JP2003263651A - Volume rendering method and its program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a volume rendering method and its program for executing volume rendering by using a normal PC or game machine, and for dealing the relatively large volume data at an interactive speed, and reducing the deterioration of image quality to the minimum. <P>SOLUTION: This volume rendering method comprises a basic slice data preparing step (A) for expressing whole volume data with the accumulation of the slide data by using a texture map function and α-blending function by using a device having graphics hardware having a texture map function, α- blending function, and multi-texture mechanism and an intermediate slice data preparing step (B) for linearly interpolating and inserting the intermediate slice data between the basic slide data. Also, the whole volume data are expressed with blocks whose resolutions are fixed without being hierarchically expressed so that the data structure can be simplified, and the data quantity is reduced by using the locality of data in each block so that the memory can be efficiently used. <P>COPYRIGHT: (C)2003,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a volume rendering method and program for performing volume rendering using a normal PC (personal computer) or game machine.

[0002]

2. Description of the Related Art Volume rendering is image display means for directly displaying volume data, which is space density data. In volume rendering, the line of sight is extended and the density of voxels on the line of sight is summed. Such a method is called "ray casting".
Therefore, in volume rendering, an image considering transparency is generated based on the final numerical value.

In other words, the volume rendering is a visualization method for displaying a three-dimensional scalar field on a two-dimensional screen, and the method can be roughly classified into an indirect method and a direct method. Indirect volume
In rendering, information to be displayed is extracted by preprocessing and a part of data is displayed. A typical example is isosurface display using the Marching Cubes method. Direct method (direct volume re)
is a method of calculating the contributions of all the sample points and displaying the whole, and integrating the luminance along the line-of-sight direction from the drawing surface (backward projecti).
on) Ray Casting, Cell Projection that projects sample points on the drawing surface (Forward projection), Shear-Wa
rp Factorization, Splattin
g etc. are known. Usually, the term volume rendering refers to the direct method, and the present invention also handles the direct method.

[0004]

The use of volume rendering makes it possible to obtain an image with high image quality and easy to grasp the overall state of data.
Widely used in entertainment and other fields. However, there is a problem that the time required for calculation is long and the required memory capacity is large. Therefore, conventionally, the use has been limited when the speed does not matter, or when a large-scale computer environment or special high-speed hardware can be used.

That is, the conventional volume rendering has the following problems. (1) The calculation cost is high. That is, unless a large-scale computer environment with a large storage capacity or special high-speed hardware can be used, it is not possible to perform drawing at an interactive speed. (2) When performing volume rendering on a normal PC (personal computer) or game machine, the volume of volume data that can be handled is small. For example, 64x
If the volume data is as small as 64 × 64, then 3
Conventionally, volume rendering can be performed at an interactive speed of 0 fps (30 frames per second) (a comfortable speed with almost no waiting time).
With volume data of about 56 × 256 × 256 (16 Mbytes), for example, it drops to 20 fps or less, and comfortable use cannot be performed. Furthermore, 512 × 512 × 512 (1
1 fp for volume data of about 28 Mbytes
It becomes less than s and becomes practically impossible. (3) When processing such as data thinning is performed in order to increase the drawing speed in a normal PC or game machine, moire and the like become severe and the image quality is conspicuously deteriorated.

The present invention was devised to solve these problems. That is, the object of the present invention is to
(1) Volume rendering can be performed using a normal PC (personal computer) or game console,
(2) To propose a volume rendering method and its program capable of handling relatively large volume data of about 128 Mbytes or more at an interactive speed, and (3) further minimizing deterioration of image quality.

[0007]

According to the present invention, a texture map function, an α-blending function and a mul are provided.
A basic slice data creation step (A) in which the entire volume data is expressed by stacking slice data using a texture map function and an α-blending function using a device having graphics hardware having a ti-texture mechanism; And a step (B) of creating intermediate slice data in which the intermediate slice data is linearly interpolated and inserted between the data, and a volume rendering method is provided. The graphics hardware is, for example, a PC graphics card, but may be graphics hardware with a similar function installed in a workstation. Further, the device is, for example, a personal computer, but may be a game machine or other device having the same function.

