Essential Image Processing and GIS for Remote Sensing

Dr Jian Guo Liu, Senior Lecturer, Department of Earth Sciences and Engineering, Imperial College London, UK Dr Philippa? Mason, Teaching Associate, Department of Earth Sciences and Engineering, Imperial College London, UK

"This book will allow interpreters to approach their work with a wider and deeper understanding of what has happened to imagery before it lands on their desk or computer." (Society of Exploration Geophysicists, 1 August 2011) "The authors have described the key concepts and ideas with clarity and in a logical manner and have also included numerous relevant conceptual illustrations. The book contains twenty three chapters, all of which are well written." (IAPR Newsletter, 1 July 2011)

• Overview of the Book xvPart One Image Processing 11 Digital Image and Display 31.1 What is a digital image? 31.2 Digital image display 41.2.1 Monochromatic display 41.2.2 Tristimulus colour theory and RGB colour display 51.2.3 Pseudo colour display 71.3 Some key points 8Questions 82 Point Operations (Contrast Enhancement) 92.1 Histogram modification and lookup table 92.2 Linear contrast enhancement 112.2.1 Derivation of a linear function from two points 122.3 Logarithmic and exponential contrast enhancement 132.3.1 Logarithmic contrast enhancement 132.3.2 Exponential contrast enhancement 142.4 Histogram equalization 142.5 Histogram matching and Gaussian stretch 152.6 Balance contrast enhancement technique 162.6.1 *Derivation of coefficients, a, b and c for a BCET parabolic function 162.7 Clipping in contrast enhancement 182.8 Tips for interactive contrast enhancement 18Questions 193 Algebraic Operations (Multi-image Point Operations) 213.1 Image addition 213.2 Image subtraction (differencing) 223.3 Image multiplication 223.4 Image division (ratio) 243.5 Index derivation and supervised enhancement 263.5.1 Vegetation indices 273.5.2 Iron oxide ratio index 283.5.3 TM clay (hydrated) mineral ratio index 293.6 Standardization and logarithmic residual 293.7 Simulated reflectance 293.7.1 Analysis of solar radiation balance and simulated irradiance 293.7.2 Simulated spectral reflectance image 303.7.3 Calculation of weights 313.7.4 Example: ATM simulated reflectance colour composite 323.7.5 Comparison with ratio and logarithmic residual techniques 333.8 Summary 34Questions 354 Filtering and Neighbourhood Processing 374.1 Fourier transform: understanding filtering in image frequency 374.2 Concepts of convolution for image filtering 394.3 Low-pass filters (smoothing) 404.3.1 Gaussian filter 414.3.2 The k nearest mean filter 424.3.3 Median filter 424.3.4 Adaptive median filter 424.3.5 The k nearest median filter 434.3.6 Mode (majority) filter 434.3.7 Conditional smoothing filter 434.4 High-pass filters (edge enhancement) 444.4.1 Gradient filters 454.4.2 Laplacian filters 464.4.3 Edge-sharpening filters 474.5 Local contrast enhancement 484.6 *FFT selective and adaptive filtering 484.6.1 FFT selective filtering 494.6.2 FFT adaptive filtering 514.7 Summary 54Questions 545 RGB–IHS Transformation 575.1 Colour coordinate transformation 575.2 IHS decorrelation stretch 595.3 Direct decorrelation stretch technique 615.4 Hue RGB colour composites 635.5 *Derivation of RGB–IHS and IHS–RGB transformations based on 3D geometry of the RGB colour cube 655.5.1 Derivation of RGB–IHS Transformation 655.5.2 Derivation of IHS–RGB transformation 665.6 *Mathematical proof of DDS and its properties 675.6.1 Mathematical proof of DDS 675.6.2 The properties of DDS 685.7 Summary 70Questions 706 Image Fusion Techniques 716.1 RGB–IHS transformation as a tool for data fusion 716.2 Brovey transform (intensity modulation) 736.3 Smoothing-filter-based intensity modulation 736.3.1 The principle of SFIM 746.3.2 Merits and limitation of SFIM 756.4 Summary 76Questions 767 Principal Component Analysis 777.1 Principle of PCA 777.2 Principal component images and colour composition 807.3 Selective PCA for PC colour composition 