume that holds reference information indicating which voxel is referenced in the volume, the fluctuation volume and the velocity field volume are linked with each other via the index volume, and the ray casting module includes the velocity field volume The reference information of the index volume is updated by reflecting the influence on the physical phenomenon by the fluctuation of the volume data for the sampling using the fluctuation function arranged in the voxel of the fluctuation volume indicated by the reference information for the update. The coordinate position of the voxel It is also preferable to moving.

  According to the volume data rendering system of the present invention, since the ray casting module changes the coordinate position of the voxel for sampling based on the fluctuation model, the increase in calculation cost is suppressed in the rendering result of the rendering object such as smoke. It is possible to obtain a volume data rendering system capable of flexibly visualizing fluctuation expression of volume data such as smoke while maintaining naturalness as three-dimensional volume data.

  According to the volume data rendering system of the present invention, when the ray casting module changes the sampled voxel coordinate position based on the fluctuation model, the movement speed of the physical phenomenon expressed by the volume data with respect to the magnitude and direction of fluctuation Because the speed field model that reflects this is added, the smoke flow rate can be appropriately expressed in the rendering result of the rendered object such as smoke that has been visualized, and the viewer can perceive the flow rate of the rendered object such as smoke. It becomes easy.

The volume data rendering system of the present invention is capable of expressing fluctuation and particle velocity while maintaining the naturalness close to the physical phenomenon of the substance while suppressing the increase in calculation cost in the rendering processing of the acquired volume data. It is a thing.
In the present invention, a concept of fluctuation volume describing characteristic fluctuations of a physical phenomenon to be drawn is introduced, and in the sampling process of each voxel, fluctuations are given to the sampling process of the voxel while referring to the fluctuation volume. It is.

  Hereinafter, a configuration example capable of expressing fluctuations as Example 1, a configuration example capable of expressing fluctuations and the flow velocity of particles as Example 2, and a configuration example capable of further application and expansion as Example 3 Show.

1 illustrates a configuration example of a volume data rendering system that enables fluctuation expression according to a first embodiment;
FIG. 1 is a diagram schematically illustrating a basic flow of a rendering process in the volume data rendering system according to the first embodiment of the present invention.

  Also in the volume data rendering system according to the first embodiment of the present invention, first, volume data acquisition processing (S1) and luminance value and opacity determination processing (S2) for each voxel are performed. These processes may be the same as the flow of the volume rendering process according to the prior art shown in FIG.

In the ray casting process of the present invention, a sampling coordinate position determination process (S10) is added to the fluctuation process, and the fluctuation process (S10) is reflected on the voxel sampling process (S3). Yes.
In the conventional voxel sampling process, as shown in FIG. 11, the voxels are sampled at regular intervals along the path of the ray, and the sampled voxels are fixed. The voxel sampling process in the prior art incorporating the concept of the sliced surface model shown in FIG. 12 also has a fixed distance between the surfaces, and eventually samples voxels at regular intervals along the path of the ray. It can be said that it is the same as the thing.

  However, in the volume data rendering system of the present invention, since the so-called spatial fluctuation is given based on the fluctuation model at the sampling position of the voxel, the luminance value and the hue of each sampled voxel are not constant, and fluctuation occurs. It becomes. The final luminance value integration calculation results in fluctuations, and the fluctuations are reflected in the luminance values and hues of the pixels that are the rendered rendering results.

FIG. 2 is a diagram schematically showing the fluctuation of the coordinate position of the voxel to be sampled in the sampling process of the volume data rendering system of the present invention.
Reference numeral 200 denotes a voxel data set of volume data. Voxels are formed as lattice units. 201 schematically represents a sampled voxel. 300 is a ray.

  The top diagram in FIG. 2 shows a state of sampling processing of a voxel data set by ray casting for drawing the first frame. It can be seen that there is a spatial fluctuation with respect to the fixed ray sampling shown in FIG. The coordinate position of the voxel 201 sampled with respect to the ray path fluctuates three-dimensionally. Based on the sampling result, the pixel value of the corresponding pixel in the first frame is determined.