According to the above-mentioned method of the present invention, the entire volume data is expressed by stacking slice data using the texture map function and the α-blending function, so that the polygon display function of the graphics hardware (for example, PC graphics card) is displayed. Can be used to realize high-speed volume rendering. Also, by inserting the intermediate slice data by linearly interpolating between the basic slice data, the lack of interpolation accuracy in the direction perpendicular to the slices is compensated for, and the image quality failure due to the mismatch of interpolation accuracy (for example, moire) Can be prevented.

According to a preferred embodiment of the present invention, the insertion interval of the intermediate slice data is set to be equal to or narrower than the sampling interval on the slice data according to the resolution required for drawing. By this method, moire can be effectively prevented.

The texture of the intermediate slice data is the multi-text of the graphics hardware.
Created at the time of drawing using the true mechanism. Since the texture of the intermediate slice data is not held in the memory by this method, the sample points can be increased in the direction perpendicular to the slice without increasing the required memory capacity.

Further, according to the present invention, the entire volume data is represented by blocks of fixed resolution instead of hierarchical representation to simplify the data structure, and the locality of the data in each block is utilized to reduce the data amount. The present invention provides a volume rendering method characterized by efficiently utilizing memory.

According to the above method of the present invention, it is possible to reduce the size of input data and draw a volume exceeding the capacity of hardware at high speed while suppressing deterioration of image quality.

According to a preferred embodiment of the present invention, a 2 ^3N block division step (C) is performed to divide the entire volume V into blocks V ^B having a fixed size of 2 ^3N , and the resolution is set to 1 in order from N for each divided block. Of the block error calculation step (D) for calculating the error with the original data that occurs at that time and the calculation error of all the blocks.
A resolution reduction step (E) for reducing the resolution of the block having the smallest error, and steps (D) and (E) are repeated to determine the optimum resolution for each block while reducing the data size as a whole. To do. By this method, even if the size of the same volume as a whole,
It is possible to maintain the high image quality of the drawing result.

Further, according to the present invention, an intermediate step is created between the basic slice data creating step (A) in which the entire volume data is expressed by stacking slice data by using the texture map function and the α-blending function, and the basic slice data. Intermediate slice data creation step (B) for linearly interpolating slice data and inserting, and 2 ^3N block division step (C) for dividing the entire volume V into blocks V ^B having a fixed size 2 ^3N , and for each divided block The block error calculation step (D) in which the resolution is sequentially decreased from N by 1 and the error with the original data generated at that time is calculated, and the resolution of the block with the smallest error among the calculation errors of all blocks is A resolution reduction step (E) of decreasing, and repeating steps (D) and (E) to obtain data as a whole. While reducing size, to determine the optimum resolution for each block, volume rendering program is provided, characterized in that.

This program has a texture map function, an α-blending function, and a multi-text function.
By executing on a device equipped with graphics hardware having a ure mechanism, high-speed volume rendering can be realized by utilizing the polygon display function of the graphics hardware, and image quality problems due to mismatch in interpolation accuracy (for example, moire) It is possible to reduce the size of the volume even with the same error as a whole, reducing the size of the input data,
A volume that exceeds the capacity of the hardware can be drawn at high speed.

[0016]

BEST MODE FOR CARRYING OUT THE INVENTION Preferred embodiments of the present invention will be described below with reference to the drawings.

1. Volume rendering is widely used in fields such as engineering, medicine, and entertainment because it makes it possible to obtain an image with high image quality and an easy understanding of the overall state of data. However, since the time required for calculation and the storage capacity are large, its use is limited when speed is not an issue, or when a large-scale computer environment or special hardware can be used. The main reasons for the high calculation cost are 1) the memory locality of the data required for the cumulative calculation of the luminance along the line of sight is low and the memory access efficiency is low, and 2) the input data is resampled. 3 needed
The points are that the linear interpolation is complicated and 3) In many cases, the amount of input data is huge. Earl aborts any unnecessary accumulations in the middle to improve speed
y ray termination optimization, oct
Improvements have been made to the algorithm, such as the method of utilizing local properties of data using ree and kd trees. However, even if these optimizations are performed, the amount of calculation is still large, so it was necessary to use a parallel computer or special dedicated hardware to draw at an interactive speed.

