View in EDS

Intensification of research work using images processing by application of parallel filtering on multi-core architectures.

Saved in:
Bibliographic Details
Title: Intensification of research work using images processing by application of parallel filtering on multi-core architectures.
Authors: Bosakova-Ardenska, Atanaska, Andreeva, Hristina
Source: AIP Conference Proceedings; 2024, Vol. 3063 Issue 1, p1-13, 13p
Subject Terms: PARALLEL processing, TIME complexity, MESSAGE passing (Computer science), IMAGE denoising, COMPUTATIONAL complexity, PARALLEL algorithms, MULTICORE processors
Abstract: As an essential operation filtering is a well-known technique for images smoothing and denoising on the level of primary images processing. Due to its computational complexity there are many approaches for decreasing its time complexity and the most often used one is the parallel processing. This paper presents a potential to improve effectiveness of images filtering by usage of computer system with multi-core architecture and message passing interface (MPI) library for communication among parallel processes. Two parallel algorithms for linear filtering are implemented and tested with two sets of experimental images. First algorithm is based on classic mean filter, but second algorithm uses partial sums in order to accelerate performance. It is observed that increasing number of parallel processes, decrease measured processing time when the number of parallel processes is not bigger than number of computational cores. Time performance of filtering process depends on size of used mask (kernel) also and increasing of mask size leads to significant increase of processing time. This effect is weaker presented when the algorithm for filtering uses partial sums. The experimental results based on processing on three different computer systems indicates that the best speed-up is achieved when the number of parallel processes is equal to number of computational cores and its value is about 4. [ABSTRACT FROM AUTHOR]
Copyright of AIP Conference Proceedings is the property of American Institute of Physics and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database: Complementary Index
Description
Abstract:As an essential operation filtering is a well-known technique for images smoothing and denoising on the level of primary images processing. Due to its computational complexity there are many approaches for decreasing its time complexity and the most often used one is the parallel processing. This paper presents a potential to improve effectiveness of images filtering by usage of computer system with multi-core architecture and message passing interface (MPI) library for communication among parallel processes. Two parallel algorithms for linear filtering are implemented and tested with two sets of experimental images. First algorithm is based on classic mean filter, but second algorithm uses partial sums in order to accelerate performance. It is observed that increasing number of parallel processes, decrease measured processing time when the number of parallel processes is not bigger than number of computational cores. Time performance of filtering process depends on size of used mask (kernel) also and increasing of mask size leads to significant increase of processing time. This effect is weaker presented when the algorithm for filtering uses partial sums. The experimental results based on processing on three different computer systems indicates that the best speed-up is achieved when the number of parallel processes is equal to number of computational cores and its value is about 4. [ABSTRACT FROM AUTHOR]
ISSN:0094243X
DOI:10.1063/5.0195739