Hyperspectral imaging : techniques for spectral detection and classification / Chein-I Chang.

Format:

Book

Author/Creator:

Chang, Chein-I.

Contributor:

Classes of 1883 and 1884 Fund.

Language:

English

Subjects (All):

Remote sensing.

Multispectral imaging.

Image processing--Digital techniques.

Image processing.

Physical Description:

xvi, 374 pages : illustrations ; 26 cm

Place of Publication:

New York : Kluwer Academic/Plenum Publishers, [2003]

Contents:

1.2.1 Stochastic Hyperspectral Measures 3

1.2.2 Subpixel Detection 4

1.2.3 Mixed Pixel classification (MPC) 5

1.2.3.1 Unconstrained MPC 7

1.2.3.2 Constrained MPC 7

1.2.3.3 Automatic Mixed Pixel Classification (AMPC) 8

Part I Hyperspectral Measures 13

2 Hyperspectral Measures for Spectral Characterization 15

2.1 Measures of Spectral Variability 15

2.1.1 Spectral Information Measure (SIM) 16

2.1.2 Hidden Markov Model(HMM)-Based Measure 17

2.2 Spectral Similarity Measures 20

2.2.1.1 Distance-Based Measures 20

2.2.1.2 Orthogonal Projection-Based Measures 20

2.2.2 Spectral Information Divergence (SID) 21

2.2.3 Hidden Markov Model-Based Information Divergence (HMMID) 23

2.3 Measures of Spectral Discriminability 23

2.3.1 Relative Spectral Discriminatory ProBability (RSDPB) 24

2.3.2 Relative Spectral Discriminatory PoWer (RSDPW) 24

2.3.3 Relative Spectral Discriminatory Entropy (RSDE) 25

2.4 Experiments 26

2.4.1 AVIRIS Data 26

2.4.2 HYDICE Data 31

Part II Subpixel Detection 37

3 Target Abundance-Constrained Subpixel Detection: Partially Constrained Least-Squares Methods 39

3.2 Linear Spectral Mixture Model 40

3.3 Orthogonal Subspace Projection (OSP) 41

3.4 Sum-to-one Constrained Least Squares Method (SCLS) 44

3.5 Nonnegativity Constrained Least Squares Method (NCLS) 45

3.6 Hyperspectral Image Experiments 48

4 Target Signature-Constrained Subpixel Detection: Linearly Constrained Minimum Variance (LCMV) 51

4.2 LCMV Target Detector 53

4.2.1 Constrained Energy Minimization (CEM) 54

4.2.2 Target-Constrained Interference-Minimized Filter (TCIMF) 55

4.3 Relationship Among Osp, Cem and Tcimf 56

4.4 A Comparartive Analysis Between Cem and Tcimf 58

4.4.1 Computer Simulations 58

4.4.2 Hyperspectral Image Experiments 61

4.5 Sensitivity of Cem and Tcimf to Level of Target Information 63

4.5.1 Computer Simulations 64

4.5.2 Hyperspectral Image Experiments 67

4.6 Real-Time Processing 68

5 Automatic Subpixel Detection: Unsupervised Subpixel Detection 73

5.2 Unsupervised Vector Quantization (UVQ)-Based Algorithm 74

5.3 Unsupervised Target Generation Process (UTGP) 75

5.4 Unsupervised Ncls (UNCLS) Algorithm 78

5.5 Experiments 80

6 Automatic Subpixel Detection: Anomaly Detection 89

6.2 RXD 91

6.3 Lptd and Utd 94

6.4 Relationship Between Cem and Rxd 97

6.5 Real-Time Processing 99

7 Sensitivity of Subpixel Detection 105

7.2 Sensitivity of Target Knowledge 107

7.3 Sensitivity of Noise 111

7.3.1 Tscsd 111

7.3.2 Hyperspectral Image Experiments 116

7.3.2.1 AVIRIS Data 116

7.3.2.2 HYDICE Data 125

7.4 Sensitivity of Anomaly Detection 129

Part III Unconstrained Mixed Pixel Classification 139

8 Unconstrained Mixed Pixel Classification: Least-Squares Subspace Projection 141

8.2 A Posteriori Osp 144

8.2.1 Signature Subspace Projection (SSP) Classifier 144

8.2.2 Target Subspace Projection (TSP) Classifier 146

8.2.3 Oblique Subspace Projection (OBSP) Classifier 147

8.2.4 Unconstrained Maximum Likelihood Estimation Classifier 148

8.3 Estimation Error Evaluated by Roc Analysis 150

8.3.1 Signature Subspace Projection (SSP) Classifier 151

8.3.2 Oblique Subspace Projection (OBSP) Classifier 153

8.4 Computer Simulations and Hyperspectral Image experiments 153

8.4.1 Computer Simulations 154

8.4.2 Hyperspectral Data 156

9 A Quantitative Analysis of Mixed-to-Pure Pixel Conversion (MPCV) 161

9.2 Conversion of MPC to PPC 162

9.2.1 Mixed-to-Pure Pixel Converter (MPCV) 163

9.2.2 Minimum Distance-Based Classification 164

9.2.3 Fisher's Linear Discriminant Analysis (LDA) 166

9.2.4 Unsupervised Classification 169

9.3 Criteria for Target Detection and Classification 169

9.4 Comparative Performance Analysis 171

Part IV Constrained Mixed Pixel Classification 179

10 Target Abundance-Constrained Mixed Pxiel Classification (TACMPC) 181

