Parallel Image Processing

Springer-Verlag Berlin Heidelberg GmbH

 T. Braunl
with S. Feyrer · W. Rapf · M. Reinhardt

Parallel
Image Processing

With 114 Figures and 7 Tables

~Springe
Prof. Dr. Thomas Braunl Dipl.-Inform. Wolfgang Rapf
The University ofWestern Australia GAOmbH
Department of Electrical and Postfach 700703
Electronic Engineering 81307 Miinchen
Nedlands [email protected]
WA 6907 Perth, Australia
[email protected]

Dipl.-Inform. Stefan Feyrer Dipl.-Inform. Michael Reinhardt

University of Tiibingen, WSI ProSieben Information Service GmbH
Kostlinstr. 6 Gutenbergstr. 3
72074 Tiibingen, Germany 85767 Unterfohring, Germany
[email protected] [email protected]

Library of Congress Cataloging-in-Publication Data

Parallel image processing I Thomas Brăunl ... [et al.].
p.cm.
Includes bibliographical references and index.
ISBN 978-3-642-08679-3 ISBN 978-3-662-04327-1 (eBook)
DOI 10.1007/978-3-662-04327-1
Image processing-Digital techniques. 2. Parallel processing (Electronic computers) I. Brăunl,
1. Thomas.
TA1637.P355 2000 621.36'7-dc21 00-064129

ISBN 978-3-642-08679-3

This work is subject to copyright. All rights are reserved, whether the whole or part of the material
is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broad-
casting, reproduction on microfilm or in any other way, and storage in data banks. Duplication of this
publication or parts thereof is permitted only under the provisions of the German Copyright Law
of September 9, 1965, in its current version, and permission for use must always be obtained from
Springer-Verlag. Violations are liable for prosecution under the German Copyright Law.
© Springer-Verlag Berlin Heidelberg 2001
Originally published by Springer-Verlag Berlin Heidelberg New York in 2001
Softcover reprint of the hardcover 1st edition 2001

The use of general descriptive names, registered names, trademarks, etc. in this publication does not
imply, even in the absence of a specific statement, that such names are exempt from the relevant protec-
tive laws and regulations and therefore free for general use.
Typesetting: Camera-ready copy produced from the author's output file using a Springer TEX macro
package
Cover design: Kiinkel + Lopka, Heidelberg
Printed on acid-free paper SPIN 10766747 41/3143/XO - 5 4 3 2 1 o
Preface

This book developed out of a series of publications in the area of image processing
with massively parallel algorithms. The topic of image processing is a particularly
promising area for the use of synchronous massively parallel or data-parallel compu-
ter systems which work according to the SIMD principle (single instruction, multiple
data). While the era of large SIMD super-computers has passed, SIMD systems have
come back as dedicated vision subsystems and will soon be found even in embedded
systems.
In comparison to conventional sequential implementations of basic image opera-
tions, this book illustrates the intrinsic parallelism which is almost always present in
image processing. By utilising parallel algorithms it is even possible to illustrate oper-
ations in a simpler and easier to understand way than for the sequential case.
The presentation method chosen for this book assumes that short, terse excerpts of
program code will significantly enhance the understanding of the material, e.g. of
image operations, while longer listings are more likely to distract from the topic. For
this reason, each chapter will not only define and explain the central image processing
algorithms with the help of examples, but will also give an excerpt of a massively par-
allel program. For image processing this means that at least virtually there should be
one processor available for each pixel. The mapping onto a smaller number of exist-
ing real processors is done transparently by the compiler, and as such is not of interest
here.
All the algorithms implemented in this book were written in the data-parallel pro-
gramming language Parallaxis. An introduction to Parallaxis III, as well as its com-
plete language definition, is given in the appendix. Programs in Parallaxis can be
ported to any data-parallel system, or they can be simulated and debugged on work-
stations or personal computers. The complete Parallaxis package, i.e., programming
environment, language description, documentation, example programs and the paral-
lel image processing library, are public domain software and available free of charge
over the Internet:
http://www.ee.uwa.edu.au/-braunl/parallaxis
The Parallaxis-III system was implemented by Eduard Kappel, Hartmut Keller,
Harald Lampke, Jorg Stippa and Jtirgen Wakunda under the supervision of Thomas
Braunl. We wish to thank all of our students who have contributed in the implementa-
tion of Parallaxis-III, as well as Michael Juschke for translating the manuscript from
German to English.
Chapters l-4, 6-7, 12, and the Appendix were written by Thomas Braunl,
Chapter 5 on skeletonizing by Michael Reinhardt, Chapters 8 and 13 on comers and
VI Preface

image sequences by Wolfgang Rapf, and Chapters 9-11 on transformations and tex-
ture by Stefan Feyrer.