  The middle diagram of FIG. 2 shows a state of sampling processing of a voxel data set by ray casting for drawing the nth frame. It can be seen that the sampled voxel 201 in the middle diagram of FIG. 2 has fluctuations relative to the sampled voxel 201 in the top diagram of FIG. This fluctuation is a fluctuation caused by the elapse of n frame time from the first frame at the top of FIG. 2, that is, a temporal fluctuation.

  As described above, the volume data rendering system of the present invention can weave spatial fluctuations in one frame, and can further incorporate temporal fluctuations between frames in a visualized image displayed as a moving image. If the description of the fluctuation model used in the fluctuation processing (S10) has temporal fluctuation as well as spatial fluctuation, the voxel coordinate position sampled based on the fluctuation model fluctuates between frames. Fluctuations are reflected in the luminance value integrated and calculated from the luminance value and opacity of the voxel, and fluctuations can be given to the rendering result.

  Here, in the above fluctuation processing, if a systematic fluctuation model that reflects fluctuation of a physical phenomenon expressed by volume data to be drawn is introduced, the physical phenomenon of the entity to be drawn is reflected in the rendering result. This makes it possible to express the fluctuations that occur. Detailed explanation of this point will be described later.

As a result of the sampling process in the ray casting process, the luminance value of the sampled voxel is i, the opacity is α, the incident luminance value incident on this voxel is I _in, and the outgoing luminance value when exiting this voxel is I _out. Then, the emission luminance value of the voxel is represented by Expression 2, which is the same mathematical expression as that described as Expression 1 in the related art.
The pixel value to be finally drawn is obtained by integrating the following formula 2 for the voxels sampled through the fluctuation process along the path of the corresponding ray.

Drawing processing of each pixel is executed, and volume data such as smoke is drawn on the monitor.
The integration process (S4) and the drawing process (S5) for each pixel shown in FIG. 1 may be the same as the flow of the volume rendering process according to the conventional technique shown in FIG.
The above is the basic processing flow of the volume rendering system of the present invention.

Next, a system configuration example will be shown.
FIG. 3 is a diagram showing an example in which the volume data rendering system 100 is specifically constructed by processing modules. Each processing module may be constituted by a combination of a general-purpose CPU and a memory system, or a dedicated accelerator such as a DSP may be used.

  Reference numeral 110 denotes a volume data acquisition module. For example, volume data given as simulation data is captured. A configuration in which an external input interface is provided and volume data is captured from the outside via a network may be employed.

  Reference numeral 120 denotes a voxel data set generation module. A luminance value determination module 121 and an opacity determination module 122 are provided. Volume data is received from the volume data acquisition module 110 and expanded for each voxel to obtain a data set of its luminance value and opacity.

  A loop is formed between the volume data acquisition module 110 and the voxel data set generation module 120. The volume data often changes as the simulation progresses. Since a new voxel data set reflecting the update of the volume data must be prepared, it is necessary to acquire new volume data via the volume data acquisition module 110 and generate a voxel data set at each update period of the volume data. There is. Therefore, as shown in the figure, the processing by the volume data acquisition module 110 and the voxel data set generation module 120 forms a loop.

  Reference numeral 130 denotes a ray casting module. The ray casting module 130 includes a sampling module 131, an integration module 132, and a fluctuation processing module 133.

  The sampling module 131 first determines the original sampling coordinates d (x, y, z) according to a predetermined rule. For example, as shown in FIG. 11, one ray corresponding to a certain pixel is selected, and the coordinate values d (x, y, z) of voxels at a predetermined interval are determined along the path through which the ray passes through the volume. The sampling module 131 passes the sampling original coordinates d (x, y, z) to the fluctuation processing module 133.

