A Parallel Algorithm for Visibility Problem on GPU<br/>

Please enable javascript in your browser.

A Parallel Algorithm for Visibility Problem on GPU

753 Viewed

Type of Document: M.Sc. Thesis
Language: Farsi
Document No: 44264 (19)
University: Sharif University of Technology
Department: Computer Engineering
Advisor(s): Ghodsi, Mohammad; Abam, Mohammad Ali
Abstract:
Finding visibility region of an observer is one of the old and well-known problems in computational geometry and computer science.In this Problem, considering an observer (point, line segment or a geometrical object)inside an environment, the goal is to find the visible portion of that environment from the observer. Today visibility is used in many fields of computer science including robotics, computer vision and graphics. Nowadays tending to parallelism due to increases in data size and the need for accurate and complicated computations is inevitable. The GPU is composed of many weak processing units, which makes it a cheap and reachable resource for high performance computing. Due to it’s high potency, reachability and scalability, the GPU is one of the best solutions for parallel computing. Although visibility is an old and important problem, There is not any impressive practical parallel algorithm for it. In this thesis, we propose an efficient and suitable parallel algorithm for computing point and line segment visibility inside a simple polygon. This algorithm is very simple and it can be easily implemented on current GPU devices
Keywords:
Visibility Algorithm ; Visibility Polygon ; Simple Polygon ; Visibility Chain