Parallel Implementation of Nonparametric Clustering Algorithm HCA-MS on GPU Using CUDA

Autor:	S. A. Rylov
Rok vydání:	2019
Předmět:	CUDA Speedup Computer science Computation ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION Mean-shift Image segmentation Parallel computing General-purpose computing on graphics processing units Cluster analysis Grid
Zdroj:	Communications in Computer and Information Science ISBN: 9783030122027
DOI:	10.1007/978-3-030-12203-4_19
Popis:	The present work explores nonparametric clustering algorithm HCA-MS. The combination of grid-based approach and Mean shift procedure allows the algorithm to discover arbitrary shaped clusters and to process large datasets, such as images. Parallel implementation of the HCA-MS algorithm on NVIDIA GPU using CUDA platform is presented. Provided experimental results on model data and multispectral images confirm the efficiency of the considered algorithm and its parallel implementation. The computation speedup on images was shown to be about 20x compared to 4 core CPU.
Databáze:	OpenAIRE
Externí odkaz:	https://explore.openaire.eu/search/publication?articleId=doi_________::a39664e685b7b4a364a3b98077adfab2 https://doi.org/10.1007/978-3-030-12203-4_19 Zobrazit plný text záznamu