Perth, November 2000 Thomas Briiunl

Stefan Feyrer
Wolfgang Rapf
Michael Reinhardt
Contents

1 Introduction
1.1 Image Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Parallel Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Data-Parallel Systems ............................................ 3
1.4 Communication and Vector Reduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.5 References ..................................................... 6

2 Point Operators
2.1 Image Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2 Conversion Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.3 Range Operations .............................................. 13
2.4 Histogram Operations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.5 References .................................................... 17

3 Local Operators
3.1 Parallel Positioning oflmage Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.2 Mean Value ................................................... 20
3.3 Median ...................................................... 22
3.4 Dithering ..................................................... 24
3.5 References .................................................... 26

4 Edge Detection
4.1 Laplace Operator............................................... 27
4.2 Edge Detection along Coordinate System Axes . . . . . . . . . . . . . . . . . . . . . . . 28
4.3 Sobel Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
4.4 References .................................................... 32

5 Skeletonizing
5.1 Introduction ................................................... 33
5.2 Requirements and Methods of Operation ............................ 34
5.3 Classification of Algorithms ...................................... 36
5.4 Naive Algorithm ............................................... 37
5.5 Definitions ................................................... 38
5.6 Algorithm of Stefanelli and Rosenfeld .............................. 40
5.7 Algorithm ofLii and Wang ....................................... 44
5.8 Algorithm of Hall and Guo ....................................... 46
5.9 Examples ..................................................... 48
5.10 References .................................................... 49
VIII Contents

6 Morphological Operators
6.1 Erosion and Dilation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
6.2 Open and Close ................................................ 53
6.3 Fill and Connected . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
6.4 Boundary and Skeleton. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
6.5 Other Morphological Methods .................................... 58
6.6 References .................................................... 58

7 Segmentation
7.1 Region Growing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
7.2 Split-and-Merge ............................................... 63
7.3 References .................................................... 63

8 Corner Detection
8.1 Introduction ................................................... 65
8.2 Extraction of the Contour Line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
8.3 Calculation of the Measure of Curvature ............................ 67
8.4 Localisation of Comers .......................................... 72
8.5 Functions for Comer Detection ................................... 77
8.6 References .................................................... 81

9 Hough Transform
9.1 Basic Idea of the Hough Transform ................................ 83
9.2 Complexity Considerations ....................................... 86
9.3 Approaches of Parallelisation ..................................... 87
9.4 Parallel Implementation ......................................... 88
9.5 Examples ..................................................... 95
9.6 References .................................................... 97

10 Fourier Transform
10.1 Two-Dimensional Discrete Fourier Transform ....................... 99
10.2 Algorithms for the Fast Fourier Transform ......................... 105
10.3 Parallel Implementation ........................................ 108
10.4 Fourier Spectrum ............................................. 113
10.5 Applications and Examples ..................................... 115
10.6 References ................................................... 118

11 Texture Recognition
11.1 Co-occurrence Matrices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
11.2 Parallel Implementation ........................................ 124
11.3 Examples .................................................... 128
11.4 References ................................................... 130
Contents IX

12 Stereo Image Processing

12.1 Random-Dot Stereo Images . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
12.2 Real Stereograms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
12.3 References ................................................... 143

13 Analysis of Image Sequences

13.1 Introduction .................................................. 145
13.2 Calculation of Displacement Vector Fields ......................... 147
13.3 Method by Hom and Schunck ................................... 151
13.4 Matching of Gray Value Comers ................................. 161
13.5 References ................................................... 168

Appendix
A The Parallel Programming Language Parallaxis ..................... 169
B Parallaxis-III Syntax ........................................... 183
C Programming Tools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
D Vision Library ................................................ 191
E References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200

Index ................................................... ·......... 201