  The fluctuation processing module 133 has a systematic fluctuation model that reflects fluctuations in physical phenomena expressed by volume data, and considers fluctuations with respect to the sampling original coordinates d (x, y, z) that were at predetermined intervals. Converted to the sampling fluctuation coordinate U (t, d). The sampling original coordinates d (x, y, z) received from the sampling module 131 are subjected to fluctuation processing based on the fluctuation model to generate sampling fluctuation coordinates U (t, d). The fluctuation processing module 133 has a fluctuation model volume in which a fluctuation value is given to each voxel. The fluctuation value is given as a vector with three components x, y and z.

The sampling module 131 receives the sampling fluctuation coordinate U (t, d) from the fluctuation processing module 133, and determines the voxel at the coordinate position as the sampling target.
The ray casting module 130 accesses the voxel data set generation module 120 based on the sampling fluctuation coordinates U (t, d), and acquires the luminance value and opacity of the voxel at each coordinate position. Thereafter, the integration module 132 performs integration based on Equation 2. This integral value is the pixel value of the target pixel.

The drawing module 140 receives the luminance value integrated from the ray casting module 130 and draws the pixel on the monitor according to the luminance value. If all the pixels are drawn by the drawing module 140, one frame in which the volume is visualized is completed.
When the simulation display of the volume data is a moving image, it is necessary to rewrite the frame every predetermined frame period. Therefore, the ray casting module 130 and the drawing module 140 form a loop. For example, when displaying at 30 frames per second, the loop processing by the ray casting module 130 and the drawing module 140 is repeated 30 times per second. In this 30-frame visualized image, fluctuation processing reflecting the fluctuation of the physical phenomenon expressed by the volume data is applied to the sampling coordinate position of the voxel. As a result, smoke and the like fluctuate in a form close to reality. It becomes possible to give such an effect.

FIG. 4 is a diagram schematically showing the concept of fluctuation processing in the present invention.
In FIG. 4, 200 is a volume data set. 210 schematically shows the fluctuation volume. The fluctuation volume 210 is a kind of mapped vector space in which a fluctuation function is arranged for each voxel in a three-dimensional space based on a fluctuation model. Given a coordinate d (x, y, z), converts it to a fluctuation function U (t, d) placed in the voxel at that coordinate position.

In the case of the prior art, as indicated by the dotted arrow, the coordinates d (x, y, z) of the volume data are directly assigned to the volume data as a clue to the coordinates d (x, y, z). Get the brightness value and opacity of the current voxel.
However, in the volume data rendering system of the present invention, the volume data is accessed after coordinate conversion is performed on the fluctuation volume. As indicated by the solid arrows, first, a certain coordinate d (x, y, z) is given to the fluctuation volume, and the fluctuation function U (t , d). Based on the converted fluctuation function U (t, d), the fluctuation function U (t1, d) at the time t1 corresponding to the frame to be rendered is specified, the volume data is accessed, and the coordinate position d + Get brightness value and opacity of voxel placed in U (t1, d).

Next, an example of the fluctuation volume 210 is shown.
The general form of the fluctuation volume is represented by the following map.
U: (x, y, z ) ∈R 3 → R 3
Moreover, it is desirable to satisfy the following properties (1) and (2).
(1) The average is 0
(2) Having a characteristic length (the frequency component obtained by Fourier transform has a characteristic peak)

  Note that it is preferable to use a systematic fluctuation model that reflects the naturalness close to the fluctuation of the physical phenomenon expressed by the volume data to be drawn as the fluctuation processing model to be added in the sampling coordinate position determination process. For example, if the volume to be drawn is smoke, adopting a fluctuation model that models the characteristic fluctuation pattern seen when smoke rises, the fluctuation of the volume data of the visualized smoke will be the fluctuation of the actual smoke The effect of getting close to condition is obtained.