On the other hand, since the drawing speed and function of the PC graphics card have improved, various researches have been actively conducted to realize various visualization methods on a normal PC. Regarding volume rendering as well, a method has been proposed in which a volume is represented by superimposing slices perpendicular to the coordinate axes and is drawn using α blending of texture-mapped polygons.

Here, the texture map or "texture mapping" is to paste an image called a texture on the surface of an object, which is one of the most basic techniques for making a 3D object look realistic. A 3D object (three-dimensional object) drawn by a computer is composed of triangular plates called "polygons", and the finer the size of the object, the more detailed the object can be expressed, so the shape becomes more realistic. The texture map is necessary because the surface pattern and texture of the object cannot be represented by this alone.

Further, α blending or “alpha blending” refers to a translucent treatment. When another image is overlaid semi-transparently on one image, the transparency value is called "Alpha value", and the image is composed (blended) based on this value. ing. In 3D processing, from simple glass plates to irregular shapes such as water, clouds and smoke,
Used for various translucent treatments.

The above-described method using slices has an advantage of being high-speed because hardware acceleration can be used for drawing, while image quality is improved because interpolation / sampling of data in a direction perpendicular to the slices is not performed. Bad again p
Due to the limited storage capacity of the C graphics card, the data size that can be handled is small.

In the present invention, based on the volume rendering based on the superposition of slices, the image quality and the data size that can be handled are improved in two points. Regarding the first point, the multi-texture function of the graphics hardware (for example, PC graphics card) is used to increase the sample points in the direction perpendicular to the slice without increasing the memory capacity necessary for drawing, and the resulting image quality is improved. Improved. Regarding the second point, adaptive sampling of a volume suitable for high-speed drawing is performed to reduce the size of input data, and it is possible to draw a volume that exceeds the capacity of hardware at high speed. In addition, the above method
It was confirmed that it could be implemented on a C graphics card and that it could draw a sufficiently large volume data with sufficient image quality and interactive speed.

Multi-texture function or "multi-texture"
And as its name implies, multiple textures (Direct
X7 is a function to combine (composite) up to 8 sheets. When the function of the video chip is generally called "multi-texture compatible", the above-mentioned overlay processing is collectively performed at the time of one polygon drawing, and "single-pass multiple texture blending (SinglePass
Multiple Texture Blendi
ng) ”. On the other hand, the method realized by drawing polygons with different textures pasted multiple times with software
Blending (MultiPass Texture)
Blen ding). Basically, the drawing results are the same with both methods, but the former can process faster.

2. Direct Volume Re
Nendering 2.1 Rendering Equation FIG. 1 is a schematic diagram of a conventional direct method volume rendering. Volume rendering is a method of accumulating luminance from the drawing surface in the direction of the line of sight (backwardproj).
section) and the method of projecting sample points on the drawing surface (forward projection), but in both cases, the drawing equation is common, and the brightness value of the volume along the ray extending from each pixel on the drawing surface in the line-of-sight direction. Is integrated. Assuming that the distance from the drawing surface along the light ray is λ and the position in the volume is x (λ), the pixel value I on the drawing surface is obtained by the equation (1). Here, c to are functions that represent colors (RGB values) emitted at each point, and τ is a function that represents the attenuation rate of the light ray at each point. With these,
Scalar value s (x) given as volume data
Assign a display color to. Normally, τ is defined as a function of s (x), but the calculation of tildec is performed in the following two stages.

