An improved Constant-Factor Approximation Algorithm for Planar Visibility Counting Problem

TL;DR

This paper introduces a randomized approximation algorithm for the planar visibility counting problem that balances preprocessing space, query time, and approximation accuracy, improving efficiency over previous methods.

Contribution

It presents a novel randomized algorithm with adjustable parameters, offering a tradeoff between space, query time, and approximation quality for the visibility counting problem.

Findings

Expected preprocessing time is $O(n^{4-3eta}\log n)$.

Expected space complexity is $O(n^{4-3eta})$.

Query time is $O(rac{1}{\delta^3}n^{eta}\log n)$.

Abstract

Given a set $S$ of $n$ disjoint line segments in $\mathbb{R}^{2}$, the visibility counting problem (VCP) is to preprocess $S$ such that the number of segments in $S$ visible from any query point $p$ can be computed quickly. This problem can trivially be solved in logarithmic query time using $O(n^{4})$ preprocessing time and space. Gudmundsson and Morin proposed a 2-approximation algorithm for this problem with a tradeoff between the space and the query time. They answer any query in $O_{\epsilon}(n^{1-\alpha})$ with $O_{\epsilon}(n^{2+2\alpha})$ of preprocessing time and space, where $\alpha$ is a constant $0\leq \alpha\leq 1$, $\epsilon > 0$ is another constant that can be made arbitrarily small, and $O_{\epsilon}(f(n))=O(f(n)n^{\epsilon})$. In this paper, we propose a randomized approximation algorithm for VCP with a tradeoff between the space and the query time. We will show that for an arbitrary constants $0\leq \beta\leq \frac{2}{3}$ and $0<\delta <1$, the expected preprocessing time, the expected space, and the query time of our algorithm are $O(n^{4-3\beta}\log n)$, $O(n^{4-3\beta})$, and $O(\frac{1}{\delta^3}n^{\beta}\log n)$, respectively. The algorithm computes the number of visible segments from $p$, or $m_p$, exactly if $m_p\leq \frac{1}{\delta^3}n^{\beta}\log n$. Otherwise, it computes a $(1+\delta)$-approximation $m'_p$ with the probability of at least $1-\frac{1}{\log n}$, where $m_p\leq m'_p\leq (1+\delta)m_p$.