Optimization Techniques for Solving Complex Problems

CHAPTER 18

Optimization of Image-Processing Algorithms Using FPGAs

M. A. VEGA

Universidad de Extremadura, Spain

A. GÓMEZ

Centro Extremeño de Tecnologias Avanzadas, Spain

J. A. GÓMEZ and J. M. SÁNCHEZ

Universidad de Extremadura, Spain

18.1 INTRODUCTION

At present, FPGAs (field-programmable gate arrays) are very popular devices in many different fields. In particular, FPGAs are a good alternative for many real applications in image processing. Several systems using programmable logic devices have been designed, showing the utility of these devices for artificial-vision applications [1]. Whereas other papers display the results of implementing image-processing techniques by means of “standard” FPGAs or reconfigurable computing systems [2]. In this chapter we present the implementation and optimization of 16 image-processing operations by means of reconfigurable hardware. Our FPGA implementations use several parallelism techniques, include other optimizations as we will see, and have high operation frequencies. Thanks to all this, they obtain very good results. In fact, our FPGA-based implementations are even more than 149 times faster than the corresponding software implementation.

The chapter is organized as follows. In Section 18.2 we give a brief background about the use of FPGAs for optimizing image-processing algorithms. In Section 18.3 we describe the main characteristics of ...