REMOTE SENSING DIGITAL IMAGE ANALYSIS
CONTENTS SOURCES AND CHARACTERISTICS OF REMOTE SENSING IMAGE DATA INTRODUCTION TO DATA SOURCES CHARACTERISTICS OF DIGITAL IMAGE DATA SPECTRAL RANGES COMMONLY USED IN REMOTE SENSING CONCLUDING REMARKS REMOTE SENSING PLATFORMS IMAGE DATA SOURCES IN THE MICROWAVE REGION SIDE LOOKING AIRBORNE RADAR AND SYNTHETIC APERTURE RADAR SPATIAL DATA SOURCES IN GENERAL TYPES OF SPATIAL DATA DATA FORMATS GEOGRAPHIC INFORMATION SYSTEMS (GIS) THE CHALLENGE TO IMAGE PROCESSING AND ANALYSIS A COMPARISON OF SCALES IN DIGITAL IMAGE DATA ERROR CORRECTION AND REGISTRATION OF IMAGE DATA SOURCES OF RADIOMETRIC DISTORTION THE EFFECT OF THE ATMOSPHERE ON RADIATION ATMOSPHERIC EFFECTS ON REMOTE SENSING IMAGERY INSTRUMENTATION ERRORS CORRECTION OF RADIOMETRIC DISTORTION DETAILED CORRECTION OF ATMOSPHERIC EFFECTS BULK CORRECTION OF ATMOSPHERIC EFFECTS CORRECTION OF INSTRUMENTATION ERRORS SOURCES OF GEOMETRIC DISTORTION EARTH ROTATION EFFECTS PANORAMIC DISTORTION EARTH CURVATURE SCAN TIME SKEW VARIATIONS IN PLATFORM ALTITUDE, VELOCITY AND ATTITUDE ASPECT RATIO DISTORTION SENSOR SCAN NONLINEARITIES CORRECTION OF GEOMETRIC DISTORTION USE OF MAPPING POLYNOMIALS FOR IMAGE CORRECTION MAPPING POLYNOMIALS AND GROUND CONTROL POINTS RESAMPLING INTERPOLATION CHOICE OF CONTROL POINTS EXAMPLE OF REGISTRATION TO A MAP GRID MATHEMATICAL MODELLING ASPECT RATIO CORRECTION EARTH ROTATION SKEW CORRECTION IMAGE ORIENTATION TO NORTH-SOUTH CORRECTION OF PANORAMIC EFFECTS COMBINING THE CORRECTIONS IMAGE REGISTRATION GEOREFERENCING AND GEOCODING IMAGE TO IMAGE REGISTRATION CONTROL POINT LOCALISATION BY CORRELATION EXAMPLE OF IMAGE TO IMAGE REGISTRATION MISCELLANEOUS IMAGE GEOMETRY OPERATIONS IMAGE ROTATION SCALE CHANGING AND ZOOMING THE INTERPRETATION OF DIGITAL IMAGE DATA APPROACHES TO INTERPRETATION FORMS OF IMAGERY FOR PHOTOINTERPRETATION COMPUTER PROCESSING FOR PHOTOINTERPRETATION AN LNTRODUCTION TO QUANTITATIVE ANALYSIS – CLASSI?CATION MULTISPECTRAL SPACE AND SPECTRAL CLASSES QUANTITATIVE ANALYSIS BY PATTERN RECOGNITION PIXEL VECTORS AND LABELLING UNSUPERVISED CLASSI?CATION SUPERVISED CLASSI?CATION RADIOMETRIC ENHANCEMENT TECHNIQUES INTRODUCTION POINT OPERATIONS AND LOOK UP TABLES SCALAR AND VECTOR IMAGES THE IMAGE HISTOGRAM CONTRAST MODI?CATION IN IMAGE DATA HISTOGRAM MODI?CATION RULE LINEAR CONTRAST MODI?CATION SATURATING LINEAR CONTRAST ENHANCEMENT AUTOMATIC CONTRAST ENHANCEMENT LOGARITHMIC AND EXPONENTIAL CONTRAST ENHANCEMENT PIECEWISE LINEAR CONTRAST MODI?CATION HISTOGRAM EQUALIZATION USE OF THE CUMULATIVE HISTOGRAM ANOMALIES IN HISTOGRAM EQUALIZATION HISTOGRAM MATCHING PRINCIPLE OF HISTOGRAM MATCHING IMAGE TO IMAGE CONTRAST MATCHING MATCHING TO A MATHEMATICAL REFERENCE DENSITY SLICING BLACK AND WHITE DENSITY SLICING COLOUR DENSITY SLICING AND