作者: JULIO CESAR TORELLI , RICARDO FABBRI , GONZALO TRAVIESO , ODEMIR MARTINEZ BRUNO
DOI: 10.1142/S0218001410008202
关键词: Mathematics 、 Workstation 、 Image (mathematics) 、 Algorithm 、 Pattern recognition (psychology) 、 Data compression 、 Implementation 、 Domain decomposition methods 、 Cluster (physics) 、 Robotics 、 Artificial intelligence
摘要: The Euclidean distance transform (EDT) is used in various methods pattern recognition, computer vision, image analysis, physics, applied mathematics and robotics. Until now, several sequential EDT algorithms have been described the literature, however they are time- memory-consuming for images with large resolutions. Therefore, parallel implementations of required specially 3D images. This paper presents a implementation based on domain decomposition well-known algorithm, analyzes its performance cluster workstations. use data compression tool to reduce communication time investigated discussed. Among obtained results, this work shows that an essential clusters low-bandwidth networks.