Hardware software partitioning using particle swarm optimization (pso) in image processing application
Loading...
Date
2018-06
Authors
Tan, Jia Zheng
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
In this project, car plate identification will be implemented in hardware-software
partitioning by using PSO algorithm. The framework for hardware-software partitioning
using PSO algorithm in MATLAB is developed. The performance between the solution
in hardware-software partitioning and the solution without partitioning is investigated.
The image processing’s formulas are verified in Visual Studio. Then the coding is written
in Verilog for hardware and C language for software to obtain the execution time and
resources consumption. Then the data will be processed in MATLAB using PSO
algorithm to determine the optimal result in partitioning. The PSO algorithm parameters
such as the number of iteration and number of particles are varied to obtain the optimum
value for the parameters. Three different constraints value, C=1022, C=681 and C=341
are take into consideration to generate an optimum solution. The solution for C=1022 use
55% of the total hardware resources (1362) in pure hardware. It is 1.11 times faster than
pure hardware and 1.45 times faster than pure software. The solution for C=681 use 49.7%
of the total hardware resources in pure hardware and it is 1.09 times faster than pure
hardware and 1.42 times faster than pure software. The solution for C=341 use 22.03%
of the total hardware resources in pure hardware and it is 1.05 times faster than pure
hardware and 1.36 times faster than pure software. Performance in hardware-software
partitioning is higher or better compare to pure hardware and pure software. Hardware-software partitioning has fast in processing speed and use less in hardware resources.
Future work of this project is to implement the hardware-software partitioning solution
in Altera DE1-SoC.