svn-gvsig-desktop / trunk / extensions / extRemoteSensing / src / org / gvsig / remotesensing / principalcomponents / PCStatisticsProcess.java @ 18828
History | View | Annotate | Download (12.8 KB)
| 1 |
/* gvSIG. Sistema de Informaci?n Geogr?fica de la Generalitat Valenciana
|
|---|---|
| 2 |
*
|
| 3 |
* Copyright (C) 2006 Instituto de Desarrollo Regional and Generalitat Valenciana.
|
| 4 |
*
|
| 5 |
* This program is free software; you can redistribute it and/or
|
| 6 |
* modify it under the terms of the GNU General Public License
|
| 7 |
* as published by the Free Software Foundation; either version 2
|
| 8 |
* of the License, or (at your option) any later version.
|
| 9 |
*
|
| 10 |
* This program is distributed in the hope that it will be useful,
|
| 11 |
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
| 12 |
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
| 13 |
* GNU General Public License for more details.
|
| 14 |
*
|
| 15 |
* You should have received a copy of the GNU General Public License
|
| 16 |
* along with this program; if not, write to the Free Software
|
| 17 |
* Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307,USA.
|
| 18 |
*
|
| 19 |
* For more information, contact:
|
| 20 |
*
|
| 21 |
* Generalitat Valenciana
|
| 22 |
* Conselleria d'Infraestructures i Transport
|
| 23 |
* Av. Blasco Iba?ez, 50
|
| 24 |
* 46010 VALENCIA
|
| 25 |
* SPAIN
|
| 26 |
*
|
| 27 |
* +34 963862235
|
| 28 |
* gvsig@gva.es
|
| 29 |
* www.gvsig.gva.es
|
| 30 |
*
|
| 31 |
* or
|
| 32 |
*
|
| 33 |
* Instituto de Desarrollo Regional (Universidad de Castilla La-Mancha)
|
| 34 |
* Campus Universitario s/n
|
| 35 |
* 02071 Alabacete
|
| 36 |
* Spain
|
| 37 |
*
|
| 38 |
* +34 967 599 200
|
| 39 |
*/
|
| 40 |
|
| 41 |
package org.gvsig.remotesensing.principalcomponents; |
| 42 |
|
| 43 |
import org.gvsig.fmap.raster.layers.FLyrRasterSE; |
| 44 |
import org.gvsig.raster.RasterProcess; |
| 45 |
import org.gvsig.raster.buffer.BufferFactory; |
| 46 |
import org.gvsig.raster.buffer.RasterBuffer; |
| 47 |
import org.gvsig.raster.buffer.RasterBufferInvalidException; |
| 48 |
import org.gvsig.raster.dataset.IRasterDataSource; |
| 49 |
import org.gvsig.raster.grid.Grid; |
| 50 |
import org.gvsig.raster.grid.GridException; |
| 51 |
import org.gvsig.raster.util.RasterToolsUtil; |
| 52 |
import org.gvsig.remotesensing.principalcomponents.gui.PrincipalComponentCalculusPanel; |
| 53 |
|
| 54 |
import Jama.EigenvalueDecomposition; |
| 55 |
import Jama.Matrix; |
| 56 |
|
| 57 |
import com.iver.andami.PluginServices; |
| 58 |
import com.iver.cit.gvsig.project.documents.view.gui.View; |
| 59 |
|
| 60 |
/**
|
| 61 |
* PCStatisticsProcess es la clase que implementa el proceso c?lculo de estad?sticas avanzado
|
| 62 |
* para el an?lisis de componentes principales. Para la imagen y las bandas de entrada se calcula
|
| 63 |
* la matriz de varianza-covarianza y los atovalores y autovectrores asociados .
