JPH08212357A - Interpolation device, interpolation method, and image generation device - Google Patents

Abstract

(57) An object of the present invention is to provide an interpolating device that operates at high speed and can generate a high-quality image. An interpolation device relates to a first original image storage unit 207 that stores a texture including a plurality of pixels and a predetermined number of pixels included in the texture regardless of the position of a reference point with respect to the texture. The pixel value supply unit 203 that supplies the information to the interpolation processing unit 204, and the interpolation processing unit 20 that executes the interpolation processing by calculating the interpolation pixel data based on the information related to the predetermined number of pixels.
4 and.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing apparatus such as texture mapping and image scaling, and an interpolation apparatus and method for executing interpolation processing to obtain a high quality image in computer graphics. The present invention relates to an image generation device including an interpolation device.

[0002]

2. Description of the Related Art In recent years, computer graphics and
In fields such as game consoles, polygon mapping is often performed by texture mapping. Usually, in a texture mapping method called an inverse map, the coordinates (u, v) of the texture pattern are calculated from the screen coordinates (Xs, Ys), and the color corresponding to the coordinates (u, v) is acquired. The coordinate (u, v) obtained by calculation has a decimal part, whereas the actual texture pattern is defined only on a finite number of grid points where the coordinate (u, v) has an integer value. For this reason,
If the coordinates (u, v) having the decimal part obtained by the calculation are simply rounded to an integer, a beautiful image cannot be obtained by texture mapping. This is because it is affected by folding back strain. A difference appears in the generated image depending on how the texture information between the grid points is interpolated. That is, a high quality image can be generated by interpolating the texture information in the texture mapping.

Interpolation is the calculation of the pixel value of a sample point using the pixel values of the pixels located around the sample point (non-integer). Pixels located around the sample point represent pixels existing in a circular area with a radius of 1 pixel centered on the sample point. There are several methods for this interpolation processing. Bilinear interpolation processing is known as a method capable of generating a high-quality image (George Wolberg, "Digital Image Warping", IEEE C
Published by omputer Society Press, 1990, pp147-pp149).

[0004]

However, since the size of the conventional texture pattern is relatively large,
How to handle the pixel data at the boundary of the texture pattern was not considered.

An object of the present invention is to provide an interpolating apparatus and an interpolating method and an image generating apparatus which are used to generate a high-quality image with fast access to textures.

[0006]

The interpolation device of the present invention comprises a first storage means for storing a texture including a plurality of pixels,
Providing means for providing information related to a predetermined number of pixels included in the texture regardless of the position of a reference point with respect to the texture, and calculating interpolation pixel data based on the information related to the predetermined number of pixels By doing so, an interpolation processing means for executing the interpolation processing is provided, and the above-mentioned object is achieved thereby.

The providing means cuts out a second texture, which is larger than the first texture to be processed, by at least one peripheral pixel, from the texture stored in the first storage means, and the second texture. A second storage unit for storing the data, and an address generation unit for generating an address for accessing a pixel of the second texture stored in the second storage unit according to the position of the reference point with respect to the first texture. You may have it.

The second texture may be cut out to form a torus.

The providing means includes a first storage means including at least a part of the textures stored in the first storage means.
The texture is at least the surrounding 1 from the first texture.
Depending on the texture conversion means for converting the second texture into pixels having a larger size, the second storage means for storing the second texture, and the position of the reference point with respect to the first texture,
Address generating means for generating an address for accessing the pixel of the second texture stored in the second storing means may be provided.

The texture conversion means may include means for copying the pixels of the boundary portion of the first texture to the periphery of the first texture.

The texture converting means may include means for inserting a predetermined fixed value around the first texture.

The fixed value of the processing may be a value representing black.

The fixed value of the processing may be a value representing the background color of the drawing target.

The providing means may include a first means including at least a part of the textures stored in the first storing means.
The second storage means stores the texture, the second storage means stores the texture, the transfer means transfers the first texture to the second storage means, and the second storage means stores the texture according to the position of the reference point with respect to the first texture. An address generation unit that generates an address for accessing a pixel of the first texture may be provided.

The providing means and the interpolation processing means may be formed on a first chip, and the first storing means may be formed on a second chip different from the first chip.

The second storage means may be a memory that operates faster than the first storage means.

The interpolation method of the present invention provides information relating to a predetermined number of pixels included in a texture regardless of the position of a reference point with respect to the texture including a plurality of pixels, and the predetermined number of pixels. Performing the interpolation process by calculating the interpolated pixel data based on the information related to the pixels, whereby the above object is achieved.

The step of providing the information includes the step of cutting out from the texture a second texture which is larger than the first texture to be processed by at least one peripheral pixel, and the position of the reference point with respect to the first texture. , Generating an address for accessing a pixel of the second texture.

The second texture may be cut out to form a torus.

The step of providing the information may include a first texture including at least a part of the textures,
Converting to a second texture that is larger than the first texture by at least one peripheral pixel, and generating an address for accessing a pixel of the second texture according to the position of a reference point with respect to the first texture. May be included.

[0021] The converting step may include a step of copying pixels of a boundary portion of the first texture around the first texture.

The converting step may include a step of inserting a predetermined fixed value around the first texture.

The fixed value of the processing may be a value representing black.

The fixed value of the processing may be a value representing the background color of the drawing target.

The step of providing the information includes the step of generating an address for accessing a pixel of the first texture according to a position of a reference point with respect to a first texture including at least a part of the texture. You may.

The image generating apparatus of the present invention is an image generating apparatus for generating an image by mapping a texture on a polygon, and means for storing a texture including a plurality of pixels, and coordinates of points on a polygon plane. To a coordinate of a point on the texture plane, a means for providing information related to a predetermined number of pixels included in the texture regardless of the value of the coordinate of the point on the texture plane, Means for performing interpolation processing by calculating interpolated pixel data based on information associated with a number of pixels of
Means for generating an image based on the interpolated pixel data, thereby achieving the above object.

[0027]

Embodiments of the present invention will be described below with reference to the drawings.

(Embodiment 1) FIG. 1 shows the configuration of a system 100 for drawing polygons by texture mapping. The system 100 includes a processor 1, a main memory 2, a rendering processor 3, a pixel memory 4, a digital-analog converter (DAC) 5,
And a monitor 6. The system 100 further includes a system bus 7. The processor 1, the main memory 2, the rendering processor 3, and the pixel memory 4 are connected to the system bus 7. Input / output devices (not shown) may also be connected to the system bus 7.

The processor 1 controls geometrical operations for three-dimensional graphic processing for texture mapping, controls the operations of the rendering processor 3 and the DAC 5, and controls the operations of the entire system.

The main memory 2 stores a program executed by the processor 1 and data required for each processing such as opaque processing and texture mapping.
Such data includes polygon data.

The rendering processor 3 generates an image representing a polygon by calculating, for each pixel included in the image stored in the pixel memory 4, pixel data representing the color of the pixel. The image produced by the rendering processor 3 is also stored in the pixel memory 4. In this way, rendering of the polygon is achieved by the rendering processor 3. The operation of the rendering processor 3 is controlled by the processor 1.

The pixel memory 4 stores an image to be drawn by the rendering processor 3 and an original image (texture) used to calculate pixel data of the image.

The image generated by the rendering processor 3 is output to the monitor 6 via the DAC 5.
In this way, the image representing the polygon is displayed on the monitor 6.

FIG. 2 shows the configurations of the rendering processor 3 and the pixel memory 4. The rendering processor 3 includes a DDA (Digital differential analyzer) coefficient calculation unit 201, a mapping address generation unit 202,
The pixel value supply unit 203, the interpolation processing unit 204, and the synthesis processing unit 205 are included. The pixel memory 4 includes a generated image storage unit 206 that stores an image generated by the rendering processor 3, and a first original image storage unit 207 that stores an original image (texture) used to generate the image. Is included.