  The smoke fluctuation volume u (t, d) is generated by the following procedure. First, all fluctuations are initialized to 0. Next, a position r (x, y, z) in the volume is selected at random, and a change of the following formula 3 that is normally distributed is added to each component of fluctuation held by each voxel d.

Where l is the characteristic size of the pattern formed by smoke. It is possible to give further fluctuations to k and l. By repeatedly adding the perturbation to the fluctuation volume according to the above equation, a fluctuation with strong short-range autocorrelation is generated.

  Next, the present invention introduces fluctuations in the determination of the coordinate position of the voxel to be sampled as described above, but it is also possible to introduce fluctuations directly to the luminance value and hue data of the sampled voxel. Is possible. Even if the change in luminance value or hue between adjacent voxels in the volume data set is small, it can be visualized to viewers by making adjustments that directly introduce fluctuations in the luminance value and hue data It is possible to ensure an easy-to-understand visual effect about the fluctuation of volume data such as smoke.

The following can be assumed as an example of the fluctuation volume that gives fluctuations to the brightness values and hue data of the sampled voxels.
For example, in the case of additive-mixed RGB data,
U: (R, G, B) ∈R ³ → R ³
For example, in the case of HSI data,
U: (H, S, I) ∈R ³ → R ³

  The rendering system of the present invention can perform not only drawing of a stationary object such as a cloud but also rendering of volume data that dynamically changes such as a fluid such as smoke. Therefore, it is necessary to periodically update the access of the volume data via the fluctuation volume shown in FIG.

  The first update process is an update process performed every frame period. For example, if display processing of 30 frames per second is performed, sampling processing and integration processing of the voxel at the coordinate position indicated by the fluctuation volume 210 are executed every 1/30 seconds to execute drawing processing of each pixel of the next frame. . In this way, the access from the fluctuation volume 210 to the volume data 200 is executed every 1/30 seconds, and the drawing process is performed while expressing the subtle fluctuations of smoke particles taking into account the fluctuation movement of the actual smoke. In consideration of the quality of drawing, for example, frames can be dropped and the update process can be performed every 1/15 seconds, or the update process can be performed every 1/10.

  The second update process is an update process performed for each update cycle of the volume data 200. When a three-dimensional object to be drawn moves to a macro or deforms, the volume data 200 itself is updated at an appropriate timing and replaced with new volume data 200. For example, in the case of expressing macro movement or deformation such as smoke moving along a wall surface or being swept away by wind, the volume data 200 itself is updated. For example, it is updated at an appropriate timing according to the macro change speed of the three-dimensional object such as every 0.5 seconds and after one second.

The volume data rendering system according to the second embodiment of the present invention is a configuration example in which the flow velocity of particles can be expressed in addition to the fluctuation expression of particles in the volume data according to the first embodiment.
In a fire in a tunnel or underground mall, the smoke not only rises naturally, but the flow of air in the tunnel or underground mall greatly affects the movement of the smoke. Here, since the macro movement of the smoke riding on the air flow is originally described in the simulation data, the rendering system may basically draw in accordance with the simulation. However, objects such as smoke are not described as individual particle movements in simulation data, but are often given as a block of volume data. For this reason, unless an appropriate fluctuation expression is performed on the rendering system side, it is expressed as a relatively lumpy movement. In addition, it was often difficult for viewers to feel the smoke flow rate.

Therefore, in the second embodiment, in addition to the fluctuation expression that reflects the characteristics of the physical phenomenon of smoke introduced in the first embodiment, the fluctuation expression that reflects the influence of the outside world such as wind is further introduced to simulate the viewer. This is a visualization that makes it easy to feel the smoke flow rate.
FIG. 5 is a diagram schematically illustrating a basic flow of a rendering process in the volume data rendering system according to the second embodiment of the present invention.

  In the volume data rendering system according to the second embodiment of the present invention, as in the volume data rendering system according to the first embodiment, fluctuation processing is added to the ray casting process. However, the fluctuation processing (S10a) according to the second embodiment is performed using particles. Fluctuation expression that incorporates the flow velocity expression, and a visual expression that makes it easier for viewers to feel the flow velocity of particles.

