KR101601687B1 - Image processing apparatus and method - Google Patents

Abstract

An image processing apparatus is provided. The image processing apparatus includes a first calculation section for calculating a projected area by projecting a mesh representing shape information of an object of a 3D model onto a plane of a first viewpoint at which the 3D model is to be rendered, And a sampling unit for adaptively allocating sampling points for the mesh based on the sampling points.

Global lighting, rendering, point sampling, shadow map, rendering, radiosity, shadow map

Description

[0001] IMAGE PROCESSING APPARATUS AND METHOD [0002]

Is associated with global illumination rendering of an object composed of a 3D model, and more particularly relates to object sampling used for creating a shadow map in the Radiosity technique.

With the development of hardware and software, there is increasing interest in real-time rendering of 3D models in various fields such as 3-dimensional (3D) games, virtual world animation, and movies.
Generally, in order to increase the rendering quality, the amount of computation increases. However, since there is a limit to increase the specification of the hardware allowed for the image processing, the problem of improvement in the computational efficiency, which reduces the amount of computation while maintaining the same quality, Which became a big target of.
In rendering a 3D model, global illumination is to consider both color values by direct illumination and color values by indirect illumination.
Here, indirect illumination can be modeled, such as light reflections, refractions, transmissions, scattered reflections, etc., at the object surface.
For example, in the rendering based on the Radiosity technique, a VPL (Virtual Point Light) is positioned at a specific position in the 3D model to model the indirect illumination, and a shadow map (Shadow) for the object shape from the point of view of the VPL map) is required.
If point sampling is performed for the object without considering the position or orientation of the camera view, such as when creating a shadow map for an object shape at this VPL point, or when performing point sampling in another rendering method, .

When the object is sampled and used in the rendering process without considering the position and direction of the camera view to which the 3D model is to be rendered, there is a waste of the calculation amount in comparison with the rendering quality. And an image processing apparatus and method for improving rendering processing speed are provided.
There is provided an image processing apparatus and method for minimizing aliasing by concentrating the amount of computation for an object in a part close to or in the direction of a camera view.

According to an aspect of the present invention, there is provided a method of calculating a projected area, comprising: a first calculation unit for projecting a mesh representing shape information of an object of a 3D model onto a plane of a first viewpoint at which the 3D model is to be rendered, And a sampling unit for adaptively assigning sampling points for the mesh based on the sampling points of the mesh.
Here, the sampling unit may allocate a sampling point proportional to the projected area to the mesh.
The first calculation unit may include a memory for temporarily storing the mesh information and a calculation module for calculating an area of the mesh stored on the memory on the plane of the first viewpoint at which the 3D model is to be rendered can do.
According to an embodiment of the present invention, the image processing apparatus further includes a second calculation section for creating a shadow map from the VPL disposed in the 3D model using the sampling point adaptively allocated to the mesh do.
The image processing apparatus may further include a rendering unit rendering the 3D model at the first viewpoint by a radiosity technique using the shadow map.
According to another aspect of the present invention, there is provided a 3D mesh modeling apparatus comprising: a sampling unit for searching at least one mesh of the 3D model corresponding to each pixel of a first viewpoint image to be rendered of the 3D model, And a shadow map computation unit for computing, for each of the at least one mesh, a shadow map for each of at least one VPL placed in the 3D model.
Here, the sampling unit may include: a depth image creating unit that creates a depth image of the 3D model at the first viewpoint; and a depth image creating unit that creates a depth image of the 3D model corresponding to each pixel of the first viewpoint image, And a search unit for searching for one mesh.
The image processing apparatus may further include a rendering unit that renders the 3D model at the first viewpoint by using a shadow map, using the shadow map.
According to another aspect of the present invention, there is provided a method for rendering a 3D model, the method comprising: a VPL arrangement unit for arranging a VPL at a first point in the 3D model, And a VPL color determining unit for determining a color.
The VPL color determining unit sets a circle having a first diameter around the first point and determines an average of object color values in the circle as a light color value of the VPL.
According to another aspect of the present invention, there is provided a method for projecting a 3D model, comprising the steps of: projecting a mesh representing shape information of an object of a 3D model onto a plane of a first viewpoint at which the 3D model is to be rendered, And a sampling step of adaptively allocating a sampling point for the mesh based on the sampling point of the mesh.
According to still another aspect of the present invention, there is provided a method for generating a 3D model, the method comprising the steps of: sampling at least one mesh of the 3D model corresponding to each pixel of a first viewpoint image to be rendered, And for each of the at least one mesh, a shadow map calculation step for calculating a shadow map for each of the at least one VPL placed in the 3D model.
According to another aspect of the present invention, there is provided a method for rendering a 3D model, the method comprising: a VPL arrangement step of arranging a VPL at a first point in the 3D model; And a VPL color determination step of determining a color.