10.2 Fully Constrained Least-Squares Approach 183

10.2.1 Fully Constrained Least-Squares Method (FCLS) 183

10.2.2 Unsupervised FCLS Method (UFCLS) 183

10.3 Modified Fully Constrained Least-Squares (MFCLS) Approach 184

10.4 Computer Simulations and Real Hyperspectral Image Experiments 186

10.4.1 Computer Simulations 186

10.4.2 AVIRIS Image Experiments 188

10.4.3 HYDICE Image Experiments 193

10.5 Near Real-Time Implementation 201

11 Target Signature-Constrained Mixed Pixel Classification (TSCMPC): LCMV Classifiers 207

11.2 LCMV Classifer 208

11.3 Bowles Et Al.'s Filter Vectors (FV) Algorithm 209

11.4 Color Assignment of LCMV Classifiers 211

11.5 Extension of Cem Filter to Classifiers 213

11.5.1 Winner-Take-All CEM (WTACEM) Classifier 213

11.5.2 Sum CEM (SCEM) Classifier 213

11.5.3 Multiple-Target CEM (MTCEM) Classifier 213

11.5.4 Target-Constrained Interference-Minimized (TCIM) Classifier 214

11.6 Computer Simulations 214

11.7 Hyperspectral Image Experiments 218

11.8 Real-Time Implementation for LCMV Classifiers 223

12 Target Signature-Constrained Mixed Pixel Classification (TSCMPC): Linearly Constrained Discriminant Analysis (LCDA) 229

12.2 LCDA 230

12.3 Whitening Process for LCDA 233

12.4 Bowles Et Al.'s Filter Vectors (FV) Algorithm 234

12.5 Computer Simulations and Hyperspectral Image Experiments 235

Part V Automatic Mixed Pixel Classification (AMPC) 243

13 Automatic Mixed Pixel Classification (AMPC): Unsupervised Mixed Pixel Classification 245

13.2 Unsupervised MPC 246

13.3 Desired Target Detection and Classification 246

13.4 Automatic Target Detection and Classification 253

14 Automatic Mixed Pixel Classification (AMPC): Anomaly Classification 257

14.2 Target Discrimination Measures 258

14.3 Anomaly Classification 260

14.4 Automatic Thresholding Method 260

14.5 Analysis on Target Correlation Using Target Discrimination Measures 265

14.6 On-Line Implementation 270

15 Automatic Mixed Pixel Classification (AMPC): Linear Spectral Random Mixture Analysis (LSRMA) 277

15.2 Independent Component Analysis (ICA) 279

15.3 ICA-Based LSRMA 280

15.3.1 Relative Entropy-Based Measure for ICA 281

15.3.2 Learning Algorithm to Find Separating Matrix W 282

15.4 Experiments 284

15.4.1 AVIRIS Image Experiments 284

15.4.2 HYDICE Image Experiments 289

15.5 3-D Roc Analysis for LSRMA 295

16 Automatic Mixed Pixel Classification (AMPC): Projection Pursuit 305

16.2 Projection Pursuit 307

16.3 Evolutionary Algorithm (EA) 308

16.4 Thresholding of Projection Images Using Zero-Detection 310

16.5 Experiments 311

16.5.1 AVIRIS Data Experiments 311

16.5.2 HYDICE Data Experiments 313

17 Estimation for Virtual Dimensionality of Hyperspectral Imagery 319

17.2 Neyman-Pearson Detection Theory-Based Eigen-Thresholding Analysis (Hfc Method) 321

17.3 Estimation of Noise Covariance Matrix 323

17.3.1 Residual Analysis (Roger, 1996) 323

17.3.2 Inter/Intra-Band Prediction Noise Estimation: Spatial/Spectral Prediction Noise Estimation (Roger and Arnold, 1996) 325

17.4 Noise Estimation-Based Eigen-Thresholding 326

17.4.1 Noise-Whitened HFC (NWHFC) Method 326

17.4.2 Noise Subspace Projection (NSP) 326

17.4.3 AIC and MDL 327

17.5 Computer Simulations and Hyperspectral Image Experiments 328

17.5.1 Computer Simulations 328

17.5.2 AVIRIS and HYDICE Image Experiments 330

18 Conclusions and Further Techniques 335

18.1 Functional Taxonomy of Techniques 335

18.2 Mathematical Taxonomy of Techniques 337

18.3 Experiments 339

18.4 Roc Analysis for Subpixel Detection and Mixed Pixel Classification 340

18.5 Sensitivity Issues 341

18.5.1 Sensitivity to Level of Target Information 341

18.5.2 Sensitivity to Noise 341

18.6 Real-Time Implementation 341

18.7.1 Generalized Orthogonal Subspace Projection 342

18.7.2 Convex Cone Analysis 343

18.7.3 Kalman Filter-Based Linear Unmixing 344

18.7.4 Interference-Annihilated Eigen-Analysis 344

18.7.5 Band Selection 345

18.7.6 Linear Mixture Analysis-Based Data Compression 346

18.7.7 Radial Basis Function Neural Network Approach 347

18.8 Applications to Magnetic Resonance Imaging 347.

Notes:

Includes bibliographical references (pages 353-364) and index.

Local Notes:

Acquired for the Penn Libraries with assistance from the Classes of 1883 and 1884 Fund.

ISBN:

0306474832

OCLC:

50803392