[0025]

[Equation 1]

First, from the scalar value s (x) to the color (R
The color contribution by the scalar value of each point is determined by passing a conversion function to (GB value) (classificativity).
on). This corresponds to the effect of the light emitted from each point of the volume. This conversion function is usually defined by the user and may change during rendering, but it is constant regardless of the viewpoint. Next, color information given from the positional relationship between each point and the light source is added (shadin
g). This includes components of diffuse reflection and specular reflection of light. Effect of classification and shadi
The effects of ng can be calculated individually and combined by addition. When the distribution of the input scalar value is visualized as it is, it can be drawn using only classification, but when displaying the geometrical shape of the object, sh is necessary to accurately recognize the three-dimensional space.
I need ading. Shad for volume
The specific method of performing ing is 5. Described in.

2.2 Discretization When the portion of the light attenuation factor of the equation (1) is discretized in order to perform drawing calculation, the equations (2) to (4) can be written. However, D is the sampling width, i is the number of the sampling point, and is set as equation (5). α _i is a value called opacity (α value).

[0028]

[Equation 2]

Using α _i defined in equation (5), c ~
= Αc _i Where c _i is an RG that does not consider transparency
B value, αc _i is capacity-weighted
color or associated color
It is called the color when considering the opacity. When the color contribution from the discretized i-th minute region area is written as equation (6), the result of discretizing equation (1) using equation (5) is equation (7). Equation (8) is obtained by modifying this equation using the cumulative luminance C _i 'from the infinite distance of the ray to i. The final desired value is I = C ₀ '.

[0030]

[Equation 3]

2.3 Slice-based Vol
The ume Rendering equation (8) shows that volume rendering can be realized by sampling the volume at regular intervals along a ray extending in the line-of-sight direction and repeating α blending of RGB values in order from the farthest from the drawing surface. ing. Therefore, it is considered that the entire volume is represented by stacking a plurality of slices that are as vertical as possible in the line-of-sight direction, and the volume rendering is performed by α blending calculation of texture-mapped polygons (FIG. 2). By expressing in this way, high-speed volume rendering can be realized by using the polygon display function of the graphics hardware (for example, PC graphics card). If the hardware supports solid textures (3D textures), the plane perpendicular to the line of sight (image
e-aligned slice: Fig. 3B),
It is possible to use a method that has a volume as a solid texture. Since the line of sight and the plane are always vertical, it is necessary to recalculate the position of the slice and the texture coordinates each time the viewpoint is changed.

When the solid texture cannot be used, a two-dimensional texture map is used. A slice obtained by cutting a volume along the coordinate axis (object-al
generated slices: FIG. 3A) is generated to map the cross-sectional image of the volume as a two-dimensional texture. At the time of display, the relationship between the line-of-sight direction and the coordinate axis is checked, and the direction of the slicing axis is selected so that the angle between the line-of-sight and the normal line of the slice is always less than a certain value. It is necessary to switch the texture when changing the slice axis, and it is necessary to prepare three sets of slices in the x-, y-, and z-axis directions in order to guarantee the interactivity. In this case, each texture image must occupy a contiguous region in memory due to hardware limitations, so the texture must have a total of three sets of duplicates in each axial direction. Two-dimensional textures are faster than solid textures on most hardware, so object-align
The method of switching and displaying ed slices has a high drawing speed.

3. Implementation of v
Whereas in normal volume rendering, input scalar values are sampled while performing cubic linear interpolation, in rendering using slices, sampling is performed while performing quadratic linear interpolation on slices. Although it is possible to sample at any resolution in the slice direction, it is possible to sample only at regular intervals in the slice normal direction, so if the drawing resolution or viewpoint is changed, the interpolation accuracy will not match, resulting in a poor image quality. May appear. The striped pattern seen in FIG. 4A is a moiré caused by insufficient interpolation accuracy in the direction perpendicular to the slice during rendering using the slice.

One of the methods of approximating the cubic interpolation to match the sampling width is to insert an intermediate slice obtained by linearly interpolating the upper and lower slices between the original slices (FIG. 5). In order to prevent moire, it is necessary to make the interval of generation of intermediate slices approximately the same as the sampling interval on slices during drawing. The texture of the intermediate slice is not stored in the memory but is multi-t
It can be created at the time of drawing using the extract mechanism. The intermediate slice S _{i +} α generated at the position of i + α between the adjacent slices S _i and S _{i + 1} can be defined as equation (9). The position of the intermediate slice is determined by the interpolation of the coordinates, and the newly generated polygon is added to the graphics hardware multi-
Map the texture generated by the texture function.

