Sensor Fusion and Its Applications

494 Pages·25.551 MB·English

by  Thomas C. (Ed.)

Sensor Fusion
and Its Applications
edited by
Dr. Ciza Thomas
SCIYO
Sensor Fusion and Its Applications
Edited by Dr. Ciza Thomas
Published by Sciyo
Janeza Trdine 9, 51000 Rijeka, Croatia
Copyright © 2010 Sciyo
All chapters are Open Access articles distributed under the Creative Commons Non Commercial Share
Alike Attribution 3.0 license, which permits to copy, distribute, transmit, and adapt the work in any
medium, so long as the original work is properly cited. After this work has been published by Sciyo,
authors have the right to republish it, in whole or part, in any publication of which they are the author,
and to make other personal use of the work. Any republication, referencing or personal use of the work
must explicitly identify the original source.
Statements and opinions expressed in the chapters are these of the individual contributors and
not necessarily those of the editors or publisher. No responsibility is accepted for the accuracy of
information contained in the published articles. The publisher assumes no responsibility for any
damage or injury to persons or property arising out of the use of any materials, instructions, methods
or ideas contained in the book.

Publishing Process Manager Jelena Marusic
Technical Editor Zeljko Debeljuh
Cover Designer Martina Sirotic
Image Copyright Olaru Radian-Alexandru, 2010. Used under license from Shutterstock.com
First published September 2010
Printed in India
A free online edition of this book is available at www.sciyo.com
Additional hard copies can be obtained from [email protected]
Sensor Fusion and Its Applications, Edited by Dr. Ciza Thomas
p. cm.
ISBN 978-953-307-101-5
SCIYO.COM
free online editions of Sciyo
Books, Journals and Videos can
WHERE KNOWLEDGE IS FREE be found at www.sciyo.com

Contents
Preface VII
Chapter 1 State Optimal Estimation for Nonstandard
Multi-sensor Information Fusion System 1
Jiongqi Wang and Haiyin Zhou
Chapter 2 Air traffic trajectories segmentation
based on time-series sensor data 31
José L. Guerrero, Jesús García and José M. Molina
Chapter 3 Distributed Compressed Sensing of Sensor Data 53
Vasanth Iyer
Chapter 4 Adaptive Kalman Filter for Navigation Sensor Fusion 65
Dah-Jing Jwo, Fong-Chi Chung and Tsu-Pin Weng
Chapter 5 Fusion of Images Recorded with Variable Illumination 91
Luis Nachtigall, Fernando Puente León and Ana Pérez Grassi
Chapter 6 Camera and laser robust integration
in engineering and architecture applications 115
Pablo Rodriguez-Gonzalvez, Diego Gonzalez-Aguilera and Javier Gomez-Lahoz
Chapter 7 Spatial Voting With Data Modeling 153
Holger Marcel Jaenisch, Ph.D., D.Sc.
Chapter 8 Hidden Markov Model as a Framework for Situational Awareness 179
Thyagaraju Damarla
Chapter 9 Multi-sensorial Active Perception for Indoor Environment Modeling 207
Luz Abril Torres-Méndez
Chapter 10 Mathematical Basis of Sensor Fusion
in Intrusion Detection Systems 225
Ciza Thomas and Balakrishnan Narayanaswamy
Chapter 11 Sensor Fusion for Position Estimation in Networked Systems 251
Giuseppe C. Calafiore, Luca Carlone and MingzhuWei
VI
Chapter 12 M2SIR: A Multi Modal Sequential Importance
Resampling Algorithm for Particle Filters 277
Thierry Chateau and Yann Goyat
Chapter 13 On passive emitter tracking in sensor networks 293
Regina Kaune, Darko Mušicki and Wolfgang Koch
Chapter 14 Fuzzy-Pattern-Classifier Based
Sensor Fusion for Machine Conditioning 319
Volker Lohweg and Uwe Mönks
Chapter 15 Feature extraction: techniques for
landmark based navigation system 347
Molaletsa Namoshe, Oduetse Matsebe and Nkgatho Tlale
Chapter 16 Sensor Data Fusion for Road
Obstacle Detection: A Validation Framework 375
Raphaël Labayrade, Mathias Perrollaz, Dominique Gruyer and Didier Aubert
Chapter 17 Biometrics Sensor Fusion 395
Dakshina Ranjan Kisku, Ajita Rattani,
Phalguni Gupta, Jamuna Kanta Sing and Massimo Tistarelli
Chapter 18 Fusion of Odometry and Visual Datas
To Localization a Mobile Robot 407
André M. Santana, Anderson A. S. Souza,
Luiz M. G. Gonçalves, Pablo J. Alsina and Adelardo A. D. Medeiros
Chapter 19 Probabilistic Mapping by Fusion of
Range-Finders Sensors and Odometry 423
Anderson Souza, Adelardo Medeiros, Luiz Gonçalves and André Santana
Chapter 20 Sensor fusion for electromagnetic stress
measurement and material characterisation 443
John W Wilson, Gui Yun Tian, Maxim Morozov and Abd Qubaa
Chapter 21 Iterative Multiscale Fusion and Night Vision
Colorization of Multispectral Images 455
Yufeng Zheng
Chapter 22 Super-Resolution Reconstruction by Image Fusion
and Application to Surveillance Videos Captured
by Small Unmanned Aircraft Systems 475
Qiang He and Richard R. Schultz
Preface
The idea of this book on Sensor fusion and its Applications comes as a response to the immense
interest and strong activities in the field of sensor fusion. Sensor fusion represents a topic of
interest from both theoretical and practical perspectives.
The technology of sensor fusion combines pieces of information coming from different
sources/sensors, resulting in an enhanced overall system performance with respect to
separate sensors/sources. Different sensor fusion methods have been developed in order to
optimize the overall system output in a variety of applications for which sensor fusion might
be useful: sensor networks, security, medical diagnosis, navigation, biometrics, environmental
monitoring, remote sensing, measurements, robotics, etc. Variety of techniques, architectures,
levels, etc., of sensor fusion enables to bring solutions in various areas of diverse disciplines.
This book aims to explore the latest practices and research works in the area of sensor fusion.
The book intends to provide a collection of novel ideas, theories, and solutions related to the
research areas in the field of sensor fusion. This book aims to satisfy the needs of researchers,
academics, and practitioners working in the field of sensor fusion. This book is a unique,
comprehensive, and up-to-date resource for sensor fusion systems designers. This book is
appropriate for use as an upper division undergraduate or graduate level text book. It should
also be of interest to researchers, who need to process and interpret the sensor data in most
scientific and engineering fields.
Initial chapters in this book provide a general overview of sensor fusion. The later chapters
focus mostly on the applications of sensor fusion. Much of this work has been published
in refereed journals and conference proceedings and these papers have been modified and
edited for content and style. With contributions from the world’s leading fusion researchers
and academicians, this book has 22 chapters covering the fundamental theory and cutting-
edge developments that are driving this field.
Several people have made valuable contributions to this book. All researchers who have
contributed to this book are kindly acknowledged: without them, this would not have
been possible. Jelena Marusic and the rest of the sciyo staff provided technical and editorial
assistance that improved the quality of this work.
Editor
Dr. Ciza Thomas
College of Engineering,
Trivandrum
India