|
| 64 |
*
|
| 65 |
* @author Alejandro Mu?oz Sanchez (alejandro.munoz@uclm.es)
|
| 66 |
* @author Diego Guerrero Sevilla (diego.guerrero@uclm.es)
|
| 67 |
* @version 19/10/2007
|
| 68 |
*/
|
| 69 |
|
| 70 |
public class PCStatisticsProcess extends RasterProcess{ |
| 71 |
|
| 72 |
private Grid inputGrid = null; |
| 73 |
private double autovalors[] = null; |
| 74 |
private Matrix coVarMatrix = null; |
| 75 |
private Matrix autoVectorMatrix = null; |
| 76 |
private int percent = 0; |
| 77 |
private boolean cancel = false; |
| 78 |
private boolean selectedBands[] =null; |
| 79 |
private FLyrRasterSE rasterSE = null; |
| 80 |
private View view = null; |
| 81 |
|
| 82 |
/**
|
| 83 |
* Constructor
|
| 84 |
*/
|
| 85 |
public PCStatisticsProcess() {
|
| 86 |
} |
| 87 |
|
| 88 |
|
| 89 |
/**
|
| 90 |
* C?lculo de la matriz de covarianza, autovalores y autovectores.
|
| 91 |
*/
|
| 92 |
public void calculate() { |
| 93 |
//Calculo de matriz de covarianza:
|
| 94 |
double coVar[][]= covarianceOptimize(); |
| 95 |
// Calculo de autovectores:
|
| 96 |
coVarMatrix = new Matrix(coVar);
|
| 97 |
EigenvalueDecomposition eigenvalueDecomp = new EigenvalueDecomposition(coVarMatrix);
|
| 98 |
//Autovectores
|
| 99 |
autoVectorMatrix= eigenvalueDecomp.getV(); |
| 100 |
// Autovalores
|
| 101 |
autovalors= eigenvalueDecomp.getRealEigenvalues(); |
| 102 |
} |
| 103 |
|
| 104 |
|
| 105 |
/**
|
| 106 |
* @return array con los autovalores
|
| 107 |
*/
|
| 108 |
public Object getResult() { |
| 109 |
return autovalors;
|
| 110 |
} |
| 111 |
|
| 112 |
/**
|
| 113 |
* @return Matriz de autovectores
|
| 114 |
*/
|
| 115 |
public Matrix getAutoVectorMatrix(){
|
| 116 |
return autoVectorMatrix;
|
| 117 |
} |
| 118 |
|
| 119 |
|
| 120 |
/**
|
| 121 |
* @return Matriz varianza-covarianza
|
| 122 |
*/
|
| 123 |
public Matrix getcoVarMatrix(){
|
| 124 |
return coVarMatrix;
|
| 125 |
} |
| 126 |
|
| 127 |
|
| 128 |
/**
|
| 129 |
* C?lculo de la matriz varianza covarianza de las bandas de un Grid.
|
| 130 |
*/
|
| 131 |
private double[][] covarianceOptimize() { |
| 132 |
|
| 133 |
double dSum = 0; |
| 134 |
int iValues = 0; |
| 135 |
double[][] coV =new double[inputGrid.getRasterBuf().getBandCount()][inputGrid.getRasterBuf().getBandCount()]; |
| 136 |
double cancelMatrix[][]= new double[][]{{0}}; |
| 137 |
double valorBandai=0, valorBandaj=0; |
| 138 |
int bandCount = inputGrid.getRasterBuf().getBandCount();
|
| 139 |
if(inputGrid.getRasterBuf().getDataType() == RasterBuffer.TYPE_BYTE){
|
| 140 |
// Se recorre el grid obener la matriz de cov
|
| 141 |
for (int i = 0; i < bandCount; i++) { |
| 142 |
for (int j = i; j < bandCount; j++) { |