[0035]

[Equation 4]

In order to prevent moire, the sampling width and the slice width on the polygon at the time of drawing the polygon may be about the same. Therefore, when drawing, the area occupied by one slice on the screen is examined and the number of intermediate slices generated is changed according to the size of the area to eliminate moire (FIG. 4B). When the number of generated intermediate slices is changed, D in equation (5) changes, so it is necessary to update the α value at each point.

4. Adaptable sample
ed 2 ^3N -block With texture maps and alpha blending, graphics hardware (eg PC graphics card)
However, the size of the video memory (VRAM) that holds textures is limited, so the speed drops dramatically when drawing huge data that exceeds the memory capacity. VR in general
AM is about 10 to 100 times smaller than main memory,
Size limitations are a big issue.

For volume data for which data locality can be assumed, the use efficiency of the memory can be expected to be improved by using the octree expression. But octr
If the hierarchical structure of ee becomes multi-tiered and becomes complicated, local continuity of data is deteriorated, so that time required for data access increases. Therefore, when processing speed is required, oc
Using tree is not always effective. An octree (octr) tree is used as an expression method that does not reduce the processing speed at the same time while ensuring the memory efficiency superiority of the multi-resolution expression.
There is a method of using a 2 ^3N branch tree (2 ^3N -tree) which is a generalization of ee). In this method, the entire scene to be processed is huge and the level of detail control (lev
els of detail) and truncation of data outside the field of view
ng) is effective.

On the other hand, in the normal volume rendering, in many cases, the resolution of data is sufficiently smaller than the resolution of the screen, so that it is not necessary to raise or lower the level of the hierarchical structure.
It also takes a relatively long time to read the texture, so it is desirable to always keep all data at the highest resolution in order to ensure interactivity.
Therefore, in the present invention, by expressing the data in a block of fixed resolution instead of the hierarchical representation, the data structure is simplified to ensure the interactivity, and at the same time, the locality in the block is used to reduce the data amount. Improve the efficiency of memory usage.

First (1) First, the entire volume V is divided into blocks V ^B of fixed size 2 ^3N (2 ^3N -block).
ing). At this time, a partial volume of 2 ^3N size is allocated to each block. Next, (2) the resolution is reduced by 1 from N in each block, and the error from the original data at that time is calculated. (3) The error calculation is performed for all blocks, and the operation of lowering the resolution of the block having the smallest error is repeated to determine the optimum resolution for each block while reducing the data size as a whole.

Assuming that the block number is B,
The volume of resolution n in V_n ^BWhen you write
Hum is V_N ^BIs expressed as When changing the resolution, V_N ^B
And V _n ^BThe error of each volume is V_N ^BThe sump
Value sampled using cubic linear interpolation at the point
It is defined by the squared error of. That is, the coordinates in block B
V in (i, j, k)_n ^BThe value of v_n ^B(I, j, k)
Then the original volume at resolution n of block B
Error R_n ^BCan be expressed as equation (10).

[0042]

[Equation 5]

Equation (10) is calculated for all blocks, the resolution n of the block having the smallest value is reduced, and the total volume size is reduced to a size required for processing at an interactive speed. To do. FIG. 7 shows the process of blocking the actual data and how the data size changes. FIG. 6 shows the algorithm of the above volume size reduction. When R _n ^B = 0, the resolution can be reduced without changing the image quality. If not, the resolution is reduced in order from the block expected to have the least loss of image quality until the required memory capacity falls within the capacity supported by the hardware.

In volume rendering using image-aligned slices, 2 ^3N -block
After ing, one solid texture and one set of slice sets are given to each block, and each block is drawn. When rendering using object-aligned slices, in addition to the same blocking,
It is possible to perform 2 ^{3 N} -blocking within each two-dimensional slice. This makes it possible to reduce the size of the volume even with the same error as a whole.
However, in this case, if the error becomes too large, the variation in resolution between adjacent slices becomes large, and as a result, moire may appear strongly.