State Optimal Estimation for Nonstandard Multi-sensor Information Fusion System 1
State Optimal Estimation for Nonstandard Multi-sensor Information
1
X Fusion System

Jiongqi Wang and Haiyin Zhou
State Optimal Estimation for Nonstandard
Multi-sensor Information Fusion System

Jiongqi Wang and Haiyin Zhou
National University of Defense Technology
China

1. Introduction
In the field of information fusion, state estimation is necessary1-3. The traditional state
estimation is a process to use statistics principle to estimate the target dynamical (or static)
state by using of measuring information including error from single measure system.
However, single measure system can’t give enough information to satisfy the system
requirement for target control, and is bad for the precision and solidity of state estimation.
Therefore, developing and researching information fusion estimation theory and method is
the only way to obtain state estimation with high precision and solidity.
The traditional estimation method for target state (parameter) can be traced back to the age
of Gauss; in 1975, Gauss presented least square estimation (LSE), which is then applied in
orbit determination for space target. In the end of 1950s, Kalman presented a linear filter
method, which is widely applied in target state estimation and can be taken as the recursion
of LSE4. At present, these two algorithms are the common algorithms in multi-sensor state
fusion estimation, which are respectively called as batch processing fusion algorithm and
sequential fusion algorithm.
The classical LSE is unbiased, consistent and effective as well as simple algorithm and easy
operation when being applied in standard multi sensor information fusion system (which is
the character with linear system state equation and measuring equation, uncorrelated plus
noise with zero mean)5. However, because of the difference of measuring principle and
character of sensor as well as measuring environment, in actual application, some
non-standard multi-sensor information fusion systems are often required to be treated,
which mainly are as follows:
1) Each system error, mixing error and random disturbed factor as well as each
nonlinear factor, uncertain factor (color noise) existing in multi sensor measuring
information6;
2) Uncertain and nonlinear factors existing in multi sensor fusion system model, which
is expressed in two aspects: one is much stronger sense, uncertain and nonlinear factors in
model structure and another is time-change and uncertain factor in model parameter7;
3) Relativity between system noise and measuring noise in dynamical system or
relativity among sub-filter estimation as well as uncertain for system parameter and
unknown covariance information8-9.
2 Sensor Fusion and Its Applications
Ignoring the above situations, the optimal estimation results cannot be obtained if still using number of parameters to be estimated, the treatment not only lowered the integration of
the traditional batch processing fusion algorithm or sequential fusion algorithm. So to estimation accuracy, but also increased the complexity of the computation of the matrix
research the optimal fusion estimation algorithm for non standard multi-sensor system with inversion. In addition, robust estimation theory and its research are designed to the problem
the above situations is very essential and significative10. of the incomplete computing of the abnormal value and the condition of systems affected by
In the next three sections, the research work in this chapter focuses on non-standard the large deviation13. A first order Gauss - Markov process is used to analyze and handle the
multi-sensor information fusion system respectively with nonlinear, uncertain and random noise in measurement information. However, most of these treatments and
correlated factor in actual multi-sensor system and then presents the corresponding researches are based on artificial experience and strong hypothesis, which are sometimes so
resolution methods. contrary to the actual situation that they can doubt the feasibility and credibility of the state
Firstly, the modeling method based on semi-parameter modeling is researched to solve state fusion estimation.
fusion estimation in nonstandard multi-sensor fusion system to eliminate and solve the The main reason for the failure of the solution of the above-mentioned problems is that there
nonlinear mixing error and uncertain factor existing in multi-sensor information and is no suitable uncertainty modeling method or a suitable mathematical model to describe
moreover to realize the optimal fusion estimation for the state. the non-linear mixed-error factors in the multi-sensor measurement information14.
Secondly, a multi-model fusion estimation methods respectively based on multi-model Parts of the linear model (or called) semi-parameter model can be used as a suitable
adaptive estimation and interacting multiple model fusion are researched to deal with mathematical model to describe the non-linear mixed-error factors in the measurement