  Other volume data acquisition processing (S1), luminance value and opacity determination processing (S2) for each voxel, sampling processing (S3), integration processing (S4), and drawing processing (S5) are the same as in the first embodiment. Good.

  The volume data rendering system according to the second embodiment introduces the concept of velocity field volume in order to allow the expression of particle flow velocity in addition to the expression of particle fluctuation in volume data. The velocity field volume is a kind of mapping vector space in which a function representing velocity is arranged in each voxel in the three-dimensional space based on the velocity field model. Given a coordinate d (x, y, z), it converts it to a velocity v (t, d) placed in the voxel at that coordinate position.

In the volume data rendering system of the second embodiment, the ray casting module performs fluctuation processing in the fluctuation processing module after reflecting the influence of the surrounding environment such as wind through the velocity field processing module. .
The volume data rendering system according to the second embodiment needs to perform calculation of functions arranged in the voxels of these two three-dimensional vector spaces while maintaining the velocity field volume and the fluctuation volume. Therefore, an index volume having voxels in the same three-dimensional vector space and holding reference points is introduced.

FIG. 6 is a diagram schematically illustrating the concept of fluctuation processing reflecting the flow velocity expression of the second embodiment.
In FIG. 6, the volume data set 200 and the fluctuation volume 210 are the same as those in FIG.

The index volume 220 is a volume that holds the reference point of the fluctuation model volume 210. The index volume 220 has voxels arranged in the same three-dimensional space as the volume data 200, the fluctuation volume 210, and the velocity field volume 230. The index volume 220 preferably contains a continuous value in each voxel at the time of initialization. For example, i _j (r) = r.

The velocity field volume 230 includes a model describing the influence of wind or the like, and each voxel has three-dimensional vector data. For example, it is expressed as v (r).
Since the fluctuation model volume 210 is referred to via the index volume 220, changes that move from moment to moment due to the influence of the velocity field can be reflected and held in the index volume. For example, the reference point is moved as shown in Equation 4 below.

In the case of the prior art, as indicated by the dotted arrow, the coordinates d (x, y, z) of the volume data are directly assigned to the volume data as a clue to the coordinates d (x, y, z). Get the brightness value and opacity of the current voxel.
However, in the volume data rendering system of the second embodiment, the volume data is accessed after reflecting the flow velocity expression by the velocity field volume 230 and the coordinate transformation by the fluctuation model volume 210 as indicated by the solid arrow.

  First, a certain coordinate d (x, y, z) is given to the index volume 220, and the reference coordinate i (t, d) of the voxel arranged at the coordinate d (x, y, z) in the index volume 220 is set. Convert. Next, the reference coordinate i (t, d) is given to the fluctuation volume 210, and the fluctuation function u (t, i (t, d) possessed by the voxel arranged at the reference coordinate i (t, d) in the fluctuation volume 210 is given. )). Based on the converted fluctuation function u (t, i (t, d)), after specifying the fluctuation function u (t1, i (t1, d)) at time t1 corresponding to the frame to be rendered, The volume data is accessed, and the brightness value and opacity of the voxel arranged at the coordinate position d + u (t1, i (t1, d)) are obtained.

  As described above, since the reference coordinates of the index volume 220 move in a manner that reflects the influence of wind or the like, the speed of change of the luminance value and opacity also changes according to the speed of movement of the reference coordinates. In addition, rendering that visualizes the flow velocity of particles such as smoke becomes possible.

  FIG. 7 is a diagram showing an example in which the volume data rendering system 100 is specifically constructed by processing modules. Each processing module may be constituted by a combination of a general-purpose CPU and a memory system, or a dedicated accelerator such as a DSP may be used.