5. Implementation 5.1 Intermediate slices In order to generate an intermediate slice, it is necessary to perform a multi-texture process while designating a mixing ratio.
In the present invention, the CO of the OpenGL graphics library is used.
This is accomplished using the MBINE texture environment.

In the environment where the multi-texture can be used, two or more texture units are connected as shown in FIG. Each unit has its own texture, texture coordinates and texture environment
In addition to RE (texture value), PRIMARY_COL
OR (diffuse value), PREVIOUS (output of immediately preceding unit), and CONSTANT (user-defined constant) can be used as inputs.

Generation of the texture of the intermediate slice is performed by the first two units (FIG. 8). First save the position of the intermediate slices as a constant, the unit 1 of the S _i, to obtain the texture value of S _{i + 1} in unit 2. Unit 1 uses the REPLACE texture environment and retrieves the texture value as is. In Unit 2, use the INTERPOLATE function of the COMBINE texture environment and
Α blends the texture values for both the B and α channels. The INTERPOLATE function takes three arguments,
When Arg0, Arg1, and Arg2 are used, the textures are mixed by the formula (11).
0 = TEXTURE (= Si), Arg1 = PREVI
OUS (= Si + 1), Arg2 = CONSTANT
When (= α), formula (11) becomes formula (12), which matches the formula of the intermediate slice of formula (9). The texture of the generated intermediate slice can be treated like a normal texture.

[0048]

[Equation 6]

Similar to the texture, when the interpolated texture is mapped to the polygon of the intermediate slice obtained by interpolating the coordinates of the polygon with α, the intermediate slice is obtained. By performing the α blending on the intermediate slice like other slices, it is possible to increase the number of slices without increasing the memory capacity required for processing.

5.2 Volume Shading To facilitate the recognition of three-dimensional shapes, it is effective to carry out shading considering the light source. Especially when the data is homogeneous and the opacity is high, it is difficult to identify other than the contour line of the original shape by volume rendering without considering the light source (FIG. 10). Here, considering only diffuse reflection as the influence of the light source, the Gourand shadin
A method of displaying using g will be described. 2.1. As described above, since it is considered that the volume emits light at each point, the drawing result is a combination of the emitted light and the diffuse reflected light. When the light source direction is 1, the brightness gradient of the volume is n, the light emission at each point of the volume is Ie, and the light reflection of the light source is Id, the shading at each point is given by Equation (13).
Calculate using.

[0051]

[Equation 7]

The gradient vector of the scalar field is calculated in advance by performing an appropriate filter process, and a volume for holding the three-dimensional vector, which is the value, as the RGB values of the voxels is prepared, so that the diffusion is efficiently performed. The reflection component can be calculated. Here, a method for interactively changing the parameters of the light source by using the DOT3_RGB function of the COMBINE texture environment will be described. The DOT3_RGB function is a function that regards the RGB channel of the texture value as a three-dimensional vector and returns the inner product result to the RGB channel of the texture value. This function takes two arguments and returns Arg0 and Arg1, respectively, and returns equation (14) as RGB values. The return value is truncated at [0,1]. By using this function, the inner product n · l necessary for the diffuse reflection calculation is calculated. The color Id of the light source can be realized by multiplying the result of n · l by Id using the MODULATE function, but here, for simplicity, only the diffuse reflection of Id = (1,1,1) (white light) is considered. . Finally, it is mixed with the reflected component of the emitted light using the ADD function.

[0053]

[Equation 8]

Specifically, as shown in FIG. 9, in the texture unit, the light source direction 1 is given to a constant, and the first texture unit obtains the brightness gradient n and the second unit obtains the emitted light I _e . . And DO in the first unit
Id (n · l) is calculated using the T3_RGB function, and 2
In the second unit, using the ADD function, Ia + Id (n
Calculate l).

