GPGPU implementation of growing neural gas: Application to 3D scene reconstruction

被引:7
作者
Orts, Sergio [1 ]
Garcia-Rodriguez, Jose [1 ]
Viejo, Diego [2 ]
Cazorla, Miguel [2 ]
Morell, Vicente [2 ]
机构
[1] Univ Alicante, Dept Comp Technol, E-03080 Alicante, Spain
[2] Univ Alicante, Inst Invest Informat, E-03080 Alicante, Spain
关键词
Growing neural gas; Parallel computing; GPU; CUDA; Multicore; 3D reconstruction; Egomotion; SURFACE RECONSTRUCTION; VECTOR QUANTIZATION; NETWORK; MAPS;
D O I
10.1016/j.jpdc.2012.05.008
中图分类号
TP301 [理论、方法];
学科分类号
081202 ;
摘要
Self-organising neural models have the ability to provide a good representation of the input space. In particular the Growing Neural Gas (GNG) is a suitable model because of its flexibility, rapid adaptation and excellent quality of representation. However, this type of learning is time-consuming, especially for high-dimensional input data. Since real applications often work under time constraints, it is necessary to adapt the learning process in order to complete it in a predefined time. This paper proposes a Graphics Processing Unit (CPU) parallel implementation of the GNG with Compute Unified Device Architecture (CUDA). In contrast to existing algorithms, the proposed CPU implementation allows the acceleration of the learning process keeping a good quality of representation. Comparative experiments using iterative, parallel and hybrid implementations are carried out to demonstrate the effectiveness of CUDA implementation. The results show that GNG learning with the proposed implementation achieves a speed-up of 6x compared with the single-threaded CPU implementation. CPU implementation has also been applied to a real application with time constraints: acceleration of 3D scene reconstruction for egomotion, in order to validate the proposal. (C) 2012 Elsevier Inc. All rights reserved.
引用
收藏
页码:1361 / 1372
页数:12
相关论文
共 56 条
[1]  
Amdahl G. M., 1967, P APR 18 20 1967 SPR, P483, DOI [10.1145/1465482.1465560, DOI 10.1145/1465482.1465560]
[2]  
Angelova Anelia., 2007, Proceedings of IEEE Computer Society Conference on Computer Vision and Pattern Recognition, P1
[3]  
[Anonymous], IEEE COMPUTER MAY
[4]  
[Anonymous], 2011, GPU Computing Gems Emerald Edition. Applications of GPU Computing Series
[5]  
[Anonymous], 2016, Programming massively parallel processors: a hands-on approach
[6]  
[Anonymous], VIEW POINT INSENSITI
[7]  
[Anonymous], 2010, CUDA Programming Guide for CUDA Toolkit 3.2: NVIDIA Corporation
[8]  
[Anonymous], 2012, INT THREAD BUILD BLO
[9]  
[Anonymous], COMPUTER BENCHMARKS
[10]  
[Anonymous], P AS C COMP VIS