Since the amount of computation is concentrated on an important part in the camera view, the rendering processing speed is greatly improved while maintaining the rendering quality as compared with the conventional image processing method.
Therefore, there is an advantageous aspect in real-time rendering.
Also, assuming the same amount of computation, aliasing can be reduced in an important part of the camera view, such as where the error is likely to be found when the image is rendered, thereby improving rendering quality.

Hereinafter, some embodiments of the present invention will be described in detail with reference to the accompanying drawings. However, the present invention is not limited to or limited by the embodiments. Like reference symbols in the drawings denote like elements.
FIG. 1 shows an image processing apparatus 100 according to an embodiment of the present invention.
The image processing apparatus 100 adaptively samples the object in consideration of the position and / or the direction of the camera view to render the 3D model, in order to increase the efficiency of the rendering operation.
The first calculation unit 110 temporarily stores each mesh of objects of the 3D model in the memory 111.
The calculation module 112 performs projection on each of the plurality of meshes stored in the memory 111 to a virtual plane positioned in the camera view for rendering the 3D model.
When projected in this way, the projected area of the mesh will vary depending on how close to the center of the viewing angle of the camera view is the mesh present and how close to the camera the view is.
Since the basic mesh sizes are mostly the same, the projected area of the mesh will be larger at the front of the camera view. Conversely, the farther away from the camera viewing angle, the smaller the projected area of the mesh.
In addition, the projected area will be large if the mesh is close to the camera view, and vice versa.
The calculation module 112 calculates the area on which the meshes are projected, and the area of these meshes is provided to the sampling unit 120.
Then, the sampling unit 120 compares the projected areas of each of the meshes to find meshes that are closer to the center of the viewing angle and closer to the center of the viewing angle, that is, meshes that are important to rendering.
And, for such meshes, sampling is performed at a high density, for example, two sampling points may be assigned to one mesh.
In addition, the sampling unit 120 may lower the sampling density as it is farther away from the camera view, for example, allocate one sampling point to two or three meshes.
On the other hand, the sampling unit 120 relatively increases the sampling density for the meshes within the viewing angle of the camera view, and allocates a large number of sampling points. On the contrary, for the meshes outside the viewing angle, the sampling density is lowered, , And one sampling point may be allocated to four to five meshes.
Then, the second calculation unit 130 creates a shadow map for each of the illustrated sampling points. The number of sampling points to be considered at the time of creating a shadow map is reduced, and the effect of reducing the amount of computation can be expected.
A more detailed operation of the image processing apparatus 100 will be described later in detail with reference to FIGS.
FIG. 2 illustrates an image processing apparatus 200 according to an embodiment of the present invention.
In the shadow map creation, the image processing apparatus 200 designates important points in the camera view and creates a shadow map of the VPL with respect to the designated points, thereby reducing aliasing and improving computation efficiency.
The sampling unit 210 refers to each pixel of the image to be rendered in the camera view, and samples the points of the 3D model object from which the shadow map is to be calculated.
First, the camera view depth image creating unit 221 creates a depth image viewed from the camera view.
Then, the mesh searching unit 222 searches the meshes corresponding to each pixel of the image to be rendered using the depth image.
By the operation of the mesh search unit 222, points to be considered for efficient shadow map creation are determined.
The shadow map calculation unit 220 creates a shadow map of the VPL by calculating the distance from the VPL to the thus determined points.
Then, the rendering unit 230 uses the shadow map to render the image by the Radiosity method or the like.
A more detailed operation of the image processing apparatus 200 will be described later with reference to Figs.
FIG. 3 illustrates an image processing apparatus 300 according to an embodiment of the present invention.
