Image analysis – Image transformation or preprocessing – Changing the image coordinates
Reexamination Certificate
1999-11-04
2003-07-15
Chang, Jon (Department: 2623)
Image analysis
Image transformation or preprocessing
Changing the image coordinates
C348S581000, C345S660000
Reexamination Certificate
active
06594404
ABSTRACT:
IMAGE PROCESSING
This invention relates to image processing and in particular to a method and apparatus for resizing digital video display signals so that the image can be displayed at a different size to the original size of the incoming display signals.
The resizing of digital video display signals is well known and is employed in a number of environments. For example it may be desired to shrink, or minify, an XGA image of 1024×768 pixels for display on a S-VGA display of 800×600 pixels. It may be desired to greatly expand, or magnify or zoom, a digital video signal (existing as, for example, a TV signal or a stored video signal such as on a video cassette or disc) for display on a multiscreen videowall.
Early attempts at resizing involved simple interpolation algorithms which involved omitting pixels and lines during minification and multiplying (repeating) pixels and lines during magnification. This often led to loss of image resolution and aliasing of images. More recent techniques have normally involved a two step process. The original, source pixels are first upsampled (oversampled) to produce intermediate pixels using an interpolator. Second, target pixels are generated from the upsampled pixels using a decimator, generally employing FIR filters. Unfortunately, the latter require substantial amounts of integrated circuitry. Other disadvantages of a two step process with FIR filters is a resolution limit, a limited processing speed and the fact that minification and magnification cannot be conducted using the same chip circuitry.
The present invention provides a new approach to image resizing using a single sampling technique so that source pixels are converted directly into target pixels without intermediate upsampling. The approach enables fast resizing to occur in real time and both minification and magnification may be conducted with the same firmware. Better image resolution is achieved by maintaining the frequency content of the original image to be proportional to the resized picture in order to minimise aliasing whilst preserving image fidelity.
The present invention is based on expressing the amount of energy (brightness) associated with each sampled source pixel to span an area of two samples in duration. The area span has a triangular area that is designed to be proportional to the linear energy distribution of the source pixel. This linear energy representation in a discrete sampled domain avoids the need for the classical intermediate up sampled step in deriving target pixels for any resized factor. From an algorithm based on this principle a set of coefficients are determined that map source to target pixels as a function of resize factor.
The present invention provides a method and apparatus for image processing as specified in the claims hereinafter.
REFERENCES:
patent: 5202670 (1993-04-01), Oha
patent: 5696848 (1997-12-01), Patti et al.
patent: 5912710 (1999-06-01), Fujimoto
patent: 6157744 (2000-12-01), Nagasaka et al.
patent: 6348929 (2002-02-01), Acharya et al.
Chang Jon
Electrosonic Limited
Fulbright & Jaworski L.L.P.
Le Brian
LandOfFree
Image processing 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 processing, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Image processing will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3081897