The DDA coefficient calculation unit 201 has the coordinates (x, y, z) of the vertices of the polygon represented by the xyz coordinate system and u.
Coordinates (u, v) of the vertex of the texture expressed in v coordinate system
The relationship between and is supplied via the system bus 7. For example, the relationship is u = f (x, y) and v = g
It is given in the form of a function (x, y). The DDA coefficient calculation unit 201 generates a DDA coefficient based on the relationship between the coordinates of the vertices of the input polygon and the coordinates of the vertices of the texture. The DDA coefficient is the differential coefficient in the y direction (du /
dy, dv / dy) and partial differential coefficient in the x direction (∂u / ∂
x, ∂v / ∂x). DDA coefficient calculator 201
The DDA coefficient generated by is supplied to the mapping address generation unit 202.

The mapping address generation unit 202 uses the DD
Based on the xy-DDA processing unit 211 that generates the coordinates (x, y) of the polygon based on the A coefficient, the alpha-DDA processing unit 212 that generates the attribute value of the polygon based on the DDA coefficient, and the DDA coefficient based on the DDA coefficient. And a uv-DDA processing unit 213 for generating texture coordinates (u, v) corresponding to the polygon coordinates (x, y). For example, the attribute value of the polygon generated by the alpha-DDA processing unit 212 may be the ratio of the area occupied by the polygon to the area occupied by one pixel (area contribution rate).

The DDA process in the mapping address generator 202 is performed based on the differential coefficient in the y direction, the partial differential coefficient in the x direction, and the initial value (coordinates of the vertices of the polygon). For example, the uv-DDA processing unit 213 is stored in the first original image storage unit 207 based on the differential coefficient in the y direction, the partial differential coefficient in the x direction, and the initial value (coordinates of the vertices of the polygon). Coordinates (u, v) indicating a reference point for the original image (texture) are generated. First original image storage unit 20
The coordinates (u, v) indicating the reference point of the original image (texture) stored in 7 are obtained as described below.

The coordinates of the texture at the edge of the polygon are calculated by applying the recurrence formula of (Equation 1) for each of the plurality of scan lines. (U (n + 1), v (n + 1)) in (Equation 1) is (u
(N), v (n)) represents the coordinates of the texture at the edge of the polygon on the scan line one level below.

[0039]

[Equation 1]

The coordinates of the texture inside the polygon on the same scan line (hereinafter referred to as "span") are calculated by the recurrence formula of (Equation 2). (U (n + 1), v (n + 1)) in (Equation 2) is (u
(N), v (n)) represents the coordinates of the texture at a point that is one pixel in the span direction.

[0041]

[Equation 2]

The pixel value supply unit 203 has coordinates (u, u) of the texture generated by the uv-DDA processing unit 213.
Regardless of the value of v), the coordinates of the predetermined number of pixels and the values of the predetermined number of pixels are supplied to the interpolation processing unit 204.

The coordinate (u, v) of the texture supplied from the uv-DDA processing unit 213 is supplied to the interpolation processing unit 204 as a reference address. In addition, the interpolation processing unit 20
4, the pixel value supply unit 203 supplies the coordinates of a predetermined number of pixels and the values of the predetermined number of pixels. Interpolation processing unit 2
Reference numeral 04 denotes a value (P_g) of the interpolated pixel data by weighting the values of a predetermined number of pixels supplied from the pixel value supply unit 203 according to the value of the reference address (u, v).
en) is calculated. This processing is called bilinear interpolation processing.

For example, FIG. 3 shows a reference address 2100.
An example in which four pixels 2110 to 2140 are arranged near (u, v) is shown. In this case, the pixel value supply unit 203
As shown in FIG. 2, the coordinate values and pixel values of the pixels 2110 to 2140 located near the reference address 2100 (u, v) are supplied to the interpolation processing unit 204. Here, in the present specification, “the vicinity of the reference address (u, v)”
Is defined as the range of a circle centered on the coordinates (u, v) of the texture and having a radius of 1 pixel. Interpolation processing unit 20
4 obtains the value (P_gen) of the interpolated pixel data according to (Equation 3).

[0045]

(Equation 3)

Here, the coordinate value of the pixel 2110 is (u0,
v0), the pixel value of the pixel 2110 is a, the coordinate value of the pixel 2120 is (u0, v1), the pixel value of the pixel 2120 is b, the coordinate value of the pixel 2130 is (u1, v1), and the pixel value of the pixel 2130 is c. , The coordinate value of the pixel 2140 is (u1, v0),
The pixel value of the pixel 2140 is d.

When the reference address 2100 is located near the boundary of the texture, four pixels may not exist in the range of a circle having a radius of 1 pixel and having the reference address as the center. In the present specification, an area in which four pixels do not exist in the range of a circle having a radius of 1 pixel centered on the reference address is referred to as a “boundary area”, and other areas are referred to as a “central area”.
Say.

FIG. 4A shows a central region (hatched portion) and a boundary region (other than the hatched portion) when the pixel value sampling point is the center of the pixel. That is, the central region is represented by (Equation 4). Here, the size of the texture is m × n pixels.

[0049]

[Equation 4]

FIG. 4B shows the central area (hatched area) and the boundary area when the sampling point of the pixel value is the upper left point of the pixel (that is, the point of the pixel closest to the origin of the coordinate system representing the pixel). (Other than the hatched part)
And That is, the central region is represented by (Equation 5). Here, the size of the texture is m × n pixels.

[0051]

(Equation 5)

In FIGS. 4A and 4B, the texture is composed of 4 × 4 pixels, and each pixel is represented by a small square.

The pixel value supply unit 203 informs the interpolation processing unit 204 of the coordinate values of a predetermined number of pixels and the predetermined number of pixels regardless of whether the reference address is in the central area or the boundary area of the texture. And the pixel value.

The pixel value supply unit 203 is connected to the address generation unit 2
08, a second original image storage unit 209, and an original image storage control unit 2
Includes 10 and.

The address generation unit 208 has coordinates (u, u) of the texture generated by the uv-DDA processing unit 213.
According to v), an address for accessing the pixel of the texture stored in the second original image storage unit 209 is generated.

The second original image storage unit 209 stores a part or all of the original image (texture) to be rendered, which is stored in the first original image storage unit 207.
The rendering processor 3 is formed on one chip,
The first original image storage unit 207 may be formed on another chip. For example, the rendering processor 3 has one LS
It can be implemented as I. In addition, the first original image storage unit 20
7 can be implemented as an external memory. In this case, the second original image storage unit 209 has the first original image storage unit 207.
Plays the role of cache / buffer. Therefore, the second
The original image storage section 209 is preferably a memory that operates faster than the first original image storage section 207. As a result, access to the texture stored in the second original image storage unit 209 becomes access inside the LSI. As a result, the texture can be accessed at high speed, and the bus bottleneck between the rendering processor 3 (LSI) and the first original image storage unit 207 (external memory) is eliminated.

The original image storage control unit 210 cuts out at least a part of the original image (texture) stored in the first original image storage unit 207, and outputs the cut out original image (texture) as a second image. It is transferred to the original image storage unit 209.

FIG. 5A shows the first original image storage unit 207.
An example of the original image (texture) 301 stored in FIG. In this example, it is assumed that the texture 301 is composed of 16 × 16 pixels, and the minimum unit for texture mapping is 4 × 4 pixels. Therefore, the texture 301 is divided into 16 4 × 4 pixels. For example, the texture 302 (FIG. 5B) is (u, v) =
This is a 4 × 4 pixel texture surrounded by (4,4), (4,8), (8,8), and (8,4). Further, the texture 304 (FIG. 5B) is (u, v) = (0,1)
2), (0, 16), (4, 16), and (4, 12) are 4 × 4 pixel textures. The minimum unit for texture mapping is determined according to the size of the texture that can be stored in the second original image storage unit 209. In the following description, the size of the texture that can be stored in the second original image storage unit 209 is larger than the minimum unit for texture mapping, which is 4 × 4 pixels, by one peripheral pixel.
Suppose there are × 6 pixels.