The image processing apparatus 300 determines the color value of the VPL for the VPL placed on the 3D model.
First, the VPL arrangement unit 310 arranges a plurality of VPLs on the object of the 3D model through a predetermined algorithm.
Then, the VPL color determining unit 320 sets a range having a certain radius in each of the plurality of VPLs.
The radius of each range may be set differently or may be set the same, which may be adjusted by the nature of the VPL, the characteristics of the object, and so on.
Then, the VPL color determining unit 320 determines a color value of the corresponding VPL as a statistical average of the object color values within the set range for each of the VPLs.
The operation of the image processing apparatus 300 will be described later in more detail with reference to FIGS. 9 to 10. FIG.
FIG. 4 illustrates shape information of an object 400 to be sampled according to an embodiment of the present invention.
As a result of 3D modeling, the shape information of each object is stored based on a mesh or a point.
In many cases, a mesh-based geometry information storage method is utilized.
The object 400 is also a case where shape information is stored based on a mesh.
The mesh 410 is a basic unit for expressing the physical position of the object. When rendering 3D models, the color values of these individual meshes when viewed from the camera-view are calculated.
In 3D rendering, the determination of the pixel value of the camera view is the process of calculating the result of the light originating from the light being reflected on the object and entering the pixels of the camera view.
In such a rendering process, it is necessary to consider not only color values by direct illumination but also physical phenomena such as indirect illumination, such as reflected light and diffuse reflection.
The indirect illumination is modeled by a virtual light source (VPL) disposed on the 3D object, a shadow map is calculated for each object, and the pixel values in the camera view are precisely calculated using the shadow maps The method to calculate is the Radiosity method.
In this process, the higher the number of VPLs, the higher the quality of the rendered image, but the greater the amount of computation.
FIG. 5 shows a state in which the object 400 of FIG. 4 is sampled regardless of the camera view 501.
Illustratively, the result of one point sampling for each of the meshes 511, 512, 513, 514, 521, and 531 of the object 400 is shown.
That is, uniform sampling was performed over the entire area of the object 400. [
A shadow map is created for the meshes within the viewing angle? 2 of the VPL 502, and a shadow map is also created for the meshes within the viewing angle? 3 of the VPL 503. These shadow maps can be viewed as a depth image representing the distance from the VPL 502 or the VPL 503 to the illustrated sampling point, respectively.
However, in the case of the mesh 521 and the mesh 531 that are not within the viewing angle? 1 of the camera view 501, the shadow map of the VPL 502 or the shadow map of the VPL 503 is utilized, 0.0 > 501). ≪ / RTI >
That is, sampling the meshes outside the &thetas; 1 range, such as the mesh 521 and the mesh 531, and creating a shadow map for them causes computation inefficiency.
Hence it is necessary to sample different meshes 511 to 514 within view angle? 1 of camera view 501, particularly mesh 511 and mesh 512 near camera view 501, with different meshes .
This process will be described later with reference to FIG.
Figure 6 shows the result of sampling the object 400 of Figure 4 in accordance with an embodiment of the present invention.
As described above, unnecessary shadow map calculations can be minimized, and the computation resources can be concentrated on the sensitive portion, such as the sampling of the object near the camera view 501, and the shadow map creation.
The 3D model to be rendered is provided and when the position and direction of the camera view 501 are designated, the first calculation unit 110 of FIG. 1 maps each mesh of the object 400 of FIG. 4 to the memory 111 Temporarily store.
The calculation module 112 performs projection on the virtual plane located in the camera view 501 for each of the plurality of meshes stored in the memory 111. [
When projecting in this way, there is a difference in the projected area of the mesh depending on how close to the center of the viewing angle &thetas; 1 of the camera view 501 is the mesh and how close the mesh is from the camera view 501 It will fly.