827.3.1 Dimensionality and colour confusion reduction 827.3.2 Spectral contrast mapping 837.3.3 FPCS spectral contrast mapping 847.4 Decorrelation stretch 857.5 Physical-property-orientated coordinate transformation and tasselled cap transformation 857.6 Statistic methods for band selection 887.6.1 Review of Chavez et al.’s and Sheffield’s methods 887.6.2 Index of three-dimensionality 897.7 Remarks 89Questions 908 Image Classification 918.1 Approaches of statistical classification 918.1.1 Unsupervised classification 918.1.2 Supervised classification 918.1.3 Classification processing and implementation 928.1.4 Summary of classification approaches 928.2 Unsupervised classification (iterative clustering) 928.2.1 Iterative clustering algorithms 928.2.2 Feature space iterative clustering 938.2.3 Seed selection 948.2.4 Cluster splitting along PC1 958.3 Supervised classification 968.3.1 Generic algorithm of supervised classification 968.3.2 Spectral angle mapping classification 968.4 Decision rules: dissimilarity functions 978.4.1 Box classifier 978.4.2 Euclidean distance: simplified maximum likelihood 988.4.3 Maximum likelihood 988.4.4 *Optimal multiple point reassignment 988.5 Post-classification processing: smoothing and accuracy assessment 998.5.1 Class smoothing process 998.5.2 Classification accuracy assessment 1008.6 Summary 102Questions 1029 Image Geometric Operations 1059.1 Image geometric deformation 1059.1.1 Platform flight coordinates, sensor status and imaging geometry 1059.1.2 Earth rotation and curvature 1079.2 Polynomial deformation model and image warping co-registration 1089.2.1 Derivation of deformation model 1099.2.2 Pixel DN resampling 1109.3 GCP selection and automation 1119.3.1 Manual and semi-automatic GCP selection 1119.3.2 *Towards automatic GCP selection 1119.4 *Optical flow image co-registration to sub-pixel accuracy 1139.4.1 Basics of phase correlation 1139.4.2 Basic scheme of pixel-to-pixel image co-registration 1149.4.3 The median shift propagation technique 1159.4.4 Summary of the refined pixel-to-pixel image co-registration and assessment 1179.5 Summary 118Questions 11910 *Introduction to Interferometric Synthetic Aperture Radar Techniques 12110.1 The principle of a radar interferometer 12110.2 Radar interferogram and DEM 12310.3 Differential InSAR and deformation measurement 12510.4 Multi-temporal coherence image and random change detection 12710.5 Spatial decorrelation and ratio coherence technique 12910.6 Fringe smoothing filter 13210.7 Summary 132Questions 134Part Two Geographical Information Systems 13511 Geographical Information Systems 13711.1 Introduction 13711.2 Software tools 13811.3 GIS, cartography and thematic mapping 13811.4 Standards, interoperability and metadata 13911.5 GIS and the Internet 14012 Data Models and Structures 14112.1 Introducing spatial data in representing geographic features 14112.2 How are spatial data different from other digital data? 14112.3 Attributes and measurement scales 14212.4 Fundamental data structures 14312.5 Raster data 14312.5.1 Data quantization and storage 14312.5.2 Spatial variability 14512.5.3 Representing spatial relationships 14512.5.4 The effect of resolution 14612.5.5 Representing surfaces 14712.6 Vector data 14712.6.1 Representing logical relationships 14812.6.2 Extending the vector data model 15312.6.3 Representing surfaces 15512.7 Conversion between data models and structures 15712.7.1 Vector to raster conversion (rasterization) 15812.7.2 Raster to vector conversion (vectorization) 16012.8 Summary 161Questions 16213 Defining a Coordinate Space 16313.1 Introduction 16313.2 Datums and projections 16313.2.1 Describing and measuring the Earth 16413.2.2 Measuring height: the geoid 16513.2.3 Coordinate systems 16613.2.4 Datums 16613.2.5 Geometric distortions and projection models 16713.2.6 Major map projections 16913.2.7 Projection specification 17213.3 How coordinate information is stored and accessed 17313.4 Selecting appropriate coordinate systems 174Questions 