Access to SPIE eBooks is limited to subscribing institutions. Access is not available as part of an individual subscription. However, books can be purchased on SPIE.Org
Chapter 9:
Algorithm Acceleration Using Parallel Computing
Author(s): Qian Kemao
Published: 2013
DOI: 10.1117/3.1002080.ch9
In the previous chapters we have analyzed various types of fringe patterns with emphasis on accuracy, automaticity, and speed. In this chapter, algorithm acceleration is considered. There are two ways to accelerate an algorithm. The first way is considered "soft." For example, in the WFR2/WFF2 algorithm, multiple convolutions are required. If the convolutions are realized directly in the spatial domain, the WFR2/WFF2 is time consuming. In contrast, if the convolutions are realized in a separable manner or by FFTs, the computation is much faster. The second way is considered "hard." If an algorithm can be parallelized and then computed by parallel computing hardware, the computation is also much faster. This hard way is increasing in popularity because the parallel computing hardware is nowcomputationally powerful, easy to program, and affordable. In this chapter we will demonstrate the benefits provided by the MATLAB® parallel computing toolbox that enables users to interface with parallel computing hardware and achieve the speedup easily. The concept of parallel computing and its application to fringe pattern analysis will be briefly covered, and an example using the WFF2 is demonstrated. This short chapter is meant to increase awareness of parallel computing, rather than to introduce the hardware and programming details.
Online access to SPIE eBooks is limited to subscribing institutions.

Back to Top