Multi-Class (Anisotropic) Electrostatic Halftoning
Novel Contributions
Our previous halftoning algorithm
achieves excellent results, but is has a number of shortcomings: Its
visual representation of very dark image regions is unsatisfactory due
to overlapping inkblots, and it is only applicable for a single class
of dots with a uniform size and colour.
The extensions and modifications introduced in our paper show how to
overcome these limitations and problems. More specifically, we make
the following contributions:
- We introduce the changes necessary to deal
with inkblots of different sizes. This also allows to create second
order screening results similar to those used in the printing
industry. As explained in detail in the paper, our approach contains
a tradeoff parameter that can be used to steer the different
requirements occurring in such situations. Here, we illustrate
the effect of this parameter.
- We give details on how to create halftones for
output devices with arbitrary colour models, including asymmetric
models such as the CMY-RB-K colour model. Here, we
show additional experiments that did not
fit into the paper in a high resolution.
- We present our approach to multi-class
sampling using different objects with varying sizes. As in the case of
multi-size halftoning, we show the effect of
the tradeoff parameter we introduced to steer the algorithm.
- Our fourth and last contribution is a novel
approach for hatching, which can even be combined with any of the
other contributions such that a mixture of dots and lines can be
created. Thereby, a user can freely specify the desired lengths and
directions of the lines depending on their positions in the
image.
Additionally, this page illustrates how to improve
visual results in
dark image areas. This is done by adjusting the underlying image
before halftoning to account for overlapping
inkblots. The deviation of the proposed
halftoning operator and the resulting table containing the
resulting tonemapping operator for 8-bit
channels, which were not included in the paper as they did not really
fit into its scope, are given on the following pages.
Main Page Deviation of Tonemapping Operator >
|