Image filtering on 3D objects using 2D manifolds

Details Image filtering on 3D objects using 2D manifolds Image filtering on 3D objects using 2D manifolds

2001-04-03

2004-06-29

Razavi, Michael (Department: 2672)

Computer graphics processing and selective visual display system

Computer graphics processing

Attributes

C345S419000

Reexamination Certificate

active

06756990

ABSTRACT:

BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to digital image and video processing. More specifically, the present invention relates to methods of image filtering the surfaces of three-dimensional (3D) objects and texture projections.
2. Discussion of Related Art
Traditionally, digital images are captured, displayed and processed as a rectangle in two dimensions. As illustrated in
FIG. 1
, a digital image
100
comprises a plurality of picture elements (pixels). Specifically, digital image
100
comprises Y rows of X pixels. For clarity, pixels in an digital image are identified using a two-dimensional coordinate system. As shown in
FIG. 1
, pixel P(0,0) is in the top left corner of digital image
100
. Pixel P(X−1,0) is in the top right corner of digital image
100
. Pixel P(0,Y−1) is in the bottom left corner and pixel P(X−1, Y−1) is in the bottom right corner.
Many linear image filtering techniques to perform such tasks as noise reduction, edge detection, and image enhancement, have been developed for use with the traditional two-dimensional digital image. A linear filter operation is usually implemented as a discrete convolution operation with the convolution kernal being the filter kernel or filter mask. This process is illustrated in
FIG. 2
, where a moving window mask, usually referred to as a filter mask, centered around a particular pixel is moved across the image. The value of the center pixel, i.e., the center pixel value, is re-assigned a value equal to the sum of each filter kernel value multiplied with the underlying pixel value.
FIG. 2
shows a digital image
200
with a filter mask
210
centered around pixel P(X, Y). Filter mask
210
is a 5×5 mask and includes 25 pixels. Specifically, filter mask
210
includes the pixels in a square having corner pixels P(X−2, Y−2), P(X+2, Y−2), P(X−2, Y+2), and P(X+2, Y+2). During an image filtering operation, filter mask
210
spans across every pixel in digital image
200
and the filtering operation is performed for each pixel. However, pixels along the 2 rows or columns on the edges of digital image
200
can not be fully processed because pixel mask
210
would not have complete data to process the pixels at the edges. Generally, these border pixels are either not filtered or filtered using predefined border parameters. For example, some image processing systems define additional pixels around digital image
200
to allow filtering of the border pixels of digital image
200
. Typically, a dark background color is chosen for the additional pixels. Alternatively, the additional pixels may be based on the border pixels or determined through an extrapolation of the border pixels.
With increasing processing and memory capabilities, computer systems can be used for advanced graphic processing of three-dimensional imaging, such as displaying environment maps and three-dimensional objects using so-called texture maps. However, many graphic hardware systems are designed for handling texture maps only as two dimensional digital images. Furthermore, computer memory addressing is not well suited for three dimensional imaging. Thus, most graphic images are still stored and processed as two dimensional images. However, in many situations the border conditions of two dimensional texture maps do not reflect the actual conditions of the three-dimensional object or environment because the surface of the three-dimensional object or environment is represented in this texture map as a two-dimensional digital image. For example, a cylinder is “unwrapped” into a two dimensional texture map. The two dimensional texture map has a left and right border whereas the actual cylinder does not have a left or right border. Hence, there is a for a method and system to allow image filtering techniques on two-dimensional digital images to be applied to three-dimensional imaging.
SUMMARY
Accordingly, the present invention provides a method and system for applying image filters on the surfaces of three dimensional objects or environments. For example, in one embodiment of the present invention a texture map of a three-dimensional object or a texture projection of an environment having a first face and a second face is filtered by using texels from the second face to filter texels on the edges of the first face. Furthermore, texels from the first face are also used to filter texels on the edges of the second face. Generally, a texel is filtered with a filter mask surrounding the texel. However, for texels near the edges of the first face, the filter mask would extend beyond the edge of first face and contain empty spaces. However, in an embodiment of the present invention, a first-face to second-face pointer, which is associated with an edge of the first face, points to a side in the second face. Using the first-face to second-face pointer, a partial filter mask is defined in the second face. The partial filter mask includes a plurality of second-face texels which are used to filter the texel in the first face. Similarly, a texel near an edge of the second face can be filtered using a plurality of first-face texels. Furthermore, some embodiments of the present invention includes a stride parameter with the first-face to second face pointer. For example, in some embodiments, the stride parameter is the amount a memory pointer must be incremented to point to an adjacent face pixel. The stride parameter can also provides information regarding the relative orientation of the first face and the second face. In accordance with a second embodiment of the present invention, the first face and second face are stored in a texture map. Furthermore, the first-face to second-face pointer is also stored in the texture map. Furthermore, additional faces and pointers can also be stored in the texture map. Some embodiments also include the stride parameter in the texture map.


REFERENCES:
patent: 3582961 (1971-06-01), Shindo
patent: 5410644 (1995-04-01), Thier et al.
patent: 5446833 (1995-08-01), Miller et al.
patent: 5877770 (1999-03-01), Hanaoka
patent: 6005580 (1999-12-01), Donovan
patent: 6009190 (1999-12-01), Szeliski et al.
patent: 6154216 (2000-11-01), Walton
patent: 6317139 (2001-11-01), Williams
patent: 6417861 (2002-07-01), Burgard et al.
patent: 6456287 (2002-09-01), Kamen et al.
patent: 6459433 (2002-10-01), Walton
patent: 6515673 (2003-02-01), Hashimoto et al.
patent: 6559853 (2003-05-01), Hashimoto et al.
patent: 2002/0070947 (2002-06-01), Dorbie et al.
patent: 2002/0075274 (2002-06-01), Walton
patent: 2002/0180748 (2002-12-01), Popescu et al.
“Creating raster omnimax images from multiple perspective views using the elliptical weighted average filter”, Greene et al, Jun. 1986, IEEE Computer graphics and applications 6(6), p. 21-27.*
“Synthetic texturing using digital filters”, Feibush et al, 1980, ACM, p. 294-301.

Affiliated with

Also associated with

Rating

Image filtering on 3D objects using 2D manifolds does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Image filtering on 3D objects using 2D manifolds, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Image filtering on 3D objects using 2D manifolds will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-3351439