  In the second embodiment, a velocity field processing module 134 is added to the ray casting module 130 in addition to the sampling module 131, the integration module 132, and the fluctuation processing module 133.

  In the case of FIG. 3 in the first embodiment, the sampling module 131 determines the sampling original coordinates d (x, y, z) according to a predetermined rule, and the sampling original coordinates d (x, y, z) is input to the fluctuation processing module 133. In the case of the second embodiment, the sampling module 131 determines the original sampling coordinates d (x, y, z) according to a predetermined rule, and first passes it to the velocity field processing module 134.

  The velocity field processing module 134 has a model describing the influence of the surrounding environment such as wind, and the smoke is generated based on the velocity field model with respect to the sampling original coordinates d (x, y, z) received from the sampling module 131. The reference coordinate i (t, d) of the voxel at the corresponding coordinate position of the index volume is generated by adding a shift process reflecting the flow velocity. That is, the sampling original coordinates d (x, y, z) are converted into the reference coordinates i (t, d) in consideration of the movement of smoke particles due to the wind.

  The fluctuation processing module 133 receives the reference coordinate i (t, d) from the velocity field processing module 134, applies fluctuation processing to the reference coordinate i (t, d) based on the fluctuation model, and obtains the sampling fluctuation coordinate u (t , i (t, d)). That is, the fluctuation processing module 133 converts the received reference coordinates i (t, d) into sampling fluctuation coordinates u (t, i (t, d)) in consideration of fluctuations. The sampling module 131 receives the sampling fluctuation coordinate u (t, i (t, d)) from the fluctuation processing module 133, and determines the voxel at the coordinate position as the sampling target.

As described above, in the second embodiment, the ray casting module 130 performs the fluctuation processing in the fluctuation processing module 133 after reflecting the influence of the surrounding environment such as wind via the velocity field processing module 134. Yes.
The ray casting module 130 accesses the voxel data set generation module 120 based on the sampling fluctuation coordinates u (t, i (t, d)), and acquires the luminance value and opacity of the voxel at each coordinate position. Thereafter, the integration module 132 performs integration based on the above-described equation 2. This integral value is the pixel value of the target pixel. The drawing module 140 receives the luminance value integrated from the ray casting module 130 and draws the pixel on the monitor according to the luminance value. If all the pixels are drawn by the drawing module 140, one frame in which the volume is visualized is completed.

  As described above, the volume data rendering system according to the second embodiment can incorporate spatial fluctuations in one frame, and further incorporates a flow velocity expression that reflects the influence of wind or the like in a visualized image displayed as a moving image. Time fluctuations between frames can be incorporated. By performing the fluctuation process (S10a) incorporating the flow velocity expression in this way, the sampled voxel coordinate position fluctuates between frames, and it is possible to give the fluctuation that incorporates the flow velocity expression in the rendering result.

Next, a device for periodically initializing the index volume will be described.
The influence of the outside world on volume data such as wind is complicated and not uniform throughout the space. If the velocity field model has different velocities depending on the location (coordinate position), it is located close to the sampling original coordinates. When the two points that were strongly correlated with each other and initially referred to nearby voxels in the fluctuation volume had different velocities in the velocity field model, the reference points in the index volume moved away from each other over time. The correlation between the voxel fluctuation coordinates referenced in the fluctuation volume is lost, and the mutual fluctuation is simply random, reflecting the fluctuation of the physical phenomenon expressed by the volume data to be drawn. It is not a systematic fluctuation. In the volume data rendering system according to the second embodiment of the present invention, it is sufficient that the flow velocity of particles due to the influence of wind or the like can be easily visualized by the viewer, so that the shift amount of the reference coordinates in the index volume is reflected based on the velocity field model. It is only necessary to initialize the reference coordinates. Therefore, it is possible to devise a method for periodically initializing the index volume so that each voxel contains a continuous value. If the index volume is periodically initialized in this way, the correlation between data that are located close to each other in the original sampling coordinates and have strong correlation with each other can be maintained appropriately.

