Browse Prior Art Database

Video Image Scaling Algorithm

IP.com Disclosure Number: IPCOM000111122D
Original Publication Date: 1994-Feb-01
Included in the Prior Art Database: 2005-Mar-26
Document File: 2 page(s) / 70K

Publishing Venue

IBM

Related People

Smith, DA: AUTHOR

Abstract

Described is an algorithm for determining the placement of video scaling suitable for hardware implementation in real time operations. The scaling algorithm uses concepts of dropping pixels, keeping pixels, dual averaging, and quad averaging of pixels [*].

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 52% of the total text.

Video Image Scaling Algorithm

      Described is an algorithm for determining the placement of
video scaling suitable for hardware implementation in real time
operations.  The scaling algorithm uses concepts of dropping pixels,
keeping pixels, dual averaging, and quad averaging of pixels [*].

      In prior art, the use of video image scaling algorithms caused
a noticeable banding effect, where a portion of the screen is scaled
differently from another section [*].  For example, one section would
be unscaled representing a series of keep operations and the next
half would be scaled with a series of dual averaging operations.  The
concept described herein improves on the prior art by minimizing the
banding effect thereby providing improved picture quality.

The described concept improves on the placement of video scaling
operations, as follows:

      Upon determining that an optimal combination of operations was
always one of a set of pairs based on the chosen scale factor, such
as 100% to 50% are keeps and duals: 50% to 25% are duals and quads;
less than 25% are quads and drops, any operation placement algorithm
using in scaling algorithms involving dropping and keeping pixels can
be modified.  For scale factors between 100% and 50%, a dual
averaging operation can be substituted for a drop operation simply by
dual averaging the pixel to be dropped with the pixel ahead.  This is
because dropping a pixel and dual averaging a pixel both reduce the
screen size by one pixel.  Therefore, any prior art algorithm
involving keep and drop operations can be used by using keeps and
duals instead of keeps and drops.

      For scale facto...