Since the basic mesh size is mostly the same, the area of the projection of the mesh will be larger at the front of the camera view 501, and conversely, the area projected from the mesh will be smaller at a direction away from the viewing angle? 1.
In addition, the projected area will be large if the mesh is close to the camera view 501, and the projected area will be small if it is far away.
The calculation module 112 calculates the area on which the meshes are projected, and the area of these meshes is provided to the sampling unit 120.
The sampling unit 120 then compares the projected areas of each of the meshes to find meshes that are closer to the camera view 501 and closer to the center of the viewing angle, i.e. meshes that are important to rendering.
And, for such meshes, sampling is performed at a high density, for example, two sampling points may be assigned to one mesh.
Points 611 and 612 represent the shape information of the objects sampled at such a high density.
The sampling unit 120 may reduce the sampling density as it is farther away from the camera view 501, for example, one sampling point may be allocated to two or three meshes.
Point 613 and point 614 represent the shape information of the object sampled at such a low density.
On the other hand, the sampling unit 120 relatively increases the sampling density for the meshes within the viewing angle &thetas; 1 of the camera view 501 and allocates a large number of sampling points. Conversely, for the meshes outside the viewing angle & , And one sampling point may be assigned to four to five meshes by lowering the sampling density, for example.
Point 621 and point 631 represent the shape information of the object sampled at such a low density.
Then, the second calculation unit 130 creates a shadow map for each of the illustrated sampling points. The number of sampling points to be considered at the time of creating a shadow map is reduced, and the effect of reducing the amount of computation can be expected.
In addition, even if there is no large difference in the total number of sampling points, since a shadow map of a high density is formed in an important part in the camera view 501, image rendering using the Radiosity method or the like is performed by the rendering unit 140 It is expected that the rendering quality is enhanced in the process of being performed.
For example, when rendering a higher resolution image in the camera view 501, aliasing can be reduced.
7 shows a state in which aliasing occurs according to the position and direction of the camera view 701 when the shadow map 720 of the VPL 702 is created for the object 703. FIG.
If the sampling of the object 703 for creation of the shadow map 720 is uniformly performed on the basis of the VPL 702, the shadow map 720 may have a uniform depth value for the entire object.
5 to 6, in the reference of the camera view 701, the distance from the camera view 701 to the center of the viewing angle of the camera view 701, Meshes are important.
This is because, when rendering the image 710 in the camera view 701, even if the shadow map of the VPL 702 is calculated with respect to the objects not to be rendered anyway, the computation is wasted.
For example, in the process of calculating the color value of the pixel 711 of the image 710 in the camera view 701, the object portion corresponding to the pixel 711 is the point 712.
In this case, a shadow test between point 712 and VPL 702 should be performed to determine whether VPL 702 affects the color value of point 712.
By the way, in the case of the pixel 721 of the shadow map 720 in the direction toward the point 712, it has the depth value 732 of the sampling point directed to the left rather than toward the point 712 exactly.
That is, the actual depth value at entry of VPL 702 towards point 712 is the distance between VPL 702 and point 722, which is the pixel 721 of shadow map 720, Which is different from the sampling point depth value 732 of FIG. In this case, since the object ID itself is different, the error is very large.
This problem introduces a problem of aliasing of the rendered image.
That is, since the point 712 is actually in the shadow state with respect to the VPL 702, the color value of the VPL 702 is influenced by the color value of the pixel 711 of the image 710 rendered in the camera view 701 Do not give.
However, as described above, a false shadow test is performed through the depth value 732 of the pixel 721, which results in the color value of the VPL 702 affecting the color value calculation of the pixel 711.
This problem may be solved by a method such as increasing the resolution of the shadow map 720 or the like, but this is undesirable because of an increase in the amount of unnecessary computation.