  The volume data rendering system according to the third embodiment of the present invention is a configuration example in which the fluctuation model of the fluctuation processing module 133 and the velocity field model of the velocity field processing module 134 shown in the first and second embodiments can be flexibly exchanged. It is.

Although the drawing target of the volume data rendering system according to the third embodiment of the present invention is not limited to smoke, the drawing target will be described as smoke as an example below.
In a fire in a tunnel or underground mall, the smoke not only rises naturally, but the flow of air in the tunnel or underground mall greatly affects the movement of the smoke. Furthermore, the flow of air may change greatly, such as the opening and closing of emergency exits in tunnels, the opening and closing of fire doors, fluctuations in blowing capacity from the blower openings, and movement of objects such as arrival and departure of trains in subway stations. When the air flow changes greatly, it is necessary to change the velocity field model describing the air flow. In the third embodiment, the fluctuation model of the fluctuation processing module 133 and the speed field model of the speed field processing module 134 can be flexibly exchanged.

  The smoke fluctuation may vary depending on the material being ignited, the oxygen concentration in the space, the temperature, the scale of the fire, and the like. Since the conditions that affect the smoke fluctuation condition vary depending on the progress of the fire, the scale of the fire, and the like, it is preferable to dynamically adopt a fluctuation model that appropriately represents the smoke fluctuation condition.

  FIG. 8 is a diagram illustrating an example in which the volume data rendering system 100 according to the third embodiment is constructed by the processing module. Each processing module may be constituted by a combination of a general-purpose CPU and a memory system, or a dedicated accelerator such as a DSP may be used.

  In the third embodiment, the ray casting module 130 includes a fluctuation model database 135 and a velocity field model database 136 in addition to the sampling module 131, the integration module 132, the fluctuation processing module 133, and the velocity field processing module 134. It has become.

The fluctuation model database 135 stores a plurality of fluctuation models according to the material being ignited, the oxygen concentration in the space, the temperature, the scale of the fire, and the like.
The speed field model database 136 stores a plurality of speed field models corresponding to the opening / closing of the emergency exit, the opening / closing of the fire door, the fluctuation of the blowing capacity from the blower opening, and the like.

  The fluctuation processing module 133 detects information such as the material that is ignited, the oxygen concentration in the space, the temperature, and the scale of the fire by itself or receives notification from the outside, and the fluctuation model corresponding to the information is received from the fluctuation model database 135. Take out and equip dynamically.

  The speed field processing module 134 detects information such as the opening / closing of the emergency exit, the opening / closing of the fire door, and the change in the air blowing capacity from the air outlet, or receives a notification from the outside, and the speed field model corresponding to the information is sent to the speed field model. Retrieve from database 136 and dynamically equip.

  As described above, the volume data rendering system according to the third embodiment of the present invention can dynamically replace the fluctuation model and the velocity field model according to various condition fluctuations, and continues the rendering process expressing the natural fluctuation. can go.

  The volume data rendering system of the present invention can be constructed using various computers by providing it as a program describing processing steps for realizing the configuration described above. The program including the processing steps for realizing the volume data rendering system of the present invention is not limited to a portable recording medium such as a CD-ROM, DVD, or flexible disk, as shown in the configuration example shown in FIG. The program may be downloaded from a program server on the network. When the program is executed, the program is loaded on the computer and executed on the main memory.

  Further, it may be configured not only to compile a source program but also to send an intermediate language format applet to a client computer via a so-called network and execute the interpreter on the client computer.

  The present invention can be applied to a volume data rendering system and volume data rendering processing method for visualizing volume data defined in a three-dimensional space.

