Translator Disclaimer
Chapter 3:
Linear Image Processing Algorithms
Abstract
Real-time processing for digital imaging concerns efficient, deterministic implementation of algorithms whose inputs include digital images and whose outputs are digital images, numerical features, symbolic representations, or decisions. Computation bottlenecks appear in many forms and, to some extent, each requires its own real-time implementation via algorithm design, software, hardware, or a combination thereof. For example, the von Neumann bottleneck prevents the parallelization of certain algorithms on serial processors. Rather than present a myriad of individual approaches to specific real-time imaging problems, we shall focus on a small number of computation tasks that occur across various application domains. The present chapter introduces basic definitions, applications, and computations pertaining to linear processing, in particular, windowed convolution. The next two chapters discuss matrix transformations and nonlinear processing. The imaging operators discussed in Chapters 3, 4, and 5 provide a class of processing problems to which we can subsequently refer when we discuss software and hardware solutions to real-time computation.
Online access to SPIE eBooks is limited to subscribing institutions.
CHAPTER 3
30 PAGES


SHARE
Advertisement
Advertisement
Back to Top