PSEUDOCOLOURING GEOMETRIC ENHANCEMENT USING IMAGE DOMAIN TECHNIQUES NEIGHBOURHOOD OPERATIONS TEMPLATE OPERATORS GEOMETRIC ENHANCEMENT AS A CONVOLUTION OPERATION IMAGE DOMAIN VERSUS FOURIER TRANSFORMATION APPROACHES IMAGE SMOOTHING (LOW PASS FILTERING) MEAN VALUE SMOOTHING MEDIAN FILTERING EDGE DETECTION AND ENHANCEMENT LINEAR EDGE DETECTING TEMPLATES SPATIAL DERIVATIVE TECHNIQUES THE ROBERTS OPERATOR THE SOBEL OPERATOR THE PREWITT OPERATOR THINNING, LINKING AND BORDER RESPONSES EDGE ENHANCEMENT BY SUBTRACTIVE SMOOTHING (SHARPENING) LINE DETECTION LINEAR LINE DETECTING TEMPLATES NON-LINEAR AND SEMI-LINEAR LINE DETECTING TEMPLATES GENERAL CONVOLUTION FILTERING DETECTING GEOMETRIC PROPERTIES TEXTURE SPATIAL CORRELATION – THE SEMIVARIOGRAM SHAPE DETECTION MULTISPECTRAL TRANSFORMATIONS OF IMAGE DATA THE PRINCIPAL COMPONENTS TRANSFORMATION THE MEAN VECTOR AND COVARIANCE MATRIX A ZERO CORRELATION, ROTATIONAL TRANSFORM EXAMPLES – SOME PRACTICAL CONSIDERATIONS THE EFFECT OF AN ORIGIN SHIFT APPLICATION OF PRINCIPAL COMPONENTS IN IMAGE ENHANCEMENT AND DISPLAY THE TAYLOR METHOD OF CONTRAST ENHANCEMENT OTHER APPLICATIONS OF PRINCIPAL COMPONENTS ANALYSIS NOISE ADJUSTED PRINCIPAL COMPONENTS TRANSFORMATION THE KAUTH-THOMAS TASSELED CAP TRANSFORMATION IMAGE ARITHMETIC, BAND RATIOS AND VEGETATION INDICES FOURIER TRANSFORMATION OF IMAGE DATA SPECIAL FUNCTIONS THE COMPLEX EXPONENTIAL FUNCTION THE DIRAC DELTA FUNCTION PROPERTIES OF THE DELTA FUNCTION THE HEAVISIDE STEP FUNCTION FOURIER SERIES THE FOURIER TRANSFORM CONVOLUTION THE CONVOLUTION INTEGRAL CONVOLUTION WITH AN IMPULSE THE CONVOLUTION THEOREM SAMPLING THEORY THE DISCRETE FOURIER TRANSFORM THE DISCRETE SPECTRUM DISCRETE FOURIER TRANSFORM FORMULAE PROPERTIES OF THE DISCRETE FOURIER TRANSFORM COMPUTATION OF THE DISCRETE FOURIER TRANSFORM DEVELOPMENT OF THE FAST FOURIER TRANSFORM ALGORITHM COMPUTATIONAL COST OF THE FAST FOURIER TRANSFORM BIT SHUF?ING AND STORAGE CONSIDERATIONS THE DISCRETE FOURIER TRANSFORM OF AN IMAGE DE?NITION EVALUATION OF THE TWO DIMENSIONAL, DISCRETE FOURIER TRANSFORM THE CONCEPT OF SPATIAL FREQUENCY IMAGE FILTERING FOR GEOMETRIC ENHANCEMENT CONVOLUTION IN TWO DIMENSIONS SUPERVISED CLASSI?CATION TECHNIQUES STEPS IN SUPERVISED CLASSI?CATION MAXIMUM LIKELIHOOD CLASSI?CATION BAYES’ CLASSI?CATION THE MAXIMUM LIKELIHOOD DECISION RULE MULTIVARIATE NORMAL CLASS MODELS DECISION SURFACES THRESHOLDS NUMBER OF TRAINING PIXELS REQUIRED FOR EACH CLASS A SIMPLE ILLUSTRATION MINIMUM DISTANCE CLASSI?CATION THE CASE OF LIMITED TRAINING DATA THE DISCRIMINANT FUNCTION DEGENERATION OF MAXIMUM LIKELIHOOD TO MINIMUM DISTANCE CLASSI?CATION PARALLELEPIPED CLASSI?CATION CLASSI?CATION TIME COMPARISON OF THE CLASSI?ERS OTHER SUPERVISED APPROACHES THE MAHALANOBIS CLASSI?ER TABLE LOOK UP CLASSI?CATION THE KNN (NEAREST