FIG. 5 is a diagram schematically illustrating a basic flow of a rendering process in the volume data rendering system according to the first embodiment of the invention. The figure which showed typically the fluctuation of the coordinate position of the voxel to sample in the sampling process of the volume data rendering system of this invention The figure which shows the example which built the system with processing module The figure which showed typically the concept of the fluctuation process in this invention The figure which showed typically the basic flow of the rendering process in the rendering system of the volume data concerning Example 2 of this invention. The figure which showed typically the concept of the fluctuation process which reflected the flow-velocity expression of Example 2. FIG. The figure which shows the example which built the system concretely with the processing module The figure which shows the example which constructed | assembled the volume data rendering system concerning Example 3 by the processing module. The figure which shows typically a mode that it provides as a program which described the processing step which implement | achieves the volume data rendering system of this invention. The figure which showed the basic flow of the rendering processing procedure when using volume data Diagram explaining the concept of ray casting processing The figure which shows the concept which introduces the expression form of the sliced surface model in the expression form of the volume data

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 Volume data rendering system 110 Volume data acquisition module 120 Voxel data set production module 121 Luminance value determination module 122 Opacity determination module 130 Ray casting module 131 Sampling module 132 Integration module 133 Fluctuation processing module 134 Speed field processing module 135 Fluctuation model database 136 Velocity field model database 140 Drawing module 200 Voxel data set of volume data 201 Voxel 202 Ray 210 Fluctuation volume 220 Index volume 230 Velocity field volume

Claims (8)

A voxel data set generation module that generates a voxel data set given a luminance value and opacity based on volume data;
A ray casting module that samples voxels along a ray from the voxel data set, and generates pixel data to be drawn based on a luminance value and opacity of the voxel involved in the sampling;
A drawing module that receives the pixel data and draws volume data;
The volume data rendering system, wherein the ray casting module changes a coordinate position of a voxel for the sampling based on a fluctuation model. The volume data rendering system according to claim 1, wherein the fluctuation model is a systematic fluctuation model reflecting fluctuation of a physical phenomenon represented by the volume data. The volume data rendering system according to claim 1 or 2, wherein the fluctuation model also includes a description of fluctuations regarding voxel hue data. Equipped with a fluctuation model database that stores multiple fluctuation models
The volume data rendering system according to claim 1 or 3, wherein the ray casting module selects and equips a systematic fluctuation model that reflects fluctuations of a physical phenomenon expressed by the volume data from the fluctuation model database. In the fluctuation of the coordinate position of the voxel for the sampling based on the fluctuation model, an adjustment by a velocity field model reflecting the movement speed of a physical phenomenon expressed by the volume data is added to the fluctuation size and the flowing direction. Item 4. The volume data rendering system according to any one of Items 1 to 3. It has a speed field model database that stores multiple speed field models,
The volume data rendering system according to claim 4, wherein the ray casting module selects and equips a speed field model reflecting a moving speed of a physical phenomenon expressed by the volume data from the speed field model database. The fluctuation model is defined in the form of a fluctuation volume in which a fluctuation function is arranged in each voxel, and the velocity field model is defined in the form of a velocity field volume in which a function representing a velocity is arranged in each voxel,
An index volume that holds reference information indicating which voxel is referenced in the fluctuation volume;
Linking the fluctuation volume and the velocity field volume via the index volume,
The ray casting module updates the reference information of the index volume by reflecting the influence of the velocity field volume on the physical phenomenon, and the fluctuation function arranged in the voxel of the fluctuation volume indicated by the reference information related to the update The volume data rendering system according to claim 5 or 6, wherein the coordinate position of the voxel of the volume data related to the sampling is changed by using. On the computer,
A voxel data set generation processing step for generating a voxel data set given a luminance value and opacity based on volume data;
A ray casting processing step of sampling voxels along a ray from the voxel data set, and generating pixel data to be drawn based on a luminance value and opacity of the voxel involved in the sampling;
A drawing processing step for receiving the pixel data and drawing the volume data ;
A volume data rendering processing program characterized in that, in the ray casting processing step, a coordinate position of a voxel for sampling is changed based on a fluctuation model.

Images