Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
10334015 | Theoretical Computer Science | 2005 | 14 Pages |
Abstract
We discuss the problem of computing all the integer sequences obtained by rounding an input sequence of n real numbers such that the discrepancy between the input sequence and each output binary sequence is less than one. The problem arises in the design of digital halftoning methods in computer graphics. We show that the number of such roundings is at most n+1 if we consider the discrepancy with respect to the set of all subintervals, and give an efficient algorithm to report all of them. Then, we give an optimal method to construct a compact graph to represent the set of global roundings satisfying a weaker discrepancy condition.
Keywords
Related Topics
Physical Sciences and Engineering
Computer Science
Computational Theory and Mathematics
Authors
Kunihiko Sadakane, Nadia Takki-Chebihi, Takeshi Tokuyama,