Occlusion points identification algorithm

F De Vivo,M Battipede,P Gili

doi:10.1016/j.cad.2017.06.005

F De Vivo, M Battipede + Show 1 more

Open Access

https://doi.org/10.1016/j.cad.2017.06.005

Copy DOI

Journal: Computer-Aided Design	Publication Date: Jul 1, 2017
Citations: 5	License type: other-oa

Affiliation: Polytechnic University of Turin

Abstract

In this paper a very simple and efficient algorithm is proposed, to calculate the invisible regions of a scene, or shadowed side of a body, when it is observed from a pre-set point. This is done by applying a deterministic numerical procedure to the portion of scene in the field of view, after having been projected in the observer reference frame. The great advantage of this approach is its generality and suitability for a wide number of applications. They span from real time renderings, to the simulation of different types of light sources, such as diffused or collimated, or simply to calculate the effective visible surface for a camera mounted on board of an aircraft, in order to optimize its trajectory if remote sensing or aerial mapping task should be carried out. Optimizing the trajectory, by minimizing at any time the occluded surface, is also a powerful solution for a search and rescue mission, because a wider area in a shorter time can be observed, particularly in situations where the time is a critical parameter, such as, during a forest fire or in case of avalanches. For its simplicity of implementation, the algorithm is suitable for real time applications, providing an extremely accurate solution in a fraction of a millisecond. In this paper, the algorithm has been tested by calculating the occluded regions of a very complex mountainous scenario, seen from a gimbal-camera mounted on board of a flying platform.

Highlights

The great advantage of this approach is its generality and suitability for a wide number of applications. They span from real time renderings, to the simulation of different types of light sources, such as diffused or collimated, or to calculate the effective visible surface for a camera mounted on board of an aircraft, in order to optimise its trajectory if remote sensing or aerial mapping task should be carried out
The algorithm has been tested by calculating the occluded regions of a very complex mountainous scenario, seen from a gimbal-camera mounted on board of a flying platform
The solution proposed in this paper is to consider the differences between the relative positions assumed by the points in the Inertial Reference. This method compares the vector with the points Pi=1...8, as they appear in the Inertial Reference Frame (IRF), and the vector containing the image projections I

Summary

Introduction

Different algorithms have been developed to takle the visibility problem: ray tracing [2, 3], beam tracing, cone tracing, frustum casting [4] and Binary Space Partitioning (BSP) trees [5,6,7] These methods share the common idea of sweeping the scene in the direction defined by a certain set of rays. The proposed algorithm has a computational complexity of O(N) and is based on a totally new approach with respect to those mentioned before It determines invisible points exploiting the difference between their relative positions in the 3D reference frame and on the projected image plane. The moving camera can be interpreted as a moving light source or a sequence of different points of view in a rendering process

Methodology

DE M j

20 Camera P3

Case study

Conclusions