Hyperspectral Unmixing on GPUs and Multi-Core Processors: A Comparison

Bernabé, Sergio; Sánchez, Sergio; Plaza, Antonio; López, S.; Benediktsson, Jón Atli; Sarmiento, Roberto

doi:10.1109/jstars.2013.2254470

Cited by 75 publications

(45 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The speedup values for this framework range from 3.3× to 6.4× if eight processors are used. The results on a multi-core system with 12 physical cores presented by [23] show that there is no significant difference between using 4 or 12 cores (Figure 11 in that paper), because the parallel implementation is not optimized to increase its scalability to a high number of cores. Our method can achieve high scalability evidenced by the results in Table 4.…”

Section: Comparing With Other Multi-core Parallel Techniquesmentioning

confidence: 92%

“…However, the parallel computing in the literature [22,23], which is built on the OpenMP, BLAS and LAPACK libraries supported by the compilers, exploits multi-threaded linear algebra subprograms and parallelized loops. Second, the latest computers containing more cores are used in this experiment.…”

Section: Comparing With Other Multi-core Parallel Techniquesmentioning

confidence: 99%

“…Wei et al [21] presented a parallel framework that optimizes the parameter determination process for various algorithms of image fusion on Linux clusters. Parallel experiments based on multi-core processors, which mainly use the OpenMP (an API for shared-memory parallel programming), basic linear algebra subprograms (BLAS), and linear algebra package (LAPACK) libraries, can be found in the literature [22,23].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

A Parallel Computing Paradigm for Pan-Sharpening Algorithms of Remotely Sensed Images on a Multi-Core Computer

2014

View full text Add to dashboard Cite

Pan-sharpening algorithms are data-and computation-intensive, and the processing performance can be poor if common serial processing techniques are adopted. This paper presents a parallel computing paradigm for pan-sharpening algorithms based on a generalized fusion model and parallel computing techniques. The developed modules, including eight typical pan-sharpening algorithms, show that the framework can be applied to implement most algorithms. The experiments demonstrate that if parallel strategies are adopted, in the best cases the fastest times required to finish the entire fusion operation (including disk input/output (I/O) and computation) are close to the time required to directly read and write the images without any computation. The parallel processing implemented on a workstation with two CPUs is able to perform these operations up to 13.9 times faster than serial execution. An algorithm in the framework is 32.6 times faster than the corresponding version in the ERDAS IMAGINE software. Additionally, no obvious differences in the fusion effects are observed between the fusion results of different implemented versions.

show abstract

Section: Comparing With Other Multi-core Parallel Techniquesmentioning

confidence: 92%

Section: Comparing With Other Multi-core Parallel Techniquesmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A Parallel Computing Paradigm for Pan-Sharpening Algorithms of Remotely Sensed Images on a Multi-Core Computer

2014

View full text Add to dashboard Cite

show abstract

“…CUDA Basic Linear Algebra Subroutines (cuBLAS) library provides the high-performance functions for the numerical operations and shows the order of magnitude performance improvements over the other libraries [23,24]. This library also contains various General Matrix-Matrix Multiply (GEMM) routines for the different types of operands (like complex, single data type, double data type etc.)…”

Section: Cublas Accelerated Matrix Multiplication Convolution (Convcamm)mentioning

confidence: 99%

Optimized Deep Neural Networks for Real-Time Object Classification on Embedded GPUs

2017

View full text Add to dashboard Cite

Abstract:Convolution is the most computationally intensive task of the Convolutional Neural Network (CNN). It requires a lot of memory storage and computational power. There are different approaches to compute the solution of convolution and reduce its computational complexity. In this paper, a matrix multiplication-based convolution (ConvMM) approach is fully parallelized using concurrent resources of GPU (Graphics Processing Unit) and optimized, considerably improving the performance of the image classifiers and making them applicable to real-time embedded applications. The flow of this CUDA (Compute Unified Device Architecture)-based scheme is optimized using unified memory and hardware-dependent acceleration of matrix multiplication. Proposed flow is evaluated on two different embedded platforms: first on an Nvidia Jetson TX1 embedded board and then on a Tegra K1 GPU of an Nvidia Shield K1 Tablet. The performance of this optimized and accelerated convolutional layer is compared with its sequential and heterogeneous versions. Results show that the proposed scheme significantly improves the overall results including energy efficiency, storage requirement and inference performance. In particular, the proposed scheme on embedded GPUs is hundreds of times faster than the sequential version and delivers tens of times higher performance than the heterogeneous approach.

show abstract

“…Thus, it is important to apply high-performance computing technologies to accelerate the hyperspectral image processing algorithms for the time-crucial scenarios [16][17][18]. CPU-GPU heterogeneous parallel mode is a tremendous potential to bridge the gap towards real-time analysis of hyperspectral image [19][20][21][22][23][24][25].…”

Section: Introductionmentioning

confidence: 99%

Optimization of minimum volume constrained hyperspectral image unmixing on CPU–GPU heterogeneous platform

et al. 2014

J Real-Time Image Proc

View full text Add to dashboard Cite

Hyperspectral unmixing is essential for efficient hyperspectral image processing. Nonnegative matrix factorization based on minimum volume constraint (MVC-NMF) is one of the most widely used methods for unsupervised unmixing for hyperspectral image without the pure-pixel assumption. But the model of MVC-NMF is unstable, and the traditional solution based on projected gradient algorithm (PG-MVC-NMF) converges slowly with low accuracy. In this paper, a novel parallel method is proposed for minimum volume constrained hyperspectral image unmixing on CPU-GPU Heterogeneous Platform. First, a optimized unmixing model of minimum logarithmic volume regularized NMF is introduced and solved based on the second-order approximation of function and alternating direction method of multipliers (SO-MVC-NMF). Then, the parallel algorithm for optimized MVC-NMF (PO-MVC-NMF) is proposed based on the CPU-GPU heterogeneous platform, taking advantage of the parallel processing capabilities of GPUs and logic control abilities of CPUs. Experimental results based on both simulated and real hyperspectral images indicate that the proposed algorithm is more accurate and robust than the traditional PG-MVC-NMF, and the total speedup of PO-MVC-NMF compared to PG-MVC-NMF is over 50 times.

show abstract

Hyperspectral Unmixing on GPUs and Multi-Core Processors: A Comparison

Cited by 75 publications

References 31 publications

A Parallel Computing Paradigm for Pan-Sharpening Algorithms of Remotely Sensed Images on a Multi-Core Computer

A Parallel Computing Paradigm for Pan-Sharpening Algorithms of Remotely Sensed Images on a Multi-Core Computer

Optimized Deep Neural Networks for Real-Time Object Classification on Embedded GPUs

Optimization of minimum volume constrained hyperspectral image unmixing on CPU–GPU heterogeneous platform

Contact Info

Product

Resources

About