Mathematical Techniques In Multisensor Data Fusion

by David Lee Hall

Overview

This invaluable reference offers the most comprehensive introduction available to the concepts of multisensor data fusion. It introduces key algorithms, provides advice on their utilization, and raises issues associated with their implementation. With a diverse set of mathematical and heuristic techniques for combining data from multiple sources, the book shows how to implement a data fusion system, describes the process for algorithm selection, functional architectures and requirements for ancillary software, and illustrates man-machine interface requirements an database issues.

Product Details

ISBN-13:	9780890065587
Publisher:	Artech House, Incorporated
Publication date:	02/01/1992
Series:	Artech House Radar Library Series
Pages:	320
Product dimensions:	6.00(w) x 9.00(h) x 0.88(d)

Table of Contents

	Preface	xiii
Chapter 1	Introduction to Multisensor Data Fusion	1
1.1	Introduction	1
1.2	Fusion Applications	3
1.3	Sensors and Sensor Data	8
1.4	The Inference Hierarchy: Output Data	16
1.5	A Data Fusion Model	18
1.6	Benefits of Data Fusion	22
1.7	Architectural Concepts and Issues	27
1.8	Limitations of Data Fusion	32
Chapter 2	Introduction to the Joint Directors of Laboratories (JDL) Data Fusion Process Model and Taxonomy of Algorithms	37
2.1	Introduction to the JDL Data Fusion Processing Model	37
2.2	Level 1 Fusion Algorithms	42
2.2.1	Data Alignment	44
2.2.2	Data/Object Correlation	44
2.2.3	Object Position, Kinematic, and Attribute Estimation	45
2.2.4	Object Identity Estimation	47
2.3	Level 2 Fusion Algorithms	54
2.4	Level 3 Fusion Algorithms	57
2.5	Level 4 Fusion Algorithms	59
2.6	Level 5 Fusion Techniques	62
2.7	Ancillary Support Functions	65
2.8	Alternative Data Fusion Process Models	66
2.8.1	Dasarathy's Functional Model	66
2.8.2	Boyd's Decision Loop	67
2.8.3	Bedworth and O'Brien's Omnibus Process Model	68
2.8.4	TRIP Model	69
Chapter 3	Level 1 Processing: Data Association and Correlation	73
3.1	Introduction	73
3.2	Process Model for Correlation	78
3.3	Hypothesis Generation	80
3.3.1	Characterizing the Hypothesis Generation Problem	85
3.3.2	Overview of Hypothesis Generation Techniques	92
3.4	Hypothesis Evaluation	99
3.4.1	Characterizing the Hypothesis Evaluation Problem	101
3.4.2	Overview of Hypothesis Evaluation Techniques	105
3.5	Hypothesis Selection Techniques	109
3.5.1	Defining the Hypothesis Selection Space	112
3.5.2	Overview of Hypothesis Selection Techniques	116
Chapter 4	Level 1 Fusion: Kinematic and Attribute Estimation	129
4.1	Introduction	129
4.2	Overview of Estimation Techniques	132
4.2.1	System Models	133
4.2.2	Optimization Criteria	136
4.2.3	Optimization Approach	140
4.2.4	Processing Approach	143
4.3	Batch Estimation	144
4.3.1	Derivation of Weighted Least Squares Solution	144
4.3.2	Processing Flow	149
4.3.3	Batch Processing Implementation Issues	152
4.4	Sequential Estimation	153
4.4.1	Derivation of Sequential Weighted Least Squares Solution	154
4.4.2	Sequential Estimation Processing Flow	156
4.4.3	Sequential Processing Implementation Issues	159
4.4.4	The Alpha-Beta Filter	160
4.5	Covariance Error Estimation	163
4.6	Recent Developments in Estimation	166
Chapter 5	Identity Declaration	171
5.1	Identity Declaration and Pattern Recognition	171
5.2	Feature Extraction	178
5.3	Parametric Templates	185
5.4	Cluster Analysis Techniques	187
5.5	Adaptive Neural Networks	193
5.6	Physical Models	196
5.7	Knowledge-Based Methods	198
5.8	Hybrid Techniques	200
Chapter 6	Decision-Level Identity Fusion	205
6.1	Introduction	205
6.2	Classical Inference	209
6.3	Bayesian Inference	214
6.4	Dempster-Shafer's Method	220
6.5	Generalized Evidence Processing (GEP) Theory	229
6.6	Heuristic Methods for Identity Fusion	231
6.7	Implementation and Trade-Offs	234
6.7.1	Inference Accuracy and Performance	235
6.7.2	Computer Resource Requirements	236
6.7.3	A Priori Data Requirements	236
Chapter 7	Knowledge-Based Approaches	239