NEIGHBOUR) CLASSI?ER GAUSSIAN MIXTURE MODELS CONTEXT CLASSI?CATION THE CONCEPT OF SPATIAL CONTEXT CONTEXT CLASSI?CATION BY IMAGE PRE-PROCESSING POST CLASSI?CATION FILTERING PROBABILISTIC LABEL RELAXATION THE BASIC ALGORITHM THE NEIGHBOURHOOD FUNCTION DETERMINING THE COMPATIBILITY COEF?CIENTS THE FINAL STEP – STOPPING THE PROCESS EXAMPLES HANDLING SPATIAL CONTEXT BY MARKOV RANDOM FIELDS NON-PARAMETRIC CLASSI?CATION: GEOMETRIC APPROACHES LINEAR DISCRIMINATION CONCEPT OF A WEIGHT VECTOR TESTING CLASS MEMBERSHIP TRAINING SETTING THE CORRECTION INCREMENT CLASSI?CATION – THE THRESHOLD LOGIC UNIT MULTICATEGORY CLASSI?CATION SUPPORT VECTOR CLASSI?ERS LINEARLY SEPARABLE DATA LINEAR INSEPARABILITY – THE USE OF KERNEL FUNCTIONS NETWORKS OF CLASSI?ERS – SOLUTIONS OF NONLINEAR PROBLEMS THE NEURAL NETWORK APPROACH THE PROCESSING ELEMENT TRAINING THE NEURAL NETWORK – BACKPROPAGATION CHOOSING THE NETWORK PARAMETERS CLUSTERING AND UNSUPERVISED CLASSI?CATION DELINEATION OF SPECTRAL CLASSES SIMILARITY METRICS AND CLUSTERING CRITERIA THE ITERATIVE OPTIMIZATION (MIGRATING MEANS) CLUSTERING ALGORITHM MERGINGS AND DELETIONS SPLITTING ELONGATED CLUSTERS CHOICE OF INITIAL CLUSTER CENTRES CLUSTERING COST UNSUPERVISED CLASSI?CATION AND CLUSTER MAPS A CLUSTERING EXAMPLE A SINGLE PASS CLUSTERING TECHNIQUE SINGLE PASS ALGORITHM ADVANTAGES AND LIMITATIONS STRIP GENERATION PARAMETER VARIATIONS ON THE SINGLE PASS ALGORITHM AN EXAMPLE AGGLOMERATIVE HIERARCHICAL CLUSTERING CLUSTERING BY HISTOGRAM PEAK SELECTION FEATURE REDUCTION 10 1 FEATURE REDUCTION AND SEPARABILITY 10 2 SEPARABILITY MEASURES FOR MULTIVARIATE NORMAL SPECTRAL CLASS MODELS DISTRIBUTION OVERLAPS DIVERGENCE 10 2 2 1 A GENERAL EXPRESSION 10 2 2 2 DIVERGENCE OF A PAIR OF NORMAL DISTRIBUTIONS 10 2 2 3 USE OF DIVERGENCE FOR FEATURE SELECTION 10 2 2 4 A PROBLEM WITH DIVERGENCE THE JEFFRIES-MATUSITA (JM) DISTANCE 10 2 3 1 DE?NITION 10 2 3 2 COMPARISON OF DIVERGENCE AND JM DISTANCE TRANSFORMED DIVERGENCE 10 2 4 1 DE?NITION 10 2 4 2 RELATION BETWEEN TRANSFORMED DIVERGENCE AND PROBABILITY OF CORRECT CLASSI?CATION 10 2 4 3 USE OF TRANSFORMED DIVERGENCE IN CLUSTERING 10 3 SEPARABILITY MEASURES FOR MINIMUM DISTANCE CLASSI?CATION 10 4 FEATURE REDUCTION BY DATA TRANSFORMATION FEATURE REDUCTION USING THE PRINCIPAL COMPONENTS TRANSFORMATION CANONICAL ANALYSIS AS A FEATURE SELECTION PROCEDURE 10 4 2 1 WITHIN CLASS AND AMONG CLASS COVARIANCE MATRICES 10 4 2 2 A SEPARABILITY MEASURE 10 4 2 3 THE GENERALISED EIGENVALUE EQUATION 10 4 2 4 AN EXAMPLE DISCRIMINANT ANALYSIS FEATURE EXTRACTION (DAFE) NON-PARAMETRIC DISCRIMINANT ANALYSIS AND DECISION BOUNDARY FEATURE EXTRACTION (DBFE) NON-PARAMETRIC WEIGHTED FEATURE EXTRACTION (NWFE) ARITHMETIC TRANSFORMATIONS IMAGE CLASSI?CATION METHODOLOGIES 11 1 INTRODUCTION 11 2 SUPERVISED CLASSI?CATION