FIG. 6 shows an example of the polygon 401 generated in the generated image storage unit 206. The texture 301 (FIG. 5A) is texture-mapped on the polygon 401. Polygon 401 has 16 like texture 301.
It is divided into polygons. The divided polygon 402 corresponds to the texture 302, and the divided polygon 403 corresponds to the texture 304.

The original image storage control unit 210, from the texture 301 stored in the first original image storage unit 207,
A texture 303 (FIG. 5C) that is larger than the texture 302 by one pixel in the surroundings is cut out, and the cut out texture 303 is transferred to the second original image storage unit 209. The texture 303 is (u, v) = (3,3),
6 × 6 surrounded by (3,9), (9,9), (9,3)
This is the pixel texture. The process of cutting out the texture 303 from the texture 301 is an example of the process by the original image storage control unit 210 when the texture 302 does not include the boundary region of the texture 301. Note that a texture larger than the texture 302 by two pixels or more may be cut out from the texture 301.

The original image storage control unit 210 extracts from the texture 301 stored in the first original image storage unit 207,
A texture 305 (FIG. 5C) that is larger than the texture 304 by one peripheral pixel is cut out, and the cut-out texture 305 is transferred to the second original image storage unit 209. The texture 305 is (u, v) = (0,11),
A texture portion of 5 × 5 pixels surrounded by (0,16), (5,16) and (5,11), and (u, v) = (15,
11), (15,16), (16,16), (16,1)
5 × 1 pixel texture part surrounded by 1), and (u,
v) = (0,0), (0,1), (5,1), (5,
1 × 5 pixel texture part surrounded by (0), and (u,
v) = (15,0), (15,1), (16,1),
It is composed of a 1-pixel texture portion surrounded by (16,0). The structure of such a texture 305 is based on the fact that the texture is formed in a torus shape. In the process of cutting out the texture 305 from the texture 301, the texture 304 is the texture 301.
Original image storage control unit 210 in the case of including the boundary area of
It is an example of the processing by. A texture larger than the texture 304 by two pixels or more may be cut out from the texture 301.

The cutout processing of the textures 303 and 305 will be described in more detail below. This cutout process is executed by the original image storage control unit 210.

In FIG. 7, the size of the divided texture is 4 × 4.
It is a pixel, and the size of the texture to be cut out is 6 x 6
A procedure of cutout processing in the case of pixels is shown.

In FIG. 7, (u, v) is a divided texture 302 (30) which is the object of texture mapping.
The coordinate point closest to the origin of 4) is shown. (Cut_u, c
ut_v) indicates the coordinate point closest to the origin of the texture 303 (305) to be cut out from the texture 301. (H, W) indicates the height and width of the texture 301. Here, H and W are powers of two. (I,
j) shows a counter for controlling the clipping process. i is the u of the texture 303 (305) to be cut out
This is a counter for counting the number of pixels in the coordinate direction, and j is v of the texture 303 (305) to be cut out.
It is a counter for counting the number of pixels in the coordinate direction. The symbol "&" indicates an AND operation, and the symbol "^" indicates an exclusive OR operation.

In FIG. 7, steps 605 to 609 form one loop. This loop is used to calculate the coordinates of the texture 303 (305) to be cut out in the u coordinate direction. Steps 602 to 611 form another loop. This loop is the v of the texture 303 (305) to be cut out.
Used to calculate coordinates in the coordinate direction.

Hereinafter, the procedure of the clipping process shown in FIG. 7 will be described in detail step by step.

At step 601, (u, v) is input. Further, the internal counter (i, j) is initialized.
The internal counters (i, j) are respectively in the u coordinate direction and v
It is a counter in the coordinate direction.

In step 602, the v coordinate cut_v of the texture 303 (305) to be cut out is calculated.

In step 603, it is judged whether or not the value of cut_v calculated in step 602 is within the range of the texture 301. If the value of cut_v is within the range of the texture 301, the process proceeds to step 60.
Go to 5. If the value of cut_v is not within the range of the texture 301, the process proceeds to step 605 via step 604.

In step 604, the value of cut_v is converted so that the texture is formed in a torus shape.

In step 605, similarly to step 602, the texture 303 (30 to be cut out)
The u coordinate cut_u of 5) is calculated.

In step 606, the cut_calculated in step 605 is calculated similarly to step 603.
It is determined whether the value of u is within the range of the texture 301. If the value of cut_u is within the range of the texture 301, the process proceeds to step 608. If the value of cut_u is not within the range of the texture 301, the process proceeds to step 608 via step 607.

In step 607, as in step 604, the value of cut_u is converted so that the texture is formed in a torus shape.

In step 608, the counter j in the v coordinate direction is incremented by 1.

In step 609, it is determined whether or not the processing in the v coordinate direction is to be ended.

At step 610, the counter i in the u-coordinate direction is incremented by 1.

In step 611, it is determined whether or not the processing in the u-coordinate direction is completed.

By executing the above steps 601-611, considering that the texture is formed in a torus shape, from the coordinate point (u, v) closest to the origin of the divided texture 302 (304), Texture 30
The coordinate point (cut_u, cut_v) closest to the origin of the texture 303 (305) to be cut out from 2 can be obtained.

As described above, according to the original image storage control unit 210, a texture that is larger than the texture corresponding to the divided polygon by at least one peripheral pixel is cut out,
The cut-out texture is the second original image storage unit 20.
9 is stored. As a result, the second original image storage unit 209
Is always arranged near the reference address regardless of whether the reference address generated by the address generation unit 208 is located within the boundary area of the texture stored in the second original image storage unit 209. It is possible to provide the four processed pixels to the interpolation processing unit 204. Also,
When the rendering processor 3 is mounted on one LSI, the texture stored in the second original image storage unit 209 can be accessed at high speed, so that the memory bottleneck can be eliminated. Furthermore, the image quality obtained by dividing a single texture into multiple textures and performing texture mapping for each of the divided textures is as follows. It is not inferior to the quality of the image obtained.

Referring again to FIG. 2, the interpolation processing unit 204 instructs the synthesis processing unit 205 to interpolate pixel data (P_ge).
n) is supplied, and the attribute value (α_ge) of the interpolation pixel data (P_gen) is supplied from the alpha-DDA processing unit 212.
n) is supplied. Further, the synthesizing processing unit 205 includes pixel data (P_acc_i) from the generated image storage unit 206.
n) and its pixel data attribute value (α_acc_i
n) is supplied. The synthesizing processing unit 205, based on these inputs, generates pixel data (P_acc_ou) of the generated image.
t) and the attribute value (α_acc_ou) of the pixel data
generate t). The pixel data (P_acc_out) of the generated image generated by the synthesis processing unit 205 and the attribute value (α_acc_out) of the pixel data are stored in the generated image storage unit 206. This compositing process is called alpha blending. The alpha blending is a process of synthesizing the interpolated pixel data supplied from the interpolation processing unit 204 and the pixel data stored in the generated image storage unit 206 at a predetermined mixing ratio. The predetermined mixing ratio is determined based on the attribute value supplied from the alpha-DDA processing unit 212 and the attribute value stored in the generated image storage unit 206. Specifically, the alpha blending by the synthesis processing unit 205 is achieved by calculating the pixel data (P_acc_out) of the generated image and the attribute value (α_acc_out) of the pixel data according to (Equation 6).