Therefore, according to the embodiment of the present invention, object sampling for creating a shadow map is adaptively performed in consideration of the camera view 701, thereby improving computation efficiency and solving the aliasing problem.
Figure 8 shows the result of creating a shadow map 810 of the VPL 702 for the object 703 of Figure 7, in accordance with an embodiment of the present invention.
The sampling unit 210 of FIG. 2 samples the points of the object 703 to which the shadow map is to be calculated by referring to each pixel of the image 710 to be rendered in the camera view 701.
Specifically, the camera view depth image creating unit 221 creates a depth image viewed from the camera view 701. [
Due to the generation of such a depth image, it can be determined which object is close to the camera view 701 to be rendered and which part of the object is important.
In this case, the mesh searching unit 222 searches the meshes corresponding to each pixel of the image 710 to be rendered using the depth image.
The operation of the mesh search unit 222 determines the points to be considered for creating the shadow map 810, such as the point 712. [
The shadow map calculation unit 220 measures the distance d2 from the VPL 702 to the point 812 of the object 703 with respect to the determined points such as the direction toward the point 712. [
When the shadow map 810 is created by this method, since the optimal sampling for rendering the image 710 is performed, aliasing can be reduced while the amount of computation is greatly reduced.
7, in the shadow test for determining whether the color value of the VPL 702 affects the calculation of the color value of the pixel 711, in the example of FIG. 7, the point 712 is not a shadow .
However, in the example of Fig. 8, since d1 is larger than d2, the point 712 is determined as a shadow.
The rendering unit 230 uses the shadow map 810 to render the image 710 according to a Radiosity method or the like.
9 shows a process of calculating the color values of VPLs when the VPLs are placed on the object 900. FIG.
In the rendering method considering global illumination such as Radiosity, a plurality of VPLs (910 to 930, etc.) may be disposed on the object 900.
In the case of the placed VPLs 910 to 930, the color value of the virtual point light may be a color value of a source light (actual light source or another VPL) to be modeled for the VPL, 900, and the color value of the portion where the VPL is disposed, the texture state, and the like.
However, unlike other VPLs such as VPL 910 and VPL 930, assuming that a thin line 901 having a color different from the entire color of the object 900 exists on the object 900, The color value of the thin line 901 is greatly affected by the color value of the thin line 901.
For example, the overall color of the object 900 is green so that the colors of the VPL 910, VPL 930, and most VPLs are also green, and the color of the thin line 901 is red, The color of the VPL 920 is determined to be red.
This may be an error within an allowable range in the entire rendering, but it may be a critical problem depending on the camera view.
Thus, in this portion, if a particular VPL 920 is placed at an outlier portion such as a spot that is different from the surrounding color by chance, the VPL color value may be distorted.
Therefore, according to an embodiment of the present invention, the distortion is minimized to provide a VPL color value capable of faithfully expressing actual indirect illumination.
FIG. 10 illustrates a manner of calculating the color values of VPLs placed on the object 900 of FIG. 9 in accordance with an embodiment of the present invention.
According to an embodiment of the present invention, the VPL arrangement unit 310 of FIG. 3 arranges a plurality of VPLs (910 to 930, etc.) on the object 900 through a predetermined algorithm.
Then, the VPL color determining unit 320 sets a range having a certain radius R in each of the plurality of VPLs (910 to 930, etc.).
For example, the range 1010 is set in the VPL 910, the range 1020 is set in the VPL 920, and the range 1030 is set in the VPL 930. The same goes for other VPLs.
The radius R of each range may be set differently or may be set the same, which may be adjusted by the nature of the VPL, the characteristics of the object, and so on.
The VPL color determining unit 320 determines the color value of the VPL 910 as a statistical average of the object color values in the range 1010 and determines the color value of the VPL 920 as a statistical average of the object color values in the range 1020. [ And determines the color value of the VPL 930 as a statistical average of the object color values in the range 1030.