OUTLINE DETERMINATION OF TRAINING DATA FEATURE SELECTION DETECTING MULTIMODAL DISTRIBUTIONS PRESENTATION OF RESULTS EFFECT OF RESAMPLING ON CLASSI?CATION 11 3 UNSUPERVISED CLASSI?CATION OUTLINE, AND COMPARISON WITH SUPERVISED METHODS 11 4 A HYBRID SUPERVISED/UNSUPERVISED METHODOLOGY THE ESSENTIAL STEPS CHOICE OF THE CLUSTERING REGIONS RATIONALISATION OF THE NUMBER OF SPECTRAL CLASSES 11 5 ASSESSMENT OF CLASSI?CATION ACCURACY USING A TESTING SET OF PIXELS THE LEAVE ONE OUT METHOD OF ACCURACY ASSESSMENT – CROSS VALIDATION 11 6 CASE STUDY 1: IRRIGATED AREA DETERMINATION BACKGROUND THE STUDY REGION CLUSTERING SIGNATURE GENERATION CLASSI?CATION AND RESULTS 11 7 CASE STUDY 2: MULTITEMPORAL MONITORING OF BUSH FIRES SIMPLE ILLUSTRATION OF THE TECHNIQUE THE STUDY AREA REGISTRATION PRINCIPAL COMPONENTS TRANSFORMATION CLASSI?CATION OF PRINCIPAL COMPONENTS IMAGERY 11 8 HIERARCHICAL CLASSI?CATION THE DECISION TREE CLASSI?ER DECISION TREE DESIGN PROGRESSIVE TWO-CLASS DECISION CLASSI?ER ERROR ACCUMULATION IN A DECISION TREE 11 9 A NOTE ON HYPERSPECTRAL DATA CLASSI?CATION MULTISOURCE, MULTISENSOR METHODS 12 1 THE STACKED VECTOR APPROACH 12 2 STATISTICAL MULTISOURCE METHODS JOINT STATISTICAL DECISION RULES COMMITTEE CLASSI?ERS OPINION POOLS AND CONSENSUS THEORETIC METHODS USE OF PRIOR PROBABILITY SUPERVISED LABEL RELAXATION 12 3 THE THEORY OF EVIDENCE THE CONCEPT OF EVIDENTIAL MASS COMBINING EVIDENCE – THE ORTHOGONAL SUM DECISION RULE 12 4 KNOWLEDGE-BASED IMAGE ANALYSIS KNOWLEDGE PROCESSING: EMULATING PHOTOINTERPRETATION FUNDAMENTALS OF A KNOWLEDGE-BASED IMAGE ANALYSIS SYSTEM 12 4 2 1 STRUCTURE 12 4 2 2 REPRESENTATION OF KNOWLEDGE: RULES 12 4 2 3 THE INFERENCE MECHANISM HANDLING MULTISOURCE AND MULTISENSOR DATA 12 4 4 1 RULES AS JUSTI?ERS FOR A LABELLING PROPOSITION 12 4 4 2 ENDORSEMENT OF A LABELLING PROPOSITION 12 4 4 3 KNOWLEDGE BASE AND RESULTS INTERPRETATION OF HYPERSPECTRAL IMAGE DATA 13 1 DATA CHARACTERISTICS 13 2 THE CHALLENGE TO INTERPRETATION DATA VOLUME REDUNDANCY THE NEED FOR CALIBRATION THE PROBLEM OF DIMENSIONALITY: THE HUGHES PHENOMENON 13 3 DATA CALIBRATION TECHNIQUES DETAILED RADIOMETRIC CORRECTION DATA NORMALISATION APPROXIMATE RADIOMETRIC CORRECTION 13 4 INTERPRETATION USING SPECTRAL INFORMATION SPECTRAL ANGLE MAPPING USING EXPERT SPECTRAL KNOWLEDGE AND LIBRARY SEARCHING LIBRARY SEARCHING BY SPECTRAL CODING 13 4 3 1 BINARY SPECTRAL CODES 13 4 3 2 MATCHING ALGORITHMS 13 5 HYPERSPECTRAL INTERPRETATION BY STATISTICAL METHODS LIMITATIONS OF TRADITIONAL THEMATIC MAPPING PROCEDURES BLOCK-BASED MAXIMUM LIKELIHOOD CLASSI?CATION 13 6 FEATURE REDUCTION SPECTRAL TRANSFORMATIONS FEATURE SELECTION FROM PRINCIPAL COMPONENTS TRANSFORMED DATA 13 7 REGULARISED COVARIANCE ESTIMATORS 13 8 COMPRESSION OF HYPERSPECTRAL DATA 13 9 SPECTRAL UNMIXING: END MEMBER ANALYSIS