ISSN : 2319-7323

INTERNATIONAL JOURNAL OF COMPUTER SCIENCE ENGINEERING

ABSTRACT

Title	:	Performance Evaluation of Basic Image Processing Algorithms in CPU, GPU, Raspberry Pi and FPGA
Authors	:	TAHAR ABBES Mounir, Selma Boumerdassi, Abdelhak Benhamada, Abdelmadjid Mhamed Allal, Mohamed Kherarba
Keywords	:	Image Processing; FPGA; CPU; GPU; Raspberry Pi.
Issue Date	:	Jul-Aug 2020
Abstract	:	This article presents a comparative and experimental study (benchmark) regarding the main execution time on diverse platforms (CPU, Graphics Processing Units (GPU), Raspberry, and The Field-Programmable Gate Arrays (FPGA)). We evaluate the time performance of standard image processing algorithms used in many vision applications. To carry out this experiment, we designed a mobile explorer robot equipped with an integrated CMOS camera. This robot has been programmed to perform a route avoiding obstacles using five image processing algorithms (re-size, erode, dilate, find contours, and distance calculation). Two levels of tests were carried out: The internal level check the efficiency of the application algorithms, and the external level measures the platform's output concerning execution time. Our results highlight the following remarks: GPU is advantageous in image processing algorithms, which process data or pixels independently. The CPU shows its power on sequential data; however, GPU is slower than CPU in those algorithms. The FPGA card's performance is ten times higher than CPU and GPU; it is possible to increase performance by processing in parallel.
Page(s)	:	312-325
ISSN	:	2319-7323
Source	:	Vol. 9, No. 4