Computer graphics processing and selective visual display system – Computer graphics processing – Character generating
Reexamination Certificate
1999-10-22
2002-05-07
Brier, Jeffery (Department: 2672)
Computer graphics processing and selective visual display system
Computer graphics processing
Character generating
C345S472000, C345S648000, C345S698000, C382S298000, C382S299000, C382S300000, C382S301000
Reexamination Certificate
active
06384828
ABSTRACT:
BACKGROUND
1. Field of the Invention
This invention relates to digital image-processing apparatus and, more particularly, to such apparatus for enlarging the size of a two-dimensional (2D) image that is displayed on a flat-panel screen composed of at least one 2D array of a predetermined fixed number of individual light-controlling elements.
2. Description of the Prior Art
As known, both television and computer-derived images may be displayed on the screen of either a CRT or a flat-panel display device. Since a CRT produces an image by intensity-modulating a scanning electron beam in accordance with an image-defining signal, there is no problem in displaying an enlarged image on the screen of a CRT display device. However, a flat-panel display device has a fixed pixel density for a given size screen, which fixed pixels comprise a predetermined fixed number of individual light-controlling (e.g., LCD, LED or gas-filled) elements. In a monochrome flat-panel display device, the predetermined fixed number of individual light-controlling elements are arranged in a single 2D array responsive to a single luminance (L) signal, while in a color flat-panel display device, the predetermined fixed number of individual light-controlling elements are arranged in separate, nearly-superimposed, red (R), green (G) and blue (B) 2D arrays responsive, respectively, to separate R, G and B chrominance (Ch) signals. The display of an enlarged computer-derived image, where L and Ch signals comprise discrete pixel sample values, on the screen of a flat-panel display device presents a problem when the ratio of the predetermined fixed number of individual light-controlling elements to the enlarged number of discrete pixel sample values is a fraction between 1 and 2.
When information created at one resolution, say VGA of pixel density 640×480 horizontal and vertical pixels, is to be displayed onto a SVGA display of pixel density 800×600 pixels the image does not fill out the display area; in fact, it covers only 80% of the viewable display area. When the same content is displayed on a XGA screen of 1024×768 pixels only 62.5% of the screen is covered. This problem has created a need for small incremental changes in image size to accommodate the resolution differences of VGA, SVGA, XGA, and UXGA in order to maintain full-screen display usage.
Image content resolution is set by the supplier and is not changeable by the viewer, as is witnessed by anyone who uses the internet. This mismatch between supplier and viewer necessitates an image transformation where the source material is scaled appropriately to match the screen resolution of a viewer's flat-panel display device. The transformation need only change image size by 1.25 for VGA to SVGA, 1.28 for SVGA to SXGA, 1.6 for VGA to XGA, and 2.0 for VGA to SXGA. Current practices are to adjust the size of the image by interpolating the image to stretch it to fit the desired display. Interpolation inserts new pixels at newly required positions based on an arithmetic combination of pixels closest to the new position in the original image.
Most often these interpolation processes impose limits on the fidelity of the newly stretched flat-panel displayed image. Common interpolation functions typically impose a loss of high frequency because of the low-pass nature of the interpolating function, and therefore, a loss in frequency resolution into the data particularly observed on sharp transition signals like text and graphics. This is obviously not desirable because now the displayed information appears to be blurred, which makes for a rather fatiguing viewing experience for the viewer. Image enlargement techniques include both linear and non-linear; however, the objective, which is to stretch the image data by using an interpolating function that combines neighboring pixel data, is the same. These approaches fall short of the desired result to faithfully reproduce the image information with the same frequency characteristics or edge profiles of the original data.
Conversely, prior-art approaches like pixel replication used in scaling infinite-frequency data like binary text and graphics would not work well in scaling gray-scaled images and anti-aliased text and graphics. And because information may come to a viewer with several of these data types mixed together, such as a page of internet data, an enlarging process that works well on both would provide great benefit.
In the prior art, interpolation is employed by digital image-processing apparatus for determining the proper value for each of the enlarged number of discrete pixel sample values. Fractional linear interpolation, taught in our U.S. Pat. No. 5,355,328, entitled “Resampling Apparatus Suitable for Resizing a Video Image,” which issued Oct. 11, 1994, is the simplest approach. Linear interpolation, however, imposes too much higher frequency loss causing blurring of sharp edges. Greggain, in his U.S. Pat. No. 5,502,662, entitled “Method And Apparatus For Quadric Interpolation,” which issued Mar. 26, 1996,”, teaches an improvement on linear interpolation by fitting a second-order curve to the image data instead, thereby reducing frequency loss and maintaining sharper edges. Liu, in his U.S. Pat. No. 5,880,767, entitled “Perceptual Image Resolution Enhancement System”, which issued Mar. 9, 1999, teaches sharpening across multiple frequency bands, and protects against edge overshoots to limit visual artifacts. His system again offers sharpened images, but boosting the signal in each band does not restore the lost resolution resulting from the spreading of the data.
None of these methods work well on the extremely hard (i.e., sharp) edges of binary text and graphics. Because binary data is the most common information displayed in daily uses of computer images, it becomes important to the viewer to have sharp edges. Eye strain and fatigue are the obvious consequences of poor edge fidelity.
Therefore, there is a need for an image-enlarging approach, suitable for use with a flat-panel displayed image, which results in a faithful reproduction of the original image for (1) hard (infinite-frequency) edge data, like binary text graphics, (2) soft (Nyquist-bounded) edge data, like natural-scene images (e.g., digital photographs) or adjacent horizontal or vertical pixels having substantially uniform intensity values, and (3) a mixture of both hard and soft edge data.
SUMMARY OF THE INVENTION
The present invention is directed to digital image-processing apparatus or method responsive to pixel values of pixels defining a digitized original 2D image for increasing the number of the pixels in at least one of horizontal and vertical dimensions of the original image by a factor F=N/M, where (1) each of the pixel values falls within a range of V pixel values which extend from a quantized pixel value of 0 to a quantized pixel value of V−1, (2) N is a first given-valued integer, (3) M is a second given-valued integer and (4) 1<N/M≦2.
Such apparatus comprises shear means incorporating upsampling means and logic means. The shear means is responsive to the pixels defining the original image for shearing the original image at certain positions of the one dimension that are determined solely by the value of factor F, thereby introducing zero-valued shear-gap pixels at each of the certain positions. The logic means is responsive to solely the 6 pixel values of those pixels within a 2×3 sub-area that borders a zero-valued shear-gap pixel at each particular certain position for filling the zero-valued shear-gap at that particular certain position with an interpolated pixel value of the original image in response to the logic means determining that that zero-valued shear-gap occurred at a soft edge of the original image or, alternatively, filling the zero-valued shear-gap at that particular certain position with a logically-chosen non-interpolated hard-edge object pixel value or non-interpolated background pixel value in response to the logic means determining that that zero-valued shear-gap
Arbeiter James Henry
Bessler Roger Frank
Blackman Anthony
Brier Jeffery
NorthShore Laboratories, Inc.
Seligsohn George J.
LandOfFree
Digital processing apparatus and method for use in enlarging... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Digital processing apparatus and method for use in enlarging..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Digital processing apparatus and method for use in enlarging... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2886811