nonlinear and time-change factors existing in multi-sensor fusion system and moreover to information 15. Semi-parametric model have both parametric and non-parametric
realize the optimal fusion estimation for the state. components. Its advantages are that it focused on the main part of (i.e. the parameter
Thirdly, self-adaptive optimal fusion estimation for non-standard multi-sensor dynamical component) the information but without neglecting the role of the interference terms
system is researched. Self-adaptive fusion estimation strategy is introduced to solve local (non-parametric component). Semi-parametric model is a set of tools for solving practical
dependency and system parameter uncertainty existed in multi-sensor dynamical system problems with a broad application prospects. On the one hand, it solves problems which are
and moreover to realize the optimal fusion estimation for the state. difficult for only parameter model or non-parametric model alone to solve, thus enhancing
the adaptability of the model; on the other, it overcome the issue of excessive loss of
information by the non-parametric method and describe practical problems closer to the real
2. Information Fusion Estimation of Nonstandard Multisensor Based on Semi and made fuller use of the information provided by data to eliminate or weaken the impact
parametric Modeling of the state fusion estimation accuracy caused by non-linear factors more effectively.
This section attempts to introduce the idea of semi-parametric modeling into the fusion state
From the perspective of parameter modeling, any system models generally consist of two estimation theory of the non-standard multi-sensor. It establishes non-standard multi-sensor
parts: deterministic model (It means that the physical model and the corresponding fusion state estimation model based on semi-parametric regression and its corresponding
parameters are determined) and non-deterministic model (It means that the physical models parameters and non-parametric algorithm. At the same time of determining the unknown
are determined but some parameter uncertainty, or physical models and parameters are not parameters, it can also distinguish between nonlinear factors and uncertainties or between
fully identified). In general case, the practical problems of information fusion can be system error and accidental error so as to enhance the state fusion estimation accuracy.
described approximately by means of parametric modeling, then to establish the compact
convergence of information processing model. Namely, the part of the systematic error of
measurement can be deduced or weaken through the establishment of the classic parametric 2.1 State Fusion Estimation Based on Mutual Iteration Semi-parametric Regression
regression model, but it cannot inhibit mixed errors not caused by parametric modeling and In terms of the optimal state fusion estimation of the multi-sensor fusion system integration,
uncertainty errors and other factors. Strictly speaking, the data-processing method of its main jobs are to determine the "measurement information" and the state of mapping
classical parametric regression cannot fundamentally solve the problem of uncertainty relationship to be assessed, to reveal statistical characteristics of measurement errors, and
factors11. Yet it is precisely multi-sensor measurement information in the mixed errors and then to reach the result to be consistent with the optimal state fusion of the project scene.
uncertainties that have a direct impact on the accuracy indicated by the model of The mapping matrix is determined by specific engineering and the model established by the
multi-sensor fusion system, then in turn will affect the state estimation accuracy to be physical background, having a clear expression generally. Therefore, the core task of the
estimated and computational efficiency. So, it is one of the most important parts to research multi-sensor consists in the statistical characteristics of the measurement error analysis. But
and resolve such error factors of uncertainty, and to establish a reasonable estimation in practice, the differences of sensor measuring principle and its properties often touch upon
method under the state fusion estimation. the existence of the observing system error and the non-standard multi-sensor data fusion
As for this problem, there are a large number of studies to obtain good results at present. system under the influence of nonlinear uncertain elements. Among them, the errors in
For instance, systematic error parameter model suitable for the engineering background is constant-value system or parameterized system are rather special but conventional system
established to deal with the system error in measurement information. error. For these systems, it is easy to deal with12. But in fact, some systematic errors,
Extended-dimensional vector is employed to directly turn systematic error into the problem non-linear uncertainties in particular, which occur in the multi-sensor information fusion
of the state fusion estimation under the standard form12. However, due to the increase of the