LCD anti-aliasing in Java

[Tim Tyler]

I have written a program to perform LCD anti-aliasing on
24-bit RGB images.

Introduction to LCD anti-aliasing
=================================

LCD anti-aliasing attains improved image quality on LCD
devices by directly addressing the red green and blue pixels
on the device to achieve increased resolution.

Colour balance is maintained by distributing colour energy
to surrounding pixels - taking advantage of the fact that
the human visual system has better monochrome resolution
than colour resolution.

The effect produces the best results when used with high
contrast images. It has no effect on images which consist
entirely of shades of one of the primary colours.

In practice the majority images have a fair degree of
contrast - due to common effects such as shadows and
highlights.

Applications
============

The number of devices with small LCD screens is rocketing
upwards - now that mobile phones are routinely including
cameras.

I expect this application to be used by those making
wallpaper, photographs or animated images for these
kinds of devices - in the cases where image quality
is regarded as important.


Requirements
============

It needs Java 1.4 - or greater - to run.


License
=======

This program and its source code are in the public domain.

Download
========

More information, example images - and free download:

--> http://mandala.co.uk/lcdaa/ <--

Enjoy,

 

[Roedy Green]

LCD anti-aliasing attains improved image quality on LCD
devices by directly addressing the red green and blue pixels
on the device to achieve increased resolution.

Do you use the same techniques as Steve Gibson?

 

[Tim Tyler]

:>LCD anti-aliasing attains improved image quality on LCD
:>devices by directly addressing the red green and blue pixels
:>on the device to achieve increased resolution.

: Do you use the same techniques as Steve Gibson?

I believe so.

He discusses using coefficients of [1/9, 2/9, 3/9, 2/9, 1/9] to smear
the colour information out further, and produce more of a central peak.

I'm using the simpler [1/3, 1/3, 1/3] coefficients - but they seem to
work well enough.