[0081]

(Equation 6)

The generated image storage unit 206 stores the pixel data of the generated image and the attribute value of the pixel data. First
The original image storage unit 207 stores the original image (texture) used for rendering. Generated image storage unit 206
A general large-capacity memory can be used as the first original image storage unit 207. Generated image storage unit 20
6 and the first original image storage unit 207 can be configured using the same bus. The generated image storage unit 206 and the first original image storage unit 207 correspond to the pixel memory 4 in FIG.

In this way, the generated image is stored in the generated image storage unit 206.

(Second Embodiment) The second embodiment of the present invention will be described below. The configuration of the system 100 and the configuration of the rendering processor 3 are similar to those in the first embodiment. Therefore, the description thereof is omitted here. Compared with the first embodiment, the processing executed by the original image storage control unit 210 is different.

FIG. 8 shows an example of textures stored in the second original image storage unit 209. Original image storage control unit 210
Reads the texture to be texture-mapped (for example, the texture of 4 × 4 pixels) from the first original image storage unit 207. Next, the original image storage control unit 210
Is a texture larger than the texture to be texture-mapped by one pixel in the surrounding (for example, a texture of 6 × 6 pixels) by copying the pixels located in the boundary area of the read texture around the boundary area. ) Is generated. Alternatively, the original image storage control unit 210
May generate a texture that is larger than the texture to be texture-mapped by 2 pixels or more around. The texture thus generated is stored in the second original image storage unit 209.

For example, the texture 501 of 6 × 6 pixels shown in FIG. 8 is obtained by copying the pixels located in the boundary area of the texture 302 of 4 × 4 pixels shown in FIG. 5B around the boundary area. can get. The 6 × 6 pixel texture 502 shown in FIG. 8 is obtained by copying the pixels located in the boundary area of the 4 × 4 pixel texture 304 shown in FIG. 5B around the boundary area.

In this way, the second original image storage unit 20
By generating one pixel around the texture stored in No. 9 by the copying process, although the image quality is slightly deteriorated as compared with the first embodiment, the first original image storage unit 207 stores the second original image. The amount of pixel data transferred to the storage unit 209 can be reduced.

The copying process by the image storage control unit 210 will be described in more detail below.

FIG. 9 shows a texture T0 which is a part of the texture stored in the first original image storage unit 207 and which is a target of texture mapping. In this example, the texture T0 is composed of 4 × 4 pixels, and the pixel value of each pixel included in the texture T0 is the array T0 [i] [j].
Suppose it is stored in. Where i = 0,1,
2,3; j = 0,1,2,3.

The image storage control unit 210 converts the texture T0 shown in FIG. 9 into the texture T1 shown in FIG. In this example, the texture T1 is 6 × 6 (= (4
+2) × (4 + 2)) pixels. The pixel value of each pixel included in the texture T1 is the array T1 [i] [j].
Stored in. Where i = 0, 1, 2, 3, 4, 5;
j = 0,1,2,3,4,5. Texture T1
Is stored in the second original image storage unit 209.

Specifically, the conversion from the texture T0 to the texture T1 is performed according to (Equation 7). Here, since the size of the original texture T0 is 4 × 4 pixels, m = 4 and n = 4.

[0092]

(Equation 7)

Here, the non-integer original image reference address for the texture T0 is (u0, v0), and the texture T1 is
Non-integer original image reference address for (u1, v1)
Then, (u1, v1) can be expressed as (Equation 8) by the conversion equation (Equation 7) from the texture T0 to the texture T1.

[0094]

(Equation 8)

The address generation unit 208 generates the address shown in (Equation 9) and uses the address to access the pixel of the texture stored in the second original image storage unit 209. As a result, the pixel data shown in (Equation 10) is input to the interpolation processing unit 204. In (Equation 10), the outer parentheses ([]) represent the array, and the inner parentheses ([]) represent the Gaussian symbol.

[0096]

[Equation 9]

[0097]

[Equation 10]

Using the pixel data shown in (Equation 10) and a part of the reference address (u1, v1) shown in (Equation 8) as input, the interpolation processing unit 204 obtains interpolated pixel data (P_gen) according to (Equation 11). To generate. Through the above processing, the reference address (u1, v
If 1) is given, the interpolation pixel data (Equation 1)
P_gen shown in 1) can be obtained.

[0099]

[Equation 11]

Next, the interpolated pixel data P_gen 'obtained when the reference address (u0, v0) is given to the texture T0 will be described. With respect to the texture T0, the range of reference addresses that can be subjected to bilinear interpolation processing is (Equation 12) to (Equation 12). In this case, the four pixel data required for the interpolation process are given by (Equation 13) (see FIG. 3). Therefore, when the reference address (u0, v0) is given to the texture T0, P_gen ′ shown in (Equation 14) can be obtained as the interpolation pixel data (see (Equation 11)).

[0101]

(Equation 12)

[0102]

(Equation 13)

[0103]

[Equation 14]

Here, it is obvious that p and q in (Equation 11) and (Equation 14) are the same as in (Equation 8). From (Equation 15), the pixel data shown in (Equation 10) and the pixel data shown in (Equation 13) match. That is, the interpolation pixel data P_gen shown in (Equation 11),
The interpolated pixel data P_gen ′ shown in (Equation 14) is
They match in the range shown in (Equation 12). That is, the same data can be obtained as the interpolation pixel data.

[0105]

(Equation 15)

Next, the reference address (u0, v0) is
The case where the value is outside the range shown in (Equation 12) will be described.
Here, it is assumed that the reference address (u0, v0) is in the range of (Equation 16).

[0107]

[Equation 16]

When referring to the original texture T0, it is impossible to align the four textures required for interpolation processing. However, the pixel value supply unit 2 of the present invention
According to No. 03, it is possible to use (Equation 10) as the texture data used in the interpolation process in the interpolation process. These texture data are (Equation 17).

[0109]

[Equation 17]

The interpolation processing unit 204 generates interpolation pixel data (P_gen) from the texture data and the reference address (u1, v1) shown in (Equation 17). Interpolation pixel data (P_gen) can be calculated from (Equation 11) to (Equation 1)
8).

[0111]

(Equation 18)

In this way, the original texture T
For 0, it is possible to realize interpolation processing by two adjacent pixels sandwiching the reference address. This is the interpolation processing when the reference address is located near the boundary on the texture in FIG. 9, but similarly when the reference address is located near the boundary other than the range shown in (Expression 16), Interpolation processing can be realized.

Thus, according to the present invention, (Equation 19)
Interpolation processing (bilinear interpolation processing) can be performed in the range. Here, it is assumed that the size of the texture is m × n pixels and the sampling point is the center of the pixels.
That is, regardless of the coordinate position of the texture, only 1
The interpolation processing can be realized by one interpolation processing (see (Equation 11)) method.

[0114]

[Formula 19]

The synthesizing unit 205, as described above,
An image is generated according to (Equation 6).

Next, the texture mapping when a certain texture is divided into a plurality of textures will be described.

FIG. 11 shows the correspondence between each pixel of the generated image stored in the generated image storage unit 206, the polygon to be drawn, and each pixel of the texture to be mapped to the polygon. It is a thing. In FIG. 11, a dotted line indicates each pixel of the generated image stored in the generated image storage unit 206, a thick line indicates a polygon to be drawn, and a thin line indicates a texture to be mapped to the polygon. Each pixel is shown.

As shown in FIG. 11, the polygon 41 and the polygon 42 are in contact with each other. It is assumed that the polygon 0 is mapped to the texture 0 and the polygon 42 is mapped to the texture 1. Texture 0
Includes a pixel 43a and a pixel 43b. The pixel 43a has a pixel value a, and the pixel 43b has a pixel value b. Texture 1 includes pixels 43c and pixels 43d. The pixel 43c has a pixel value c, and the pixel 43d has a pixel value d. Here, it is assumed that the pixel of interest is the pixel 44. The pixel of interest 44 is indicated by a cross in FIG. The pixel of interest represents the position of the pixel to which the interpolation pixel data is applied in the generated image storage unit 206.