Of course, the color values may be determined by other methods than the statistical averages. Obviously, according to the present embodiment, unlike the example of FIG. 9, the VPL 910, the VPL 920, the VPL 930 ) Are generally similar to each other, and the actual indirect illumination can be better reflected.
11 illustrates an image processing method according to an embodiment of the present invention.
In step S1110, each mesh of objects of the 3D model is temporarily stored in the memory.
In step S1120, for each of the plurality of meshes stored in the memory, projection to a virtual plane located in a camera view for rendering the 3D model is performed.
In this case, the larger the projection area of the mesh is at the front of the camera view, and the smaller the projected area of the mesh is at the farther away from the camera viewing angle.
In addition, the projected area will be large if the mesh is close to the camera view, and vice versa.
In step S1130, the projected area of the meshes is calculated.
Then, in step S1140, the sampling point is adaptively allocated on the object of the 3D model.
The details are as described above with reference to Figs.
Then, for the adaptively sampled points in step S1150, the shadow maps of each VPL of the 3D model are calculated.
In step S1160, the image is rendered by the Radiosity method or the like.
12 illustrates an image processing method according to an embodiment of the present invention.
In step S1210, a camera view depth image is generated in the camera view in which the 3D model is to be rendered.
When the depth image is generated, in step S1220, meshes corresponding to each pixel of the image to be rendered are searched.
Through this process, among the whole parts of the object of the 3D model, the points to be considered for efficient shadow map creation are determined.
Then, in step S1230, for the determined points, a shadow map of the VPL is created by calculating the distance from the VPL.
Then, using the shadow map, the image in the camera view is rendered by the Radiosity method or the like in step S1240.
The contents of the above steps are as described above with reference to Figs.
13 illustrates an image processing method according to an embodiment of the present invention.
In step S1310, a plurality of VPLs are placed on the object of the 3D model.
In step S1320, a range having a predetermined radius is set in each of the plurality of VPLs, and a color value of the corresponding VPL is determined as a statistical average of object color values within the set range for each VPL.
The radii of the respective ranges may be set differently or may be set to be the same, which may be adjusted by the nature of the VPL, the characteristics of the objects, and the like.
Then, in step S1330, the final VPL color value is determined in consideration of the characteristics of the actual light source, the layout, the texture state of the object, and the like in the determined VPL color value.
This process is as described above with reference to Figs. 9 to 10.
The method according to an embodiment of the present invention can be implemented in the form of a program command which can be executed through various computer means and recorded in a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, and the like, alone or in combination. The program instructions recorded on the medium may be those specially designed and constructed for the present invention or may be available to those skilled in the art of computer software. Examples of computer-readable media include magnetic media such as hard disks, floppy disks and magnetic tape; optical media such as CD-ROMs and DVDs; magnetic media such as floppy disks; Magneto-optical media, and hardware devices specifically configured to store and execute program instructions such as ROM, RAM, flash memory, and the like. Examples of program instructions include machine language code such as those produced by a compiler, as well as high-level language code that can be executed by a computer using an interpreter or the like. The hardware devices described above may be configured to operate as one or more software modules to perform the operations of the present invention, and vice versa.
While the invention has been shown and described with reference to certain preferred embodiments thereof, it will be understood by those of ordinary skill in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims. This is possible.
Therefore, the scope of the present invention should not be limited to the described embodiments, but should be determined by the equivalents of the claims, as well as the claims.