17514 Operations 17714.1 Introducing operations on spatial data 17714.2 Map algebra concepts 17814.2.1 Working with null data 17814.2.2 Logical and conditional processing 17914.2.3 Other types of operator 17914.3 Local operations 18114.3.1 Primary operations 18114.3.2 Unary operations 18214.3.3 Binary operations 18414.3.4 N-ary operations 18514.4 Neighbourhood operations 18514.4.1 Local neighbourhood 18514.4.2 Extended neighbourhood 19114.5 Vector equivalents to raster map algebra 19214.6 Summary 194Questions 19515 Extracting Information from Point Data: Geostatistics 19715.1 Introduction 19715.2 Understanding the data 19815.2.1 Histograms 19815.2.2 Spatial autocorrelation 19815.2.3 Variograms 19915.2.4 Underlying trends and natural barriers 20015.3 Interpolation 20115.3.1 Selecting sample size 20115.3.2 Interpolation methods 20215.3.3 Deterministic interpolators 20215.3.4 Stochastic interpolators 20715.4 Summary 209Questions 20916 Representing and Exploiting Surfaces 21116.1 Introduction 21116.2 Sources and uses of surface data 21116.2.1 Digital elevation models 21116.2.2 Vector surfaces and objects 21416.2.3 Uses of surface data 21516.3 Visualizing surfaces 21516.3.1 Visualizing in two dimensions 21616.3.2 Visualizing in three dimensions 21816.4 Extracting surface parameters 22016.4.1 Slope: gradient and aspect 22016.4.2 Curvature 22216.4.3 Surface topology: drainage networks and watersheds 22516.4.4 Viewshed 22616.4.5 Calculating volume 22816.5 Summary 229Questions 22917 Decision Support and Uncertainty 23117.1 Introduction 23117.2 Decision support 23117.3 Uncertainty 23217.3.1 Criterion uncertainty 23317.3.2 Threshold uncertainty 23317.3.3 Decision rule uncertainty 23417.4 Risk and hazard 23417.5 Dealing with uncertainty in spatial analysis 23517.5.1 Error assessment (criterion uncertainty) 23517.5.2 Fuzzy membership (threshold uncertainty) 23617.5.3 Multi-criteria decision making (decision rule uncertainty) 23617.5.4 Error propagation and sensitivity analysis (decision rule uncertainty) 23717.5.5 Result validation (decision rule uncertainty) 23817.6 Summary 239Questions 23918 Complex Problems and Multi-Criteria Evaluation 24118.1 Introduction 24118.2 Different approaches and models 24218.2.1 Knowledge-driven approach (conceptual) 24218.2.2 Data-driven approach (empirical) 24218.2.3 Data-driven approach (neural network) 24318.3 Evaluation criteria 24318.4 Deriving weighting coefficients 24418.4.1 Rating 24418.4.2 Ranking 24518.4.3 Pairwise comparison 24518.5 Multi-criteria combination methods 24818.5.1 Boolean logical combination 24818.5.2 Index-overlay and algebraic combination 24818.5.3 Weights of evidence modelling based on Bayesian probability theory 24918.5.4 Belief and Dempster–Shafer theory 25118.5.5 Weighted factors in linear combination 25218.5.6 Fuzzy logic 25418.5.7 Vectorial fuzzy modelling 25618.6 Summary 258Questions 258Part Three Remote Sensing Applications 25919 Image Processing and GIS Operation Strategy 26119.1 General image processing strategy 26219.1.1 Preparation of basic working dataset 26319.1.2 Image processing 26619.1.3 Image interpretation and map composition 27019.2 Remote-sensing-based GIS projects: from images to thematic mapping 27119.3 An example of thematic mapping based on optimal visualization and interpretation of multi-spectral satellite imagery 27219.3.1 Background information 27219.3.2 Image enhancement for visual observation 27419.3.3 Data capture and image interpretation 27419.3.4 Map composition 27819.4 Summary 279Questions 28020 Thematic Teaching Case Studies in SE Spain 28120.1 Thematic information extraction (1): gypsum natural outcrop mapping and quarry change assessment 28120.1.1 Data preparation and general visualization 28120.1.2 Gypsum enhancement and extraction based on spectral analysis 28320.1.3 Gypsum quarry changes during 1984–2000 28420.1.4 Summary of the case study 28720.2 Thematic information extraction (2): spectral enhancement and mineral mapping of epithermal gold