The interpolation processing unit 204 determines the pixel values a, b, c,
Interpolation pixel data is generated based on d, and the interpolation pixel data is applied to the target pixel 44.

FIG. 12 shows the arrangement of the pixel of interest 44, the pixels 43a and 43b of texture 0, and the pixels 43c and 43d of texture 1 when the coordinates of the pixel of interest 44 are represented by texture coordinates ((u, v) coordinates). For example: The target pixel 44 and the pixels 43a to 43 shown in FIG.
d are the same as those shown in FIG. The pixel of interest 44 shown in FIG. 12 is the generated image storage unit 206.
Of the pixel of interest is projected onto the second original image storage unit 209, and is located at a non-integer point in the (u, v) coordinate system.

In the case of the arrangement shown in FIG. 12, the pixel value a,
Interpolation pixel data based on b, c, and d (Equation 20)
Should be obtained.

[0122]

(Equation 20)

Now, how to obtain (Equation 20) using the divided texture 0 and texture 1 will be described. Here, it is assumed that the texture 1 is mapped to the polygon 42 after the texture 0 is mapped to the polygon 41. However, the processing is the same when the order of mapping is reversed.

First, the interpolation processing section 204 generates interpolation pixel data (P_gen0) for texture 0. The interpolated pixel data (P_gen0) is calculated according to (Equation 11). As a result, (Equation 21) is obtained.
When the ratio of the area occupied by the polygon 0 (contribution rate) to the area occupied by the pixel of interest 44 is 0.4, the attribute value (α0_gen) is represented by (Equation 22). The value stored in the target pixel 44 of the generated image storage unit 206 (P
_Acc_out) is calculated according to (Equation 6) based on the interpolated pixel data (P_gen0) and the attribute value (α0_gen) thus obtained. as a result,
(Equation 23) is obtained.

[0125]

[Equation 21]

[0126]

[Equation 22]

[0127]

(Equation 23)

Next, the interpolation processing section 204 determines the texture 1
Generate interpolated pixel data (P_gen1).
The interpolated pixel data (P_gen1) is calculated according to (Equation 11). As a result, (Equation 24) is obtained. When the ratio (contribution rate) of the area occupied by the polygon 1 to the area occupied by the pixel of interest 44 is 0.6, the attribute value (α1
_Gen) is represented by (Equation 25). The value stored in the pixel of interest 44 of the generated image storage unit 206 (P_ac
c_out) is the interpolation pixel data (P_gen1) thus obtained, the attribute value (α1_gen) and the value P_
It is calculated according to (Equation 6) based on acc_in and the value α_acc_in. As a result, (Equation 26) and (Equation 27) are obtained.

[0129]

[Equation 24]

[0130]

(Equation 25)

[0131]

(Equation 26)

[0132]

[Equation 27]

Thus, (Equation 20) and (Equation 26) match. From this, it is understood that even if one certain texture is divided into a plurality of textures, ideal interpolation can be realized by texture mapping.

By dividing a certain texture into a plurality of textures, discontinuous images are not generated between textures, that is, between polygons. Therefore, it is possible to realize texture mapping that provides a high-quality image.

Further, the pixel data finally generated is set to P
Let n be the number of interpolation pixels associated with the pixel data, and P_gen (i) be the interpolation pixel data of each of the n interpolation pixels. When assuming that the length of the square is 1, the area of the shared area between the pixel on the screen and the interpolation pixel to be generated) is α_gen (i),
The sum of _gen (i) is 1.0 as shown in (Equation 28).
And the value P is obtained according to (Equation 29).

[0136]

[Equation 28]

[0137]

[Equation 29]

From (Equation 7), (Equation 11), (Equation 28), (Equation 29), the texture data corresponding to the polygon associated with the pixel should be reflected in the finally generated pixel data P. It turns out that is possible. In other words, it becomes possible to refer to the pixel data of each texture reflected between the adjacent polygons.
That is, it becomes possible to eliminate the discontinuity between polygons that occurs in the conventional interpolation device. As a result, even if the textures are independent, if they are between adjacent polygons, it is possible to obtain the same effect as if they were treated as continuous textures, and it is possible to realize high-quality texture mapping. Become.

The present invention relates to bilinear interpolation processing which is one method of interpolation processing. The bilinear interpolation process is an essential process for generating a high quality image in computer graphics, especially in image processing such as texture mapping and image scaling.

Further, the present invention reduces the circuit scale for performing the bilinear interpolation processing and realizes the interpolation processing in high-speed and high-quality texture mapping with high processing capability.

In the second embodiment, the sunring point of the texture is the center of the pixel, but the position of the sampling point is arbitrary.

(Third Embodiment) In the second embodiment, a pixel located in the boundary area of the texture 302 is copied around the boundary area to generate a texture 501 which is larger than the texture 302 by at least one peripheral pixel. did. The same effect as that of the copying process can be obtained by changing the process of the address generation unit 208 while storing the texture 302 in the second original image storage unit 209. That is, the uv-DDA processing unit 2
When the texture coordinates (u, v) generated by 13 are located in the boundary area of the texture 302, the address generation unit 208 may not generate an address beyond the area of the texture 302. This allows the second
The storage area required for the original image storage unit 209 can be reduced.

The third embodiment of the present invention will be described below.
Configuration of system 100 and rendering processor 3
The configuration of is the same as that of the second embodiment. Therefore, the description thereof is omitted here. Compared with the second embodiment, the processing executed by the original image storage control unit 210 and the address generation unit 208 is different.

The original image storage control unit 210 reads out the texture (for example, the texture of 4 × 4 pixels) to be texture-mapped from the first original image storage unit 207. Next, the original image storage control unit 210 stores the read texture in the second original image storage unit 209.

Now, the texture T0 shown in FIG. 9 is the second
It is assumed that the original image is stored in the original image storage unit 209. Here, the texture T0 is composed of 4 × 4 pixels, and the pixel value of each pixel included in the texture T0 is the array T0 [i].
It is stored in [j]. Where i = 0, 1, 2,
3; j = 0,1,2,3.

The address generator 208 gives a reference address (u0, v0) to the texture T0 stored in the second original image storage 209. Reference address (u
When 0, v0) is within the range of (Equation 12), the interpolation processing unit 204 can obtain the pixel data required for the interpolation processing from the second original image storage unit 209. As described above, the pixel data required for the interpolation processing is represented by (Equation 13).

Texture mapping when the reference address (u0, v0) is (Equation 30) will be described below. The reference address shown in (Equation 30) is located in the range of (Equation 16) mentioned in the second embodiment.

[0148]

[Equation 30]

FIG. 13 shows that in the texture T0 (Equation 3)
0) indicates the point of the reference address (u0, v0). The point is represented by a small circle 71. Further, the X mark in FIG. 13 represents the sampling point of the texture 0.

FIG. 14 is an enlarged view of the vicinity of the reference address (u0, v0) shown in FIG. Reference numeral 81 indicates a reference address (u0, v0), and reference numerals 82a to 82d indicate four peripheral pixels. The pixel 82b and the pixel 82d are pixels of the texture 0 that actually exist, and have T0 [1] [0] and T0 [2] [0], respectively, as their pixel values. On the other hand, the pixel 82a and the pixel 82c are virtual pixels that do not actually exist.

Thus, the reference address (u0, v0)
Is located in the boundary area of the texture 0, the address generation unit 208 generates an address for obtaining pixel data (see (Equation 13) and (Equation 31)) necessary for performing the interpolation process. To do.