FIG. 1 illustrates an image processing apparatus according to an embodiment of the present invention.
FIG. 2 illustrates an image processing apparatus according to an embodiment of the present invention.
FIG. 3 illustrates an image processing apparatus according to an embodiment of the present invention.
FIG. 4 illustrates shape information of an object to be sampled according to an embodiment of the present invention.
FIG. 5 shows the object of FIG. 4 sampled regardless of the camera view.
FIG. 6 shows a result of sampling the object of FIG. 4 according to an embodiment of the present invention.
FIG. 7 shows how aliasing occurs according to the position and direction of the camera view when a shadow map of VPL is created for an object.
Figure 8 shows the result of creating a shadow map of the VPL for the object of Figure 7 in accordance with one embodiment of the present invention.
9 shows a process of calculating the color value of the VPL when the VPL is placed on the object.
FIG. 10 illustrates a manner of calculating a color value of a VPL disposed on the object of FIG. 9 according to an embodiment of the present invention.
11 illustrates an image processing method according to an embodiment of the present invention.
12 illustrates an image processing method according to an embodiment of the present invention.
13 illustrates an image processing method according to an embodiment of the present invention.

Claims (20)

A first calculator for projecting a mesh representing shape information of an object of the 3D model to a plane of a first viewpoint at which the 3D model is to be rendered and calculating an area projected on the basis of the plane; And A sampling unit for adaptively allocating a sampling point for the mesh based on the projected area, And the image processing apparatus. The method according to claim 1, Wherein the sampling unit assigns a number of sampling points proportional to the projected area to the mesh. The method according to claim 1, The first calculation unit may calculate, A memory for temporarily storing the mesh information; And A calculation module for calculating an area of the mesh stored in the memory on the plane of the first viewpoint at which the 3D model is to be rendered, And the image processing apparatus. The method according to claim 1, And a second calculation unit for generating a shadow map from the VPL placed in the 3D model using the sampling point adaptively allocated to the mesh, Further comprising: 5. The method of claim 4, A rendering unit that renders the 3D model at the first viewpoint by a radiosity technique using the shadow map; Further comprising: A sampling unit for searching at least one mesh of the 3D model corresponding to each pixel of a first viewpoint image to be rendered of the 3D model among a plurality of shape information meshes constituting the 3D model; And For each of the at least one mesh, a shadow map calculation unit for calculating a shadow map for each of at least one VPL placed in the 3D model, And the image processing apparatus. The method according to claim 6, Wherein the sampling unit comprises: A depth image creating unit for creating a depth image for the 3D model at the first time point; And And a search unit for searching at least one mesh of the 3D model corresponding to each pixel of the first viewpoint image using the depth image, And the image processing apparatus. The method according to claim 6, A rendering unit that renders the 3D model at the first viewpoint using a shadow map, Further comprising: A VPL placement unit for placing a VPL at a first point in the 3D model, for rendering a 3D model; And A VPL color determining unit for determining a light color of the VPL in consideration of an object color value of the first point surrounding area, And the image processing apparatus. 10. The method of claim 9, Wherein the VPL color determining unit sets a circle having a first diameter around the first point and determines an average of object color values in the circle as a light color value of the VPL. Projecting a mesh representing shape information of an object of the 3D model to a plane of a first viewpoint at which the 3D model is to be rendered and calculating an area projected on the basis of the plane; And A sampling step of adaptively allocating a sampling point for the mesh based on the projected area And an image processing method. 12. The method of claim 11, Wherein the sampling step comprises: And assigning a number of sampling points proportional to the projected area to the mesh. 12. The method of claim 11, Wherein calculating the projected area comprises: Temporarily storing the mesh information in a memory; And Calculating an area of the mesh stored on the memory, the area projected on a plane of a first viewpoint at which the 3D model is to be rendered And the image processing method. 12. The method of claim 11, Creating a shadow map from a VPL disposed in the 3D model using a sampling point adaptively allocated to the mesh; Further comprising the steps of: 15. The method of claim 14, Rendering the 3D model at the first viewpoint by means of a radiosity technique using the shadow map; Further comprising the steps of: A sampling step of searching at least one mesh of the 3D model corresponding to each pixel of a first viewpoint image to be rendered of the 3D model among a plurality of shape information meshes constituting the 3D model; And Calculating, for each of said at least one mesh, a shadow map for each of at least one VPL placed in said 3D model; And an image processing method. 17. The method of claim 16, Wherein the sampling step comprises: Creating a depth image for the 3D model at the first viewpoint; And Searching at least one mesh of the 3D model corresponding to each pixel of the first viewpoint image using the depth image, And an image processing method. A VPL placement step of placing a VPL at a first point in the 3D model, for rendering a 3D model; And A VPL color determination step of determining a light color of the VPL in consideration of an object color value of the first point surrounding area And an image processing method. 19. The method of claim 18, The VPL color determination step may include: Setting a circle having a first diameter around the first point; Obtaining an average of object color values in the circle; And Determining a light color value of the VPL And an image processing method. A computer-readable recording medium storing a program for performing the image processing method according to any one of claims 11 to 19.

Images