alteration, and iron ore deposits in ferroan dolomite 28720.2.1 Image datasets and data preparation 28720.2.2 ASTER image processing and analysis for regional prospectivity 28820.2.3 ATM image processing and analysis for target extraction 29220.2.4 Summary 29620.3 Remote sensing and GIS: evaluating vegetation and land-use change in the Nijar Basin, SE Spain 29620.3.1 Introduction 29620.3.2 Data preparation 29720.3.3 Highlighting vegetation 29820.3.4 Highlighting plastic greenhouses 30020.3.5 Identifying change between different dates of observation 30220.3.6 Summary 30420.4 Applied remote sensing and GIS: a combined interpretive tool for regional tectonics, drainage and water resources 30420.4.1 Introduction 30420.4.2 Geological and hydrological setting 30520.4.3 Case study objectives 30620.4.4 Land use and vegetation 30720.4.5 Lithological enhancement and discrimination 31020.4.6 Structural enhancement and interpretation 31320.4.7 Summary 318Questions 320References 32121 Research Case Studies 32321.1 Vegetation change in the three parallel rivers region, Yunnan province, China 32321.1.1 Introduction 32321.1.2 The study area and data 32421.1.3 Methodology 32421.1.4 Data processing 32621.1.5 Interpretation of regional vegetation changes 32821.1.6 Summary 33221.2 Landslide hazard assessment in the three gorges area of the Yangtze river using ASTER imagery: Wushan–Badong–Zogui 33421.2.1 Introduction 33421.2.2 The study area 33421.2.3 Methodology: multi-variable elimination and characterization 33621.2.4 Terrestrial information extraction 33921.2.5 DEM and topographic information extraction 34421.2.6 Landslide hazard mapping 34721.2.7 Summary 34921.3 Predicting landslides using fuzzy geohazard mapping; an example from Piemonte, North-west Italy 35021.3.1 Introduction 35021.3.2 The study area 35221.3.3 A holistic GIS-based approach to landslide hazard assessment 35421.3.4 Summary 35721.4 Land surface change detection in a desert area in Algeria using multi-temporal ERS SAR coherence images 35921.4.1 The study area 35921.4.2 Coherence image processing and evaluation 36021.4.3 Image visualization and interpretation for change detection 36121.4.4 Summary 366Questions 366References 36622 Industrial Case Studies 37122.1 Multi-criteria assessment of mineral prospectivity, in SE Greenland 37122.1.1 Introduction and objectives 37122.1.2 Area description 37222.1.3 Litho-tectonic context – why the project’s concept works 37322.1.4 Mineral deposit types evaluated 37422.1.5 Data preparation 37422.1.6 Multi-criteria spatial modelling 38122.1.7 Summary 384Acknowledgements 38622.2 Water resource exploration in Somalia 38622.2.1 Introduction 38622.2.2 Data preparation 38722.2.3 Preliminary geological enhancements and target area identification 38822.2.4 Discrimination potential aquifer lithologies using ASTER spectral indices 39022.2.5 Summary 397Questions 397References 397Part Four Summary 39923 Concluding Remarks 40123.1 Image processing 40123.2 Geographical information systems 40423.3 Final remarks 407Appendix A: Imaging Sensor Systems and Remote Sensing Satellites 409A.1 Multi-spectral sensing 409A.2 Broadband multi-spectral sensors 413A.2.1 Digital camera 413A.2.2 Across-track mechanical scanner 414A.2.3 Along-track push-broom scanner 415A.3 Thermal sensing and thermal infrared sensors 416A.4 Hyperspectral sensors (imaging spectrometers) 417A.5 Passive microwave sensors 418A.6 Active sensing: SAR imaging systems 419Appendix B: Online Resources for Information, Software and Data 425B.1 Software – proprietary, low cost and free (shareware) 425B.2 Information and technical information on standards, best practice, formats, techniques and various publications 426B.3 Data sources including online satellite imagery from major suppliers, DEM data plus GIS maps and data of all kinds 426References 429General references 429Image processing 429GIS 430Remote sensing 430Part One References and further reading 430Part Two References and further reading 433Index 437