A Neural Halftoning Algorithm Suiting VLSI Implementation

@inproceedings{bgz_icassp90,
    author  = "Bernard, T. and Garda, P. and Zavidovique, B. ",
    title   = "A Neural Halftoning Algorithm Suiting {VLSI} Implementation",
  booktitle = "Proc. IEEE Int. Conf. on Acoustics, Speech and Signal Processing",
    address = "Albuquerque, New Mexico",
    pages   = "981-984",
    month   = apr,
    year    = 1990
}

Abstract
We present halftoning as an examplary application where the use of simple neural networks proves of immediate interest. Halftoning is a non standard A/D conversion that we treat as an optimization problem, subject to a frequency weighted MSE criterion. We implement the frequential weight thanks to a specific neural interconnection network based on current diffusion in resistive grids. This "physical" choice not only leads to a dramatically compact VLSI switched capacitor implementation, but also turns the whole process into a "clean" 2-D isotropic generalization of sigma-delta modulation. For the first time, isotropy and shift-invariance cooperate within the same halftoning process for the sake of image rendition. The reached performances prove equal to the deep underlying harmony between the theoretical, algorithmic and material aspects of the procedure.

Paper not available

Back to global view of Thierry Bernard's publications