Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Temporal-Based Dithering for Reducing Perceived Quantization Error in Video Displays

IP.com Disclosure Number: IPCOM000116359D
Original Publication Date: 1995-Sep-01
Included in the Prior Art Database: 2005-Mar-30
Document File: 4 page(s) / 128K

Publishing Venue

IBM

Related People

Lee, WR: AUTHOR [+3]

Abstract

Computer systems are now being used to display natural images similar to television images. Computer systems internally must represent the image as a sequence of quantified samples instead of the smooth analog values in television systems. Increasing the fineness of the quantization removes visual artifacts at the cost of increased system cost.

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

Temporal-Based Dithering for Reducing Perceived Quantization Error
in Video Displays

      Computer systems are now being used to display natural images
similar to television images.  Computer systems internally must
represent the image as a sequence of quantified samples instead of
the smooth analog values in television systems.  Increasing the
fineness of the quantization removes visual artifacts at the cost of
increased system cost.

      This problem is solved by implementing a novel adaptation of
dithering to decrease the quantization error without increasing the
system cost.  The use of dithering to decrease quantization error is
well known, particularly in the production of still print images.
Dithering improves the perceived image quality in systems that
transform an image from a higher resolution system into a lower
resolution system.

      When an analog signal is digitized with simple thresholding, an
error termed "quantization error" results.  The error in the
digitized image will be less than the resolution of a single least
significant bit in the digital code.  Any input signal whose value is
between two adjacent digital code points is given the digital code
for the lower of the two digital codes it is between.  This
"quantization" of the signal can produce ugly "stair step" artifacts
in the resulting digital image.  Images that were smoothly shaded in
the analog domain become badly "banded" in the digital domain.
Dithering is a widely used method to decrease these artifacts.  It
consists of adding a noise signal to the input signal before
quantizing the signal.  This noise signal is usually limited in
magnitude to the value represented by a single least significant bit
of the digital code.  When this noise signal is added to the
original, it will cause some of the values near the quantization
points to be pushed to the opposite side of the quantization point.
This has the effect of smoothing the abrupt "stair step" boundaries
that the image would have without dithering.  However, adding noise
to the image can produce perceptible high frequency artifacts.  The
type of noise signal to add is determined by finding a noise signal
that does not produce objectionable artifacts.  This same
"quantization error" occurs when a digital signal is transformed from
a digital code to a new digital code with less bits of resolution.
This type of transformation is used to decrease the size of a
digitally encoded image.

      The purpose of this invention is to reduce the high frequency
artifacts created by transforming an 8-bit digital encoded image into
a n bit digitally encoded image.  (n being less than 8) In full
motion video, many consecutive images are shown.  Our visual system
integrates these discrete images over time giving us the illusion of
continuous, smooth motion.  In this invention, the noise signal for
the dithering is prod...