[0152]

[Equation 31]

Pixel data (Equation 3) thus obtained
1) is the same as the pixel data obtained by giving the reference address (u1, v1) to the texture T1 mentioned in the second embodiment. This means that by changing the address generation processing of the address generation unit 208, the texture T1 is not generated, and
It means that the same effect as that of and can be obtained.

The address generator 208 in the third embodiment will be described in detail below.

FIG. 15 shows the internal structure of the address generator 208 in the third embodiment. Address generator 208
Is input with a two-dimensional address. The two-dimensional address is a u address (u_adr) and a v address (v_a
dr)).

The address generator 208 includes the address incrementers 111 and 112 and the address selector 12.
1-124 are included.

The address incrementer 111 increments the input u address (u_adr) by 1, and outputs the address (u_adr ') indicating the result. The address incrementer 112 increments the input v address (v_adr) by 1, and outputs the address (v_adr ') indicating the result.

The address selector 121 selectively outputs u_adr or 0 by comparing u_adr with 0. The output is u_adr_out0.
The address selector 122 selectively outputs v_adr or 0 by comparing v_adr with 0.
Its output is v_adr_out0. The address selector 123 has a size m next to u_adr ′ and the original image.
By comparing with (4 in this example)
It selectively outputs r'or m. The output is u_adr
_Out1. The address selector 124 uses v_a
By comparing dr ′ with the vertical size n (4 in this example) of the original image, v_adr ′ or n is selectively output. The output is v_adr_out1.

Address generation unit 20 having the above-mentioned configuration
When the reference address shown in (Equation 30) is input to 8, the u_adr and v_adr are
Represented by 2).

[0160]

[Equation 32]

First, the address incrementer 111
The operations of 112 and 112 will be described. Input to address incrementers 111 and 112 (u_adr, v_ad
r) and the outputs (u_adr ′, v_adr ′) from the address incrementers 111 and 112 are as follows.
It is represented by (Equation 33).

[0162]

[Expression 33]

Next, the operation of the address selectors 121 to 124 will be described. Here, the function that defines the operation of the address selector 121 is sel1, and the address selector 122 is
The function that defines the operation of sel2, address selector 1
The function defining the operation of 23 is sel3, and the function defining the operation of the address selector 124 is sel4. Using these functions, the output values u_adr_out0, v_
adr_out0, u_adr_out1, v_adr
When _out1 is represented, it becomes as shown in (Equation 34).

[0164]

(Equation 34)

The address generator 208 uses (u_adr_out0, v_) as the pixel data address 1 for interpolation processing.
adr_out0) and outputs (u_adr_out0, v_adr_) as the pixel data address 2 for interpolation processing.
out1) and outputs pixel data address 3 for interpolation processing
As (u_adr_out1, v_adr_out
0) is output, and (u_adr_out1, v_adr_out1) is output as the pixel data address 4 for interpolation processing.

In this way, the address generator 208
Outputs four addresses (pixel data addresses for interpolation processing) to the second original image storage unit 209. Those four addresses are represented by (Equation 35).

[0167]

[Equation 35]

As described above, the address generator 208 having the configuration shown in FIG. 15 can generate an address (Equation 35) for referring to pixel data.
This means that instead of addressing pixel 82a, pixel 8
This means addressing 2b and addressing pixel 82d instead of addressing pixel 82c. as a result,
The pixel value supply unit 203 determines whether the reference address is located in the boundary area of the texture or not.
4 will be supplied with a predetermined number of pixel data. Accordingly, the interpolation processing unit 204 can calculate the interpolated pixel data (P_gen) according to (Equation 11) (see (Equation 36)).

[0169]

[Equation 36]

The subsequent processing is the same as in the second embodiment. That is, in the third embodiment, the address generator 2
08 is the texture T generated according to the second embodiment.
Generate an address as if there were one.

As described above, according to the third embodiment, similarly to the second embodiment, it is possible to realize the interpolation processing at the time of texture mapping with high image quality and high speed. Moreover,
Compared to the second embodiment, there is an advantage that the texture conversion processing by the original image storage control unit 210 is unnecessary and the storage capacity of the second original image storage unit 209 can be small.

(Embodiment 4) In Embodiment 2, the pixels located in the boundary area of the texture 302 are copied around the boundary area to generate a texture 501 which is larger than the texture 302 by at least one peripheral pixel. did. The same effect as that of the copying process can be obtained by changing the process of the interpolation processing unit 204 while storing the texture 302 in the second original image storage unit 209. That is, when the number of pixel data input to the interpolation processing unit 204 is smaller than the number of pixel data required for the interpolation processing (for example, when the number of pixel data input to the interpolation processing unit 204 is less than 4). Includes an interpolated pixel data generation unit 93a that generates pixel data required for interpolation processing based on the input interpolated pixel data.
Should be provided. As a result, the original image storage control unit 210
Since the conversion processing of the texture due to is unnecessary, the storage area required for the second original image storage unit 209 can be reduced as compared with the second embodiment.

The fourth embodiment of the present invention will be described below.
Configuration of system 100 and rendering processor 3
The configuration of is the same as that of the second embodiment. Therefore, the description thereof is omitted here. Compared to the second embodiment, the processing executed by the original image storage control unit 210 and the interpolation processing unit 204 is different.

The original image storage control unit 210 reads out the texture (for example, the texture of 4 × 4 pixels) to be texture-mapped from the first original image storage unit 207. Next, the original image storage control unit 210 stores the read texture in the second original image storage unit 209.

Now, the texture T0 shown in FIG. 9 is the second
It is assumed that the original image is stored in the original image storage unit 209. Here, the texture T0 is composed of 4 × 4 pixels, and the pixel value of each pixel included in the texture T0 is the array T0 [i].
It is stored in [j]. Where i = 0, 1, 2,
3; j = 0,1,2,3.

FIG. 16 shows the internal structure of the interpolation processing unit 204 according to the fourth embodiment. The interpolation processing unit 204 includes an interpolation data generation unit 93a and an interpolation actual processing unit 93b.

The operations of the interpolation data generator 93a and the interpolation actual processor 93b will be described below. Here, it is assumed that the sampling of the texture is performed at the center of the pixel and the texture to be texture-mapped has 4 × 4 pixels.

The reference address (u0, v0) given to the texture T0 stored in the second original image storage unit 209 by the address generation unit 208 is (Equation 12).
If it is within the range, four pixel data are input to the interpolation data generation unit 93a. These four pixel data are an appropriate number of pixel data required by the bilinear interpolation processing. Therefore, the interpolation data generation unit 9
3a outputs the pixel data to the interpolation actual processing unit 93b without performing any processing on the input four pixel data. The interpolation actual processing unit 93a executes bilinear interpolation processing. The operation of the interpolation actual processing unit 93a is the same as the operation of the interpolation processing unit 204 in the first to third embodiments.

On the other hand, when the reference address (u0, v0) is outside the range of (Equation 12), the number of pixel data input to the interpolation processing data generation unit 93a is required for the bilinear interpolation processing. It is less than the number of pixel data. In this case, the interpolation processing data generation unit 93a generates insufficient pixel data. As a result, the interpolation actual processing unit 93b
Will output four pixel data.

Texture mapping when the reference address (u0, v0) is (Equation 30) will be described below. The reference address shown in (Equation 30) is located in the range of (Equation 16) mentioned in the second embodiment.

Since the address generation unit 208 gives the reference address (u0, v0) to the texture T0 stored in the second original image storage unit 209, the interpolation processing data generation unit 93a has The pixel data shown in () is input. However, among the four pixel data shown in (Equation 37), the pixel data shown in (Equation 38) is invalid data or there is no data. The interpolation processing data generation unit 93a determines the value of the given reference address (u0, v0) and the size of the texture T0 (4 × 4).
The two pixel data shown in (Equation 38) are identified as invalid based on (Pixel).