[0055]

[Example] 6. Experiments were performed on a PC utilizing the Experiments OpenGL graphics library. The environment used is that the CPU is Pentiu
m4 1.7 GHz, main memory 1 GB, PC graphics card used is GeForce3 AGP4x6
4MB DDR, OS is Microsoft Wind
ows 2000.

3. FIG. 4 shows the result of approximation of the cubic linear interpolation using the intermediate slice described above. The input size is 256 × 256 × 128, and the drawing surface size is about double. According to the original result, moiré can be observed especially in the case where the brightness changes drastically (FIG. 4A), but the image quality is improved as a result of generating one intermediate slice between each slice and doubling the number of slices (FIG. 4B). ).

4. Table 1 shows the change in size when the adaptive resolution change is performed in block units as described above.
The allowable value of the error is 5% of the difference between the original volume and the empty volume (all voxels are 0). 2 ^3N block (2 ^N
25% to 75% when using (× 2 ^N × 2 ^N ),
It can be seen that the data can be further reduced when the 2 ^{3 N} block is used.

[0058]

[Table 1]

In general, 70% to 95% of the volume data obtained by tomography is a transparent area, and the amount of data in this area can be effectively reduced by blocking. Generally, as the volume is reduced, the image quality deteriorates. However, when 2 ^{3 N} -blocking is used, most of the quality does not deteriorate and the data amount can be greatly reduced. Figure 1
1 shows a change in image quality when the actual data is blocked and the data is reduced.

The optimum value depends on the processing system and the data, because the smaller the block size is, the more the overhead at the time of drawing increases, and the larger the block size becomes, the more difficult it is to use the locality of the data. In the experiment, 256 × 256 × 256 size 8
As a result of comparing drawing speeds for various sizes with bit volume data, the highest speed was obtained when N = 5, that is, when the length of one side was 32 (FIG. 12). When N = 4, the data size becomes the smallest, but the reason why the drawing speed does not become high is considered to be that the drawing overhead increases due to the increase in the number of blocks.

Finally, Table 2 shows the results of investigating changes in the rendering speed by subjecting various data to 2 ^3N blocking. As a result of simplification of various data with no change in resolution, error tolerance of 5% and tolerance of 10%, the simplification effect depends on the nature of the data, but a certain speed improvement is obtained. Therefore, it can be seen that a huge volume can be drawn at an interactive speed by setting an allowable value of 10%. Generally, 30 fps (1
It can be said that the interactive speed (very comfortable speed) with almost no waiting time is 30 frames or more per second) and 10 fps or more is a practical range.

[0062]

[Table 2]

7. We proposed a method of interactive volume rendering using the functions of the Confusions graphics hardware (eg PC graphics card). Use intermediate slices to solve the visual quality problem that occurs when rendering using slices to utilize the acceleration of the graphics card, and simplify the problem of the size that can be handled to slices. Improved. As a result, high-quality volume rendering can be realized at an interactive speed without special hardware, and a maximum of 512 x 512 can be achieved.
Even with enormous data of size * 512, it was possible to draw at a sufficient speed with almost no loss of image quality. Volume rendering requires a large amount of calculation and requires a special computer and dedicated hardware to be interactive. By using the method of the present invention, it is possible to interactively render a large size volume data even on a normal PC, and therefore, it can be expected to be applied to a field in which a volume could not be handled in view of the conventional speed.

In the present invention, only the direct method volume rendering is targeted, but the present invention can be applied to a method of performing isosurface extraction in the same framework, a method of simultaneously displaying polygons and a volume, and the like.

[0065]

As described above, according to the present invention, in order to reduce the calculation cost of volume rendering, a device (personal computer) equipped with graphics hardware (PC graphics card) is used to represent a volume by stacking slices. However, high-speed drawing is possible by using the texture map function and α-blending function that utilize hardware acceleration. Further, in this method, since the image quality is seriously deteriorated due to lack of data interpolation, the image quality is improved by increasing the number of sampling points by using the multi-texture function. In addition, since the data size that can be handled is small due to the limited memory capacity of the graphics hardware (PC graphics card), adaptive sampling of a single layer using 2 ^3N block ^formation is performed to minimize the deterioration of image quality and make it interactive. It became possible to draw at various speeds.

