Pattern-based compression of multi-band image data for landscape analysis

We offer here a non-conventional approach to muhivariate ima- structured data for which the basis is well tested but the analytical ramifi cations are still unfolding. Although we do not formally pursue them, there are several parallels with the nature of neural networks. We employ a systematic set of statistical heuristics for modeling multivariate image data in a quasi-perceptual manner. When the human eye perceives a scene, the elements of the scene are segregated heuristically into compo nents according to similarity and dissimilarity, and then the relationships among the components are interpreted. Similarly, we segregate or seg ment the scene into hierarchically organized components that are subject to subsequent statistical analysis in many modes for interpretive purposes. We refer to the segregated scene segments as patterns, since they provide a basis for perception of pattern. Since they are also hierarchically organ ized, we refer to them further as polypatterns. This leads us to our acro nym of Progressively Segmented Image Modeling As Poly-Patterns (PSIMAPP). Likewise, we formalize our approach in terms of pattern processes and segmentation sequences. In alignment with the terminology of image analysis, we refer to our multivariate measures as being signal bands.

Author Notes

Dr. Wayne L. Myers earned M.F. and Ph.D. degrees in forest ecology and forest entomology at the University of Michigan. He began his professional career in Canada as a research forest entomologist and biometrician. He then joined the faculty of forestry at Michigan State University specializing in biometrics and remote sensing. The position at Michigan State also encompassed consultancies with the U.S. Forest Service and a work in Brazil. He moved to Penn State University in 1978 in the School of Forest Resources. He is professor of forest biometrics and Director of the Office for Remote Sensing and Spatial Information Resources (ORSSIR) in the Penn State Institutes of Environment.

He has thirty-five years of experience in research on development of remote sensing, geographic information systems, and related spatial technologies with applications focusing on natural resources and environment. This extends back to participation as a co-investigator in early investigations of ERTS/LANDSAT as the first spaceborne civilian multispectral sensor.

His recent research has focused on dual level progressive segmentation of multispectral images for purposes of compression, integration with geographic information systems and pattern-based change detection. He has developed concepts and computation of echelons of spatial structure in digital surfaces that facilitate extracting major change features from change indicator images. Echelons offer alternatives to thresholding in surface or pseudo-surface rasters. Dome domains provide a further generalization of topological structure in signal surfaces.

He has extensive international experience including long-term advisory for the U.S. Agency for International Development in India and research fellowships in Malaysia. He has placed special emphasis on interdisciplinary research and team approach.

G.P. Patil: is Distinguished Professor of Mathematical and Environmental Statistics in the Department of Statistics at the Pennsylvania State University, and is a former Visiting Professor of Biostatistics at Harvard University in the Harvard School of Public Health.

He has a Ph.D. in Mathematics, D.Sc. in Statistics, one Honorary Degree in Biological Sciences, and another in Letters. GP is a Fellow of American Statistical Association, Fellow of American Association of Advancement of Science, Fellow of Institute of Mathematical Statistics, Elected Member of the International Statistical Institute, Founder Fellow of the National Institute of Ecology and the Society for Medical Statistics in India.

GP has been a founder of Statistical Ecology Section of International Association for Ecology and Ecological Society of America, a founder of Statistics and Environment Section of American Statistical Association, and a founder of the International Society for Risk Analysis. He is founding editor-in-chief of the international journal, Environmental and Ecological Statistics and founding director of the Penn State Center for Statistical Ecology and Environmental Statistics. He has published thirty volumes and three hundred research papers. GP has received several distinguished awards which include: Distinguished Statistical Ecologist Award of the International Association for Ecology, Distinguished Achievement Medal for Statistics and the Environment of the American Statistical Association, Distinguished Twentieth Century Service Award for Statistical Ecology and Environmental Statistics of the Ninth Lukacs Symposium, Best Paper Award of the American Fisheries Society, and lately, the Best Paper Award of the American Water Resources Association, among others.

Currently, GP is principal investigator of a multi-year NSF grant for surveillance geoinformatics for hotspot detection and prioritization across geographic regions and networks for digital government in the 21st Century.

