BERTUD: Building footprint extraction and regularization in LiDAR datasets through utilization of a distributed system

Issue Date

2016

Abstract

The Philippines is situated inside the pacific ring of fire, which makes it vulnerable to natural disasters such as typhoons, tsunamis, volcanic eruptions and earthquakes. Disaster risk management plans must be developed and updated to keep up with the effects of climate change. Currently, such plans in the Philippines often lack updated data on the location of infrastructure and residential areas. Thus, building footprint extraction is an important task. One of the tasks of our project, UPLB Phil-LiDAR 1, is to extract building footprints in the Laguna and MIMAROPA areas of the Philippines. Manual digitization of building footprints using GIS is tedious. With limited personnel and an extensive area to cover, devising an automated workflow is important. Through the use of remote sensing techniques, building footprint extraction can be automated. Light Detection and Ranging (LiDAR) is one of the most powerful remote sensing technologies nowadays. We have developed an automated workflow for building footprint extraction and regularization from LiDAR datasets. One of the problems encountered with the created workflow is its running time. A 1km × 1km LiDAR tile of an urban area would take hours to finish. In order to address this problem, we developed BERTUD, a distributed system to enable full utilization of the available computing resources of our project. Distributed systems are systems that make networked computers finish a big task. The system was able to fully utilize the computing resources of our project by providing two layers of maximization: macro-level and micro-level. On the macro-level, BERTUD divides large areas for building extraction to multiple computers effectively speeding up the process. On the micro-level, the slave program maximizes the available resources of its host computer by utilizing multiple CPU cores, without interrupting its current user.

Source or Periodical Title

37th Asian Conference on Remote Sensing, ACRS 2016

Page

1-10

Document Type

Conference Paper

Physical Description

illustrations, graphs

Language

English

Subject

Disaster risk management, Distributed computing, Multi-core processing

Digital Copy

yes

Link to Full Text

Share

COinS