| 143 |
// si cancelado se devuelve cancelMatrix
|
| 144 |
if(cancel)
|
| 145 |
return cancelMatrix;
|
| 146 |
iValues=0;
|
| 147 |
dSum = 0;
|
| 148 |
// Calculo covarianza entre las bandas i,j
|
| 149 |
for (int k=0; k<inputGrid.getNX(); k++){ |
| 150 |
for (int l=0;l<inputGrid.getNY();l++){ |
| 151 |
try{
|
| 152 |
inputGrid.setBandToOperate(i); |
| 153 |
valorBandai=inputGrid.getCellValueAsByte(k,l) -inputGrid.getMeanValue(); |
| 154 |
inputGrid.setBandToOperate(j); |
| 155 |
valorBandaj=inputGrid.getCellValueAsByte(k,l) -inputGrid.getMeanValue(); |
| 156 |
} catch (GridException e) {
|
| 157 |
RasterToolsUtil.messageBoxError(PluginServices.getText(this, "grid_error"), this, e); |
| 158 |
} |
| 159 |
dSum += valorBandai*valorBandaj; |
| 160 |
iValues++; |
| 161 |
} |
| 162 |
} |
| 163 |
// Se asigna el valor a la matriz
|
| 164 |
if (iValues>1) |
| 165 |
coV[i][j]=dSum/(double)(iValues);
|
| 166 |
else
|
| 167 |
coV[i][j]= inputGrid.getNoDataValue(); |
| 168 |
} |
| 169 |
|
| 170 |
if (bandCount>1) |
| 171 |
percent = (i+1)*100/(bandCount-1); |
| 172 |
else
|
| 173 |
percent= (i+1)*100/(1); |
| 174 |
} |
| 175 |
|
| 176 |
} // Fin tipo Byte
|
| 177 |
|
| 178 |
|
| 179 |
if(inputGrid.getRasterBuf().getDataType() == RasterBuffer.TYPE_SHORT){
|
| 180 |
// Se recorre el grid obener la matriz de cov
|
| 181 |
for (int i = 0; i < bandCount; i++) { |
| 182 |
for (int j = i; j < bandCount; j++) { |
| 183 |
if(cancel)
|
| 184 |
return cancelMatrix;
|
| 185 |
iValues=0;
|
| 186 |
dSum = 0;
|
| 187 |
// Calculo covarianza entre las bandas i,j
|
| 188 |
for (int k=0; k<inputGrid.getNX(); k++){ |
| 189 |
for (int l=0;l<inputGrid.getNY();l++){ |
| 190 |
try{
|
| 191 |
inputGrid.setBandToOperate(i); |
| 192 |
valorBandai=inputGrid.getCellValueAsShort(k,l) -inputGrid.getMeanValue(); |
| 193 |
inputGrid.setBandToOperate(j); |
| 194 |
valorBandaj=inputGrid.getCellValueAsShort(k,l) -inputGrid.getMeanValue(); |
| 195 |
} catch (GridException e) {
|
| 196 |
RasterToolsUtil.messageBoxError(PluginServices.getText(this, "grid_error"), this, e); |
| 197 |
} |
| 198 |
dSum += valorBandai*valorBandaj; |
| 199 |
iValues++; |
| 200 |
} |
| 201 |
} |
| 202 |
// Se asigna el valor a la matriz
|
| 203 |
if (iValues>1) |
| 204 |
coV[i][j]=dSum/(double)(iValues);
|
| 205 |
else
|
| 206 |
coV[i][j]= inputGrid.getNoDataValue(); |
| 207 |
|
| 208 |
} |
| 209 |
|
| 210 |
if (bandCount>1) |
| 211 |
percent = (i+1)*100/(bandCount-1); |
| 212 |
else
|
| 213 |
percent= (i+1)*100/(1); |
| 214 |
} |
| 215 |
|
| 216 |
} // Fin tipo Short
|
| 217 |
|
| 218 |
|
| 219 |
if(inputGrid.getRasterBuf().getDataType() == RasterBuffer.TYPE_INT){