[0182]

(37)

[0183]

(38)

[0184]

[Formula 39]

If the pixel data input to the interpolation processing data generation unit 93a includes invalid pixel data, the interpolation processing data generation unit 93a uses valid pixel data instead of the invalid pixel data. Interpolation actual processing unit 93b
Output to. For example, the pixel data shown in (Formula 39) is output instead of the pixel data shown in (Formula 38). The interpolation actual processing unit 93b executes interpolation processing based on these pixel data. The interpolated pixel data obtained by this interpolating process matches (Equation 36).

Here, the number of valid pixel data among the pixel data input to the interpolation processing data generation unit 93a is 4
It is either one, two, or one. Also, the pixel data required for the interpolation processing is uniquely determined by the behavior of the value of the reference address (u0, v0). From this, it will be easily understood by those skilled in the art that the interpolation processing data generating unit 93a can be realized by a simple multiplexer.

The subsequent processing is similar to that of the second embodiment.

(Fifth Embodiment) The fifth embodiment of the present invention will be described below. The configuration of the system 100 and the configuration of the rendering processor 3 are similar to those in the first embodiment. Therefore, the description thereof is omitted here. Compared with the first embodiment, the processing executed by the original image storage control unit 210 is different.

FIG. 17 shows an example of textures 503 and 504 stored in the second original image storage section 209. The original image storage control unit 210 uses a texture to be texture-mapped (for example, a texture of 4 × 4 pixels).
Is read from the first original image storage unit 207. Next, the original image storage control unit 210 inserts a predetermined fixed value into one pixel around the read texture,
A texture (for example, a 6 × 6 pixel texture) larger by one pixel than the texture to be texture-mapped is generated. Alternatively, the original image storage control unit 210 may generate a texture that is larger than the texture to be texture-mapped by 2 pixels or more around. The texture thus generated is stored in the second original image storage unit 209.

For example, the 6 × 6 pixel texture 503 shown in FIG. 17 is obtained by inserting a fixed value into one pixel around the 4 × 4 pixel texture 302 shown in FIG. 5B. The 6 × 6 pixel texture 504 shown in FIG.
Can be obtained by inserting a fixed value into one pixel around the 4 × 4 pixel texture 304 shown in FIG. 5B. For example, the fixed value is the pixel value 0 (black).

By inserting a pixel value 0 (black) into one pixel around the texture to be texture-mapped, the pixel value gradually approaches 0 from the inside of the polygon toward the boundary of the polygon. This means that the polygon edge has been filtered. As a result, the alias of the polygon edge can be removed.

As described above, according to the fifth embodiment, the interpolation processing and the edge anti-aliasing processing can be realized at the same time.

The fixed value inserted in one pixel around the texture is not limited to the pixel value 0 (black). It is possible to insert an arbitrary value as the fixed value.

The insertion process performed by the image storage control unit 210 will be described in more detail below.

Similar to the second embodiment, the image storage controller 210 converts the texture T0 into the texture T1. However, in the fifth embodiment, this conversion is
0). In (Equation 40), the pixel value representing black is represented as 0.

[0196]

(Equation 40)

The reference address (u0, v0) is (equation 1
If it is within the range shown in 2), the interpolation processing result is the same as that of the second embodiment. However, when the reference address (u0, v0) is outside the range shown in (Equation 12), the interpolation processing result is different from that of the second embodiment.
That is, the reference address (u0, v0) is the texture 0.
When it is located in the boundary area of, the interpolation processing is performed by increasing the ratio of black pixels (pixel value 0) as the texture T0 approaches the boundary (see (Equation 11)).

The subsequent processing is similar to that of the second embodiment. The calculation of the pixel value of the polygon boundary serves as a reference to the vicinity of the texture boundary. This is equivalent to performing processing for smoothing the edges of polygons.

The same effect as that of the inserting process can be obtained by changing the process of the address generating unit 208 while storing the texture 302 in the second original image storage unit 209. That is, when the texture coordinate (u, v) generated by the uv-DDA processing unit 213 is located in the boundary area of the texture 302, the address generation unit 208 does not generate an address beyond the area of the texture 302. do it. Instead, the address generator 208 refers to a specific register in which a fixed value is stored. As a result, the second original image storage unit 20
The storage area required for 9 can be reduced.

Furthermore, it is possible to generate an anti-aliasing process with a high quality background by inserting a pixel value representing the background color of the target on which the polygon is drawn into one pixel around the texture 302.

In the above-described embodiment, the example in which four pixels located near the reference address are used for the interpolation processing has been described. However, the number of pixels used for the interpolation processing is not limited to four. Hereinafter, the case where the number of pixels used for the interpolation processing is 9 and 16 will be described.

FIG. 18 shows the arrangement of pixels in the case of performing interpolation processing using nine pixels located near the reference address.

In FIG. 18, a point 700 indicates the position of the reference address. In the vicinity of the reference point 700, the pixel 701 (pixel value a), the pixel 702 (pixel value b), the pixel 7
03 (pixel value c), pixel 704 (pixel value d), pixel pixel 705 (pixel value e), pixel 706 (pixel value f), pixel 7
07 (pixel value g), pixel 708 (pixel value h), pixel 70
It is assumed that 9 (pixel value i) exists. Further, reference point 70
0 is located at a position surrounded by pixels 701, 702, 704, and 705. As illustrated in FIG. 18, the reference point 700 and the pixel 701 are separated by p in the u-axis direction and only q in the v-axis direction. Suppose they are far apart. Where 0 ≦ p <1, 0 ≦ q
<1.

In this case, the interpolated pixel data (P_gen) for the reference point 700 is calculated according to (Equation 41). This is performed by processing the pixels 701 to 709 for every four pixels, and thus the pixels are located near the reference address.
This means performing interpolation processing using pixels.
The sets for every four pixels in the interpolation processing are shown as sets 711 to 714 in FIG.

[0205]

[Formula 41]

FIG. 19 shows the arrangement of pixels when the interpolation processing is performed using 16 pixels located near the reference address.

In FIG. 19, a point 800 indicates the position of the reference address. In the vicinity of the reference point 800, a pixel 801 (pixel value a), a pixel 802 (pixel value b), a pixel 8
03 (pixel value c), pixel 804 (pixel value d), pixel pixel 805 (pixel value e), pixel 806 (pixel value f), pixel 8
07 (pixel value g), pixel 808 (pixel value h), pixel 80
9 (pixel value i), pixel 810 (pixel value j), pixel 811
(Pixel value k), pixel 812 (pixel value l), pixel pixel 81
3 (pixel value m), pixel 814 (pixel value n), pixel 815
(Pixel value o) and pixel 816 (pixel value r) exist. Further, the reference point 800 has pixels 801, 802, 8
It is assumed that the reference point 800 and the pixel 801 are located in a position surrounded by 05 and 806 and are apart from each other by p in the u-axis direction and q in the v-axis direction, as shown in FIG. Here, 0 ≦ p <1 and 0 ≦ q <1.

In this case, the interpolated pixel data (P_gen) for the reference point 800 is calculated according to (Equation 42). This is because a pixel located in the vicinity of the reference address is processed by processing the pixels 801 to 816 every four pixels.
This means that the interpolation processing is performed using 6 pixels. The set for every 4 pixels in the interpolation processing is shown in FIG.
9 as sets 821-829.

[0209]

(Equation 42)

In the above-described embodiment, the bilinear processing is used as the interpolation processing, and a part of the reference address (u, v) for the original image is input to the interpolation processing unit 204. However, when the interpolation range is wide, simply inputting part of the reference address (u, v) deteriorates the amount of information. Therefore, it is preferable to input the reference address (u, v) capable of calculating at least the values of p and q with high accuracy.