1 Innovative Imaging, Parsing Patterns and Motivating Models	p. 1
1.1 Image Introductory	p. 2
1.2 Satellite Sensing Scenario	p. 9
1.3 Innovative Imaging of Ecological and Environmental Indicators	p. 11
1.4 Georeferencing and Formatting Image Data	p. 16
1.5 The 4CS Pattern Perspective On Image Modeling	p. 18
References	p. 21
2 Pattern Progressions and Segmentation Sequences for Image Intensity Modeling and Grouped Enhancement	p. 23
2.1 Pattern Process, Progression, Prominence and Potentials	p. 23
2.2 Polypatterns	p. 25
2.3 Pattern Pictures, Ordered Overtones and Mosaic Models of Images	p. 26
2.4 Pattern Processes for Image Compression by Mosaic Modeling	p. 29
2.5 [alpha]-Scenario Starting Stages	p. 31
2.6 [alpha]-Scenario Splitting Stage	p. 32
2.7 [alpha]-Scenario Shifting Stage	p. 33
2.8 [beta]-Scenario Starting Stages	p. 36
2.9 [beta]-Scenario Splitting Stage	p. 37
2.10 Tree Topology and Level Loss	p. 39
2.11 [gamma]-Scenario for Parallel Processing	p. 40
2.12 Regional Restoration	p. 42
2.13 Relative Residuals	p. 42
2.14 Pictorial Presentation and Grouped Versus Global Enhancement	p. 47
2.15 Practicalities of Pattern Packages	p. 47
References	p. 48
3 Collective and Composite Contrast for Pattern Pictures	p. 51
3.1 Indirect Imaging by Tabular Transfer	p. 51
3.2 Characteristics of Colors	p. 53
3.3 Collective Contrast	p. 54
3.4 Integrative Image Indicators	p. 55
3.5 Composite Contrast for Pattern Pictures	p. 60
3.6 Tailored Transfer Tables	p. 61
References	p. 62
4 Content Classification and Thematic Transforms	p. 63
4.1 Interpretive Identification	p. 64
4.2 Thematic Transforms	p. 67
4.3 Algorithmic Assignments	p. 69
4.4 Adaptive Assignment Advisor	p. 70
4.5 Mixed Mapping Methods	p. 75
References	p. 78
5 Comparative Change and Pattern Perturbation	p. 79
5.1 Method of Multiple Mappings	p. 80
5.2 Compositing Companion Images	p. 81
5.3 Direct Difference Detection	p. 82
5.4 Pattern Perturbation	p. 87
5.5 Integrating Indicators	p. 90
5.6 Spanning Three or More Dates	p. 92
References	p. 95
6 Conjunctive Context	p. 99
6.1 Direct Detrending	p. 99
6.2 Echelons of Explicit Spatial Structure	p. 103
6.3 Disposition and Situation	p. 106
6.4 Joint Disposition	p. 106
6.5 Edge Affinities	p. 109
6.6 Patch Patterns and Generations of Generalization	p. 114
6.7 Parquet Polypattern Profiles	p. 115
6.8 Conformant/Comparative Contexts and Segment Signal Sequences	p. 117
6.9 Principal Properties of Patterns	p. 125
References	p. 128
7 Advanced Aspects and Anticipated Applications	p. 129
7.1 Advantageous Alternative Approaches	p. 129
7.2 Structural Sectors of Signal Step Surfaces	p. 131
7.3 Thematic Tracking	p. 133
7.4 Compositional Components	p. 134
7.5 Scale and Scope	p. 136
References	p. 136
Appendix A Public Packages for Portraying Polypatterns	p. 139
A.1 MultiSpec for Multiband Images and Ordered Overtones	p. 139
A.2 ArcExplorer	p. 147
Appendix B [alpha]-Scenario with PSIMAPP Software	p. 149
B.1 Polypatterns from Pixels	p. 151
B.2 Supplementary Statistics	p. 153
B.3 Collective Contrast	p. 153
B.4 Tonal Transfer Tables	p. 156
B.5 Combinatorial Contrast	p. 159
B.6 Regional Restoration	p. 160
B.7 Relative Residuals	p. 161
B.8 Direct Differences	p. 163
B.9 Detecting Changes from Perturbed Patterns	p. 165
B.10 Edge Expression	p. 167
B.11 Covariance Characteristics	p. 168
Appendix C Details of Directives for PSIMAPP Modules	p. 171
Glossary	p. 175
Index	p. 177

Available:*

On Order

Summary

Summary

Author Notes

Table of Contents