|
| 220 |
// Se recorre el grid obener la matriz de cov
|
| 221 |
for (int i = 0; i < bandCount; i++) { |
| 222 |
for (int j = i; j < bandCount; j++) { |
| 223 |
if(cancel)
|
| 224 |
return cancelMatrix;
|
| 225 |
iValues=0;
|
| 226 |
dSum = 0;
|
| 227 |
// Calculo covarianza entre las bandas i,j
|
| 228 |
for (int k=0; k<inputGrid.getNX(); k++){ |
| 229 |
for (int l=0;l<inputGrid.getNY();l++){ |
| 230 |
try{
|
| 231 |
inputGrid.setBandToOperate(i); |
| 232 |
valorBandai=inputGrid.getCellValueAsInt(k,l) -inputGrid.getMeanValue(); |
| 233 |
inputGrid.setBandToOperate(j); |
| 234 |
valorBandaj=inputGrid.getCellValueAsInt(k,l) -inputGrid.getMeanValue(); |
| 235 |
} catch (GridException e) {
|
| 236 |
RasterToolsUtil.messageBoxError(PluginServices.getText(this, "grid_error"), this, e); |
| 237 |
} |
| 238 |
dSum += valorBandai*valorBandaj; |
| 239 |
iValues++; |
| 240 |
} |
| 241 |
} |
| 242 |
// Se asigna el valor a la matriz
|
| 243 |
if (iValues>1) |
| 244 |
coV[i][j]=dSum/(double)(iValues);
|
| 245 |
else
|
| 246 |
coV[i][j]= inputGrid.getNoDataValue(); |
| 247 |
} |
| 248 |
|
| 249 |
if (bandCount>1) |
| 250 |
percent = (i+1)*100/(bandCount-1); |
| 251 |
else
|
| 252 |
percent= (i+1)*100/(1); |
| 253 |
} |
| 254 |
|
| 255 |
} // Fin tipo Int
|
| 256 |
|
| 257 |
|
| 258 |
if(inputGrid.getRasterBuf().getDataType() == RasterBuffer.TYPE_FLOAT){
|
| 259 |
// Se recorre el grid obener la matriz de cov
|
| 260 |
for (int i = 0; i < bandCount; i++) { |
| 261 |
for (int j = i; j < bandCount; j++) { |
| 262 |
if(cancel)
|
| 263 |
return cancelMatrix;
|
| 264 |
iValues=0;
|
| 265 |
dSum = 0;
|
| 266 |
// Calculo la covarianza entre las bandas i,j
|
| 267 |
for (int k=0; k<inputGrid.getNX(); k++){ |
| 268 |
for (int l=0;l<inputGrid.getNY();l++){ |
| 269 |
try{
|
| 270 |
inputGrid.setBandToOperate(i); |
| 271 |
valorBandai=inputGrid.getCellValueAsFloat(k,l) -inputGrid.getMeanValue(); |
| 272 |
inputGrid.setBandToOperate(j); |
| 273 |
valorBandaj=inputGrid.getCellValueAsFloat(k,l) -inputGrid.getMeanValue(); |
| 274 |
} catch (GridException e) {
|
| 275 |
RasterToolsUtil.messageBoxError(PluginServices.getText(this, "grid_error"), this, e); |
| 276 |
} |
| 277 |
dSum += valorBandai*valorBandaj; |
| 278 |
iValues++; |
| 279 |
} |
| 280 |
} |
| 281 |
// Se asigna el valor a la matriz
|
| 282 |
if (iValues>1) |
| 283 |
coV[i][j]=dSum/(double)(iValues);
|
| 284 |
else
|
| 285 |
coV[i][j]= inputGrid.getNoDataValue(); |
| 286 |
} |
| 287 |
|
| 288 |
if (bandCount>1) |
| 289 |
percent = (i+1)*100/(bandCount-1); |
| 290 |
else
|
| 291 |
percent= (i+1)*100/(1); |
| 292 |
} |
| 293 |
|
| 294 |
} // Fin tipo Float
|