[0211]

According to the present invention, the information related to the number of pixels requested by the interpolation processing unit is interpolated regardless of whether the reference address is located in the central area of the texture or in the boundary area. Can be provided to the department. As a result, the entire area of the texture pattern can be interpolated by a single interpolation process, and thus the exceptional interpolation process required in the conventional interpolation device can be eliminated. As a result, the circuit scale can be reduced.

Furthermore, the image quality does not deteriorate due to the exclusion of the exceptional interpolation process. Moreover, since the interpolation processing is uniform over the entire area of the texture pattern, a very high quality image can be obtained.

Further, according to the present invention, one texture can be divided into a plurality of textures and each texture can be treated independently. Furthermore, in the case where the texture before division is stored in the first memory and the texture after division is stored in the second memory, if a second memory that operates much faster than the first memory is used, The second memory acts as a cache / buffer for the first memory. As a result, it becomes possible to access the texture to be texture-mapped at high speed, and the bus bottleneck is eliminated.

[Brief description of drawings]

FIG. 1 is a diagram showing a configuration of a system 100 for drawing a polygon by texture mapping.

FIG. 2 is a diagram showing a configuration of a rendering processor 3.

FIG. 3 is a diagram showing an example of a reference address and an arrangement of pixels located in the vicinity of the reference address.

4A and 4B are diagrams showing examples of a central region and a boundary region of a texture.

5A is a diagram showing an example of an original image (texture) 301 stored in a first original image storage unit 207, FIG.
(B) is a diagram showing an example of original images (textures) 302 and 304 which are part of the original image (texture) 301,
(C) is a diagram showing an example of original images (textures) 303 and 305 cut out from the original image (texture) 301.

FIG. 6 is a polygon 4 generated in the generated image storage unit 206.
It is a figure which shows the example of 01.

7 is a diagram showing a procedure of cutout processing by the original image storage control unit 210. FIG.

FIG. 8 is a diagram showing an example of textures stored in a second original image storage unit 209.

FIG. 9 is a diagram showing an example of a texture T0 stored in a first original image storage unit 207.

FIG. 10 is a diagram showing an example of a texture T1 stored in a second original image storage unit 209.

11 is a diagram showing a correspondence relationship between each pixel of a generated image stored in a generated image storage unit 206, a polygon to be drawn, and each pixel of a texture to be mapped to the polygon. FIG. .

FIG. 12 is a target pixel 44 and its peripheral pixels 43a to 43d.
It is a figure which shows the example of arrangement | positioning.

FIG. 13 shows a reference address (u
It is a figure which shows the point of 0, v0).

FIG. 14 is a diagram showing reference addresses (u0, v shown in FIG.
It is the figure which expanded the vicinity of 0).

15 is a diagram showing an internal configuration of an address generator 208. FIG.

16 is a diagram showing an internal configuration of an interpolation processing unit 204. FIG.

FIG. 17 is a diagram showing an example of textures stored in a second original image storage unit 209.

FIG. 18 is a diagram showing an arrangement of pixels in the case of performing interpolation processing using nine pixels located near a reference address.

FIG. 19 is a diagram showing an arrangement of pixels when performing interpolation processing using 16 pixels located near a reference address.

[Explanation of symbols]

 1 Processor 2 Main Memory 3 Rendering Processor 4 Pixel Memory 5 Digital-to-Analog Converter (DAC) 6 Monitor 7 System Bus 201 DDA Coefficient Operation Unit 202 Mapping Address Generation Unit 203 Pixel Value Supply Unit 204 Interpolation Processing Unit 205 Synthesis Processing Unit 206 Generated Image Storage Part 207 First original image storage unit 208 Address generation unit 209 Second original image storage unit 210 Original image storage control unit

Claims (21)

[Claims] 1. A first storage means for storing a texture including a plurality of pixels, and a providing means for providing information relating to a predetermined number of pixels included in the texture regardless of the position of a reference point with respect to the texture. And an interpolation processing unit that executes interpolation processing by calculating interpolation pixel data based on information related to the predetermined number of pixels. 2. The cutting means for cutting out, from the texture stored in the first storage means, a second texture that is larger than the first texture to be processed by at least one peripheral pixel. Second storage means for storing a texture, and address generation means for generating an address for accessing a pixel of the second texture stored in the second storage means according to a position of a reference point with respect to the first texture. The interpolating device according to claim 1, comprising: 3. The interpolation device according to claim 2, wherein the second texture is cut out so as to form a torus. 4. The providing means converts a first texture including at least a part of the textures stored in the first storage means into a second texture which is larger than the first texture by at least one peripheral pixel. Texture conversion means, second storage means for storing the second texture, and for accessing the pixels of the second texture stored in the second storage means according to the position of the reference point with respect to the first texture The interpolating device according to claim 1, further comprising an address generating unit that generates the address. 5. The interpolating device according to claim 4, wherein the texture conversion means includes means for copying pixels at a boundary portion of the first texture around the first texture. 6. The interpolation device according to claim 4, wherein the texture conversion means includes means for inserting a predetermined fixed value around the first texture. 7. The interpolation device according to claim 6, wherein the fixed value of the processing is a value representing black. 8. The interpolation device according to claim 6, wherein the fixed value of the processing is a value representing a background color of a drawing target. 9. The providing means includes a second storage means for storing a first texture including at least a part of the textures stored in the first storage means, and the first texture for the second storage means. Transfer means for transferring to the first texture, and address generation means for generating an address for accessing a pixel of the first texture stored in the second storage means according to the position of the reference point with respect to the first texture. The interpolation device according to claim 1, wherein 10. The first providing means and the interpolation processing means
2. The first storage means is formed on a second chip different from the first chip, and the first storage means is formed on a second chip different from the first chip.
The interpolating device according to. 11. The second storage means is a memory that operates faster than the first storage means, 2.
9. The interpolation device according to item 9. 12. A step of providing information related to a predetermined number of pixels included in the texture regardless of the position of a reference point with respect to the texture including a plurality of pixels, and information related to the predetermined number of pixels. Performing interpolation processing by calculating interpolation pixel data based on the above. 13. The step of providing the information comprises the step of cutting out from the texture a second texture that is larger than the first texture to be processed by at least one peripheral pixel, and at the position of a reference point with respect to the first texture. Responsively, generating an address for accessing a pixel of the second texture. 14. The interpolation method according to claim 13, wherein the second texture is cut out so as to form a torus. 15. The step of providing the information comprises: converting a first texture including at least a part of the texture into a second texture that is larger than the first texture by at least one peripheral pixel. 13. The interpolation method according to claim 12, further comprising the step of generating an address for accessing a pixel of the second texture according to a position of a reference point with respect to the texture. 16. The conversion step includes a step of copying pixels of a boundary portion of the first texture around the first texture.
The interpolation method according to item 5. 17. The interpolation method according to claim 15, wherein the converting step includes a step of inserting a predetermined fixed value around the first texture. 18. The interpolation method according to claim 17, wherein the fixed value of the processing is a value representing black. 19. The interpolation method according to claim 17, wherein the fixed value of the processing is a value representing a background color of a drawing target. 20. The step of providing the information includes the step of generating an address for accessing a pixel of the first texture according to a position of a reference point with respect to a first texture including at least a part of the texture. The interpolation method according to claim 12, which comprises: 21. An image generating apparatus for generating an image by mapping a texture onto a polygon, comprising means for storing a texture including a plurality of pixels, and coordinates of a point on the polygon plane to a point on the texture plane. Associated with the predetermined number of pixels, and means for providing information related to the predetermined number of pixels included in the texture, regardless of the coordinate values of the points on the texture plane. An image generation apparatus including means for executing interpolation processing by calculating interpolated pixel data based on the information described above, and means for generating an image based on the interpolated pixel data.