GA based adaptive sampling for image-based walkthrough

Abstract

This paper presents an adaptive sampling method for image-based walkthrough. Our goal is to select minimal sets from the initially dense sampled data set, while guaranteeing a visual correct view from any position in any direction in walkthrough space. For this purpose we formulate the covered region for sampling criteria and then regard the sampling problem as a set covering problem. We estimate the optimal set using Genetic algorithm, and show the efficiency of the proposed method with several experiments.