| 295 |
|
| 296 |
|
| 297 |
if(inputGrid.getRasterBuf().getDataType() == RasterBuffer.TYPE_DOUBLE){
|
| 298 |
// Se recorre el grid obener la matriz de cov
|
| 299 |
for (int i = 0; i < bandCount; i++) { |
| 300 |
for (int j = i; j < bandCount; j++) { |
| 301 |
if(cancel)
|
| 302 |
return cancelMatrix;
|
| 303 |
iValues=0;
|
| 304 |
dSum = 0;
|
| 305 |
// Calculo la covarianza entre las bandas i,j
|
| 306 |
for (int k=0; k<inputGrid.getNX(); k++){ |
| 307 |
for (int l=0;l<inputGrid.getNY();l++){ |
| 308 |
try{
|
| 309 |
inputGrid.setBandToOperate(i); |
| 310 |
valorBandai=inputGrid.getCellValueAsDouble(k,l) -inputGrid.getMeanValue(); |
| 311 |
inputGrid.setBandToOperate(j); |
| 312 |
valorBandaj=inputGrid.getCellValueAsDouble(k,l) -inputGrid.getMeanValue(); |
| 313 |
} catch (GridException e) {
|
| 314 |
RasterToolsUtil.messageBoxError(PluginServices.getText(this, "grid_error"), this, e); |
| 315 |
} |
| 316 |
dSum += valorBandai*valorBandaj; |
| 317 |
iValues++; |
| 318 |
} |
| 319 |
} |
| 320 |
// Asigno el valor a la matriz
|
| 321 |
if (iValues>1) |
| 322 |
coV[i][j]=dSum/(double)(iValues);
|
| 323 |
else
|
| 324 |
coV[i][j]= inputGrid.getNoDataValue(); |
| 325 |
} |
| 326 |
|
| 327 |
if (bandCount>1) |
| 328 |
percent = (i+1)*100/(bandCount-1); |
| 329 |
else
|
| 330 |
percent= (i+1)*100/(1); |
| 331 |
} |
| 332 |
|
| 333 |
} // Fin tipo Double
|
| 334 |
|
| 335 |
for (int i = 0; i < bandCount; i++) { |
| 336 |
for (int j = 0; j < bandCount; j++) { |
| 337 |
if(j<i)
|
| 338 |
coV[i][j]=coV[j][i]; |
| 339 |
} |
| 340 |
} |
| 341 |
|
| 342 |
return coV;
|
| 343 |
} |
| 344 |
|
| 345 |
|
| 346 |
|
| 347 |
/**
|
| 348 |
* Construye el grid con las bandas seleccionadas
|
| 349 |
*/
|
| 350 |
public void setGrid(){ |
| 351 |
|
| 352 |
IRasterDataSource dsetCopy = null;
|
| 353 |
dsetCopy = rasterSE.getDataSource().newDataset(); |
| 354 |
BufferFactory bufferFactory = new BufferFactory(dsetCopy);
|
| 355 |
|
| 356 |
int longitud=0; |
| 357 |
for (int i=0; i<selectedBands.length;i++) |
| 358 |
if (selectedBands[i]) longitud++;
|
| 359 |
|
| 360 |
int bands[]= new int[longitud]; |
| 361 |
int j=0; |
| 362 |
for (int i=0; i<selectedBands.length; i++) |
| 363 |
if (selectedBands[i])
|
| 364 |
{ bands[j]=i;
|
| 365 |
j++; |
| 366 |
} |
| 367 |
try {
|
| 368 |
inputGrid = new Grid(bufferFactory, bands);
|
| 369 |
} catch (RasterBufferInvalidException e) {
|
| 370 |
e.printStackTrace(); |
| 371 |
} |
| 372 |
} |
| 373 |
|
| 374 |
|
| 375 |
/*
|
| 376 |
* (non-Javadoc)
|
| 377 |
* @see org.gvsig.gui.beans.incrementabletask.IIncrementable#getLabel()
|
| 378 |
*/
|
| 379 |
public String getLabel() { |
| 380 |
