Computer graphics processing and selective visual display system – Computer graphics processing – Attributes
Reexamination Certificate
2000-05-16
2004-11-30
Luu, Matthew (Department: 2672)
Computer graphics processing and selective visual display system
Computer graphics processing
Attributes
C345S589000, C345S593000, C345S603000, C345S604000
Reexamination Certificate
active
06825852
ABSTRACT:
BACKGROUND
The invention relates to techniques for combining two or more images defined by color and alpha information to produce a new image also defined by color and alpha information.
The compositing of digital images is done with images defined by color information and alpha (also called transparency) information. The color information describes what color the image is at a particular location. The alpha information describes how much color there is at a particular location—i.e., in simple, normal mode compositing, how much the color obscures the data that lies beneath it. If the images are considered as painted on sheets of acetate, then the color information describes the color of the paint while the alpha information describes its density.
Colors are generally represented by one or more color components, sometimes referred to as colorants. For example, a color video cathode ray tube can reproduce a significant fraction of the range of visible colors using a mixture of red, green, and blue light. Because of impurities in the inks and other physical limitations, printed color generally uses four inks: cyan, magenta, yellow, and black (CMYK).
To represent the color and alpha information in an image at a particular location in the image, a number for every color component and a number for the alpha value is required. The color values can cover arbitrary ranges; however, these are generally normalized to the range of zero to one, and the range of zero to one will be used in the examples of this specification. Similarly, this specification will treat alpha values in the conventional way as having a range from zero (fully transparent) to one (fully opaque).
In many compositing operations, the compositing operation uses the result of multiplying the color component values by the alpha value. This results in a format known as pre-multiplied alpha. One benefit of this format is that it resolves the question of what color the fully transparent regions have by multiplying it with zero and thereby arriving at zero as the value for the pre-multiplied color components. Thus, <C
0
, C
1
, . . . , Cn, A> is transformed into <C
0
*A, C
1
*A, . . . , Cn*A, A>. (The notation < . . . > indicates a vector of elements that defines a color, where Ci is the value of the i-th color component, A is the value of alpha, and * is the scalar multiplication operator. The values Ci and A are generally normalized to range from zero to one, inclusive.)
The case of pre-multiplied alpha can be generalized to matted images by specifying that the color values are stored as if the image had been blended (matted) against some particular background using linear interpolation on the components. Pre-multiplied alpha corresponds to matting against the color with all components zero. The vector from above now becomes <C
0
*A+M
0
*(1−A), . . . , Cn*A+Mn*(1−A), A>. (The Mi is value of the i-th color component of the matting color.)
What has been described so far is a shared-alpha format, in which all of the color components have the same alpha value. The shared-alpha format can be generalized to a per-component alpha by associating an alpha value Ai with each component, thus yielding:
<<C
0
, A
0
>, <C
1
, A
1
>, . . . , <Cn, An >>.
As long as compositing operations are separable—i.e., can be applied to each color component independently—each component is composited as if it were a grayscale image with a single alpha value. The matted version of per-component alpha looks like this:
<<C
0
*A
0
+M
0
*(1−A
0
), A
0
>, . . . , <Cn*An+Mn*(1−An), An >>.
SUMMARY
The present invention provides methods and apparatus including computer program products for combining source images to form a final image, each image being a digital image defined by color component values and a transparency value. The invention performs the steps of receiving two or more source images, each defined by color component and alpha information; assembling a new, mixed image in per-component format by applying a selection operation on the color and alpha information of the source images to define color component and alpha information of the new image according to a source selection definition; and converting the new per-component image into a shared-alpha format to generate the final image in shared-alpha format.
In general, in one aspect, the invention provides a technique for combining source images to form a final image, each image being a digital image defined by color component values and a transparency value. The technique includes converting each of two or more shared-alpha source images into corresponding per-component source images in which each color component value is paired with a per-component alpha value, the alpha value indicating to what extent color exists in a range of fully transparent to fully opaque; assembling a new image in per-component format by selecting each color and alpha pair for each component of the new image from exactly one of the source images in per-component format according to a source selection definition; and converting the new per-component image into shared-alpha format to generate the final image in shared-alpha format.
Embodiments may include one or more of the following features.
The new per-component image is a matted image having color and alpha pairs at each location of the image <<X
0
, A
0
>, . . . , <Xn, An>>, the Xi being the color value of the new image of the i-th color component, the Ai being the alpha value paired with the i-th color component, and the image being matted with a matting color M having color component values Mi. The final image is in a matted shared-alpha format <C
0
, . . . , Cn, A>, the Ci being the color value of the i-th color component and the A being the shared alpha value. The Ci and A satisfy the two relationships A≧Max (Ai), where i ranges over the color components, and Ci=Xi. The A satisfies the relationship A=Max (Ai). The source images are converted into a matted format with the matting color M before the assembling of the new per-component image occurs.
The new per-component image can be a matted image having color and alpha pairs at each location of the image <<X
0
, A
0
>, . . . , <Xn, An>>, the Xi being the matted color value of the i-th color component, the Ai being the alpha value paired with the i-th color component, and the image being matted with a matting color M having color component values Mi. The final image can be in an unmatted shared-alpha format.
The source images can be converted into a matted format with the matting color M before the assembling of the new per-component image occurs. The source images can be converted from an unmatted format. The source images can be converted from a matted format with a matting color different from M. The source selection definition may be location independent. The number of source images can be two. The color components can be cyan, magenta, yellow, and black and the matting color M can be white.
In another aspect, the present invention provides methods and apparatus including computer program products for combining source images to form a final image, each image being a digital image defined by color component values and a transparency value. The invention performs the steps of converting each of two or more unmatted, shared-alpha source images into a corresponding unmatted per-component source image in which each color component value is paired with a per-component alpha value, the alpha value indicating to what extent color exists in a range of fully transparent to fully opaque; assembling a new image in per-component format <<P
0
, A
0
>, . . . , <Pn, An>> by selecting each color and alpha pair for each component of the new image from exactly one of the per-component source images according to a source selection definition; and converting the new per-component image into a final unmatted shared-alpha image <C
0
, . . . , Cn,
Adobe Systems Incorporated
Chung Daniel J
Fish & Richardson P.C.
Luu Matthew
LandOfFree
Combining images including transparency by selecting color... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Combining images including transparency by selecting color..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Combining images including transparency by selecting color... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3344203