7.1	Brief Introduction to Artificial Intelligence	239
7.2	Overview of Expert Systems	245
7.2.1	Expert System Concept	245
7.2.2	The Inference Process	247
7.2.3	Forward and Backward Chaining	249
7.2.4	Knowledge Representation	250
7.2.5	Representing Uncertainty	253
7.2.6	Search Techniques	260
7.2.7	Architectures for Knowledge-Based Systems	263
7.3	Implementation of Expert Systems	266
7.3.1	Life-Cycle Development Model for Expert Systems	266
7.3.2	Knowledge Engineering	269
7.3.3	Test and Evaluation	272
7.3.4	Expert System Development Tools	275
7.4	Logical Templating Techniques	278
7.5	Bayes Belief Systems	283
7.6	Intelligent Agent Systems	285
Chapter 8	Level 4 Processing: Process Monitoring and Optimization	291
8.1	Introduction	291
8.2	Extending the Concept of Level 4 Processing	297
8.3	Techniques for Level 4 Processing	300
8.3.1	Sensor Management Functions	300
8.3.2	General Sensor Controls	302
8.3.3	Optimization of System Resources	305
8.3.4	Measures of Effectiveness and Performance	306
8.4	Auction-Based Methods	308
8.4.1	Market Components	309
8.4.2	Multiattribute Auctions	310
8.4.3	Multiattribute Auction Algorithms	311
8.5	Research Issues in Level 4 Processing	311
Chapter 9	Level 5: Cognitive Refinement and Human-Computer Interaction	315
9.1	Introduction	315
9.2	Cognitive Aspects of Situation Assessment	317
9.3	Individual Differences in Information Processing	320
9.4	Enabling HCI Technologies	320
9.4.1	Visual and Graphical Interfaces	321
9.4.2	Aural Interfaces and Natural Language Processing (NLP)	325
9.4.3	Haptic Interfaces	327
9.4.4	Gesture Recognition	328
9.4.5	Wearable Computers	329
9.5	Computer-Aided Situation Assessment	330
9.5.1	Computer-Aided Cognition	330
9.5.2	Utilization of Language Constructs	331
9.5.3	Areas for Research	334
9.6	An SBIR Multimode Experiment in Computer-Based Training	336
9.6.1	SBIR Objective	336
9.6.2	Experimental Design and Test Approach	337
9.6.3	CBT Implementation	338
9.6.4	Summary of Results	340
9.6.5	Implications for Data Fusion Systems	341
Chapter 10	Implementing Data Fusion Systems	345
10.1	Introduction	345
10.2	Requirements Analysis and Definition	349
10.3	Sensor Selection and Evaluation	351
10.4	Functional Allocation and Decomposition	356
10.5	Architecture Trade-Offs	358
10.6	Algorithm Selection	364
10.7	Database Definition	369
10.8	HCI Design	373
10.9	Software Implementation	377
10.10	Test and Evaluation	379
Chapter 11	Emerging Applications of Multisensor Data Fusion	385
11.1	Introduction	385
11.2	Survey of Military Applications	386
11.3	Emerging Nonmilitary Applications	392
11.3.1	Intelligent Monitoring of Complex Systems	393
11.3.2	Medical Applications	396
11.3.3	Law Enforcement	397
11.3.4	Nondestructive Testing (NDT)	398
11.3.5	Robotics	398
11.4	Commercial Off The Shelf (COTS) Tools	399
11.4.1	Survey of COTS Software	399
11.4.2	Special Purpose COTS Software	399
11.4.3	General Purpose Data Fusion Software	402
11.4.4	A Survey of Surveys	406
11.5	Perspectives and Comments	408
Chapter 12	Automated Information Management	415
12.1	Introduction	415
12.2	Initial Automated Information Manager: Automated Targeting Data Fusion	419
12.3	Automated Targeting Data Fusion: Structure and Flow	424
12.4	Automatic Information Needs Resolution Example: Automated Imagery Corroboration	433
12.4.1	Automated Image Corroboration Example	436
12.5	Automated Information Manager: Ubiquitous Utility	441
	About The Authors	445
	Index	447

Editorial Reviews

Research into how to combine data from multiple (and sometimes different) sensors to find out what is going on, has been split between military (battlefield surveillance, etc.) and non-military (robotics, etc.) concerns. The two groups don't speak to each other, and have developed different approaches, techniques, and terminology. Here is a distillation of the mathematics they both use for data- fusion reduced to the bare bones to be comprehensible to researchers and application engineers in many fields. Annotation c. Book News, Inc., Portland, OR (booknews.com)

Booknews