Therefore, the volume rendering method and the program thereof according to the present invention can (1) perform volume rendering using a normal PC (personal computer) or game machine, and (2) relatively large, such as 128 Mbytes or more. The volume data can be handled at an interactive speed, and (3) further, the deterioration of the image quality can be minimized, which is an excellent effect.

[Brief description of drawings]

FIG. 1 is a schematic diagram of a conventional direct method volume rendering.

FIG. 2 is a schematic diagram of volume rendering of a direct method based on slice data of the present invention.

FIG. 3 is an explanatory diagram of a slice direction.

FIG. 4 is an image on a display when interpolation accuracy is insufficient and when linear interpolation is performed.

FIG. 5 is a schematic diagram showing linear interpolation according to the present invention.

FIG. 6 is a flowchart showing an algorithm for volume size reduction according to the present invention.

FIG. 7 is an image on a display showing a process of blocking data and a change in data size according to the present invention.

FIG. 8 is a schematic diagram showing texture generation of an intermediate slice according to the present invention.

FIG. 9 is a schematic diagram showing texture generation of an intermediate slice by Shading.

FIG. 10 is an image on a display due to a difference in shading considering a light source.

FIG. 11 is an image on a display showing a change in image quality when data is reduced by performing blocking on actual data.

FIG. 12 is a relationship diagram of a block size of 8-bit volume data, a data size, and a rendering speed.

   ─────────────────────────────────────────────────── ─── Continued front page    (72) Inventor Katsushi Ikeuchi             2-12 Umegaoka, Aoba-ku, Yokohama-shi, Kanagawa F term (reference) 5B080 AA13 AA20 CA04 CA05 FA03                       FA09 FA16 FA17 GA14 GA22

Claims (6)

[Claims] 1. Using a device having graphics hardware having a texture map function, an α-blending function, and a multi-texture mechanism, the entire volume data is represented by stacking slice data using the texture map function and the α-blending function. And a basic slice data creating step (A) and an intermediate slice data creating step (B) of linearly interpolating and inserting intermediate slice data between the basic slice data. 2. The insertion interval of the intermediate slice data is
The volume rendering method according to claim 1, wherein the sampling interval on the slice data is set to be equal to or narrower than the sampling interval according to the resolution required for drawing. 3. The texture of the intermediate slice data is a multi-text of graphics hardware.
The volume rendering method according to claim 1, wherein the volume rendering method is created at the time of drawing using the ure mechanism. 4. The whole volume data is represented by fixed resolution blocks instead of hierarchical representation to simplify the data structure, and the locality of the data in each block is used to reduce the amount of data for memory utilization. A volume rendering method characterized by increasing efficiency. 5. A 2 ^3N block dividing step (C) of dividing the entire volume V into blocks V ^B of fixed size 2 ^3N
And a block error calculation step (D) in which the resolution is sequentially decreased from N for each of the divided blocks and an error with the original data generated at that time is calculated, and among the calculation errors of all blocks, A resolution reduction step (E) for reducing the resolution of the block having the smallest error, and steps (D) and (E) are repeated to determine the optimum resolution for each block while reducing the data size as a whole. It is characterized by the above-mentioned.
Volume rendering method described in. 6. A basic slice data creating step (A) in which the entire volume data is expressed by stacking slice data using a texture map function and an α-blending function, and intermediate slice data is linearly interpolated between the basic slice data. The intermediate slice data creation step (B) for inserting the whole volume V and the 2 ^3N block dividing step (C) for dividing the entire volume V into blocks V ^B of fixed size 2 ^3N , and the resolution of each divided block in order from N. The block error calculation step (D) of reducing one by one and calculating the error with the original data generated at that time, and the resolution reduction step of decreasing the resolution of the block with the smallest error among the calculation errors of all blocks ( E) and repeat steps (D) and (E) to reduce the size of the data as a whole. Reluctant, volume rendering program for determining the optimum resolution for each block, characterized in that