return PluginServices.getText(this,"procesando"); |
| 381 |
} |
| 382 |
|
| 383 |
/*
|
| 384 |
* (non-Javadoc)
|
| 385 |
* @see org.gvsig.gui.beans.incrementabletask.IIncrementable#getLog()
|
| 386 |
*/
|
| 387 |
public String getLog() { |
| 388 |
return PluginServices.getText(this,"calculando_estadisticas")+"..."; |
| 389 |
} |
| 390 |
|
| 391 |
|
| 392 |
/*
|
| 393 |
* (non-Javadoc)
|
| 394 |
* @see org.gvsig.gui.beans.incrementabletask.IIncrementable#getPercent()
|
| 395 |
*/
|
| 396 |
public int getPercent() { |
| 397 |
return percent;
|
| 398 |
} |
| 399 |
|
| 400 |
/*
|
| 401 |
* (non-Javadoc)
|
| 402 |
* @see org.gvsig.gui.beans.incrementabletask.IIncrementable#getTitle()
|
| 403 |
*/
|
| 404 |
public String getTitle() { |
| 405 |
return PluginServices.getText(this,"principal_components"); |
| 406 |
} |
| 407 |
|
| 408 |
/**
|
| 409 |
* @return grid
|
| 410 |
*/
|
| 411 |
public Grid getInputGrid() {
|
| 412 |
return inputGrid;
|
| 413 |
} |
| 414 |
|
| 415 |
/**
|
| 416 |
* @return raster de entrada
|
| 417 |
* */
|
| 418 |
public FLyrRasterSE getRasterLayer(){
|
| 419 |
return rasterSE;
|
| 420 |
} |
| 421 |
|
| 422 |
|
| 423 |
public void init() { |
| 424 |
// Se toman los parametros del proceso
|
| 425 |
selectedBands = (boolean []) getParam("selectedBands"); |
| 426 |
view = (View)getParam("view"); |
| 427 |
rasterSE = getLayerParam("layer");
|
| 428 |
|
| 429 |
} |
| 430 |
|
| 431 |
|
| 432 |
/**
|
| 433 |
* Proceso de calculo de estadisticas para Principal Component
|
| 434 |
* */
|
| 435 |
public void process() throws InterruptedException { |
| 436 |
try{
|
| 437 |
// Se establece el grid con las bandas seleccionadas
|
| 438 |
setGrid(); |
| 439 |
// Calculo de matriz de covarianza:
|
| 440 |
double coVar[][]= covarianceOptimize(); |
| 441 |
// Calculo de autovectores:
|
| 442 |
coVarMatrix = new Matrix(coVar);
|
| 443 |
EigenvalueDecomposition eigenvalueDecomp = new EigenvalueDecomposition(coVarMatrix);
|
| 444 |
//Autovectores
|
| 445 |
autoVectorMatrix= eigenvalueDecomp.getV(); |
| 446 |
// Autovalores
|
| 447 |
autovalors= eigenvalueDecomp.getRealEigenvalues(); |
| 448 |
|
| 449 |
}finally{
|
| 450 |
if(!cancel){
|
| 451 |
incrementableTask.processFinalize(); |
| 452 |
PrincipalComponentCalculusPanel altPrincipalComponentCalculusPanel = new PrincipalComponentCalculusPanel(view,this); |
| 453 |
PluginServices.getMDIManager().addWindow(altPrincipalComponentCalculusPanel); |
| 454 |
} |
| 455 |
} |
| 456 |
|
| 457 |
} |
| 458 |
|
| 459 |
/**
|
| 460 |
* @return vista de la aplicacion
|
| 461 |
*/
|
| 462 |
public View getView(){ |
| 463 |
return view;
|
| 464 |
} |
| 465 |
} |