ecological-engine-geospatia.../src/main/java/org/gcube/dataanalysis/geo/utils/VectorOperations.java

272 lines
9.7 KiB
Java

package org.gcube.dataanalysis.geo.utils;
import java.util.ArrayList;
import java.util.List;
import org.gcube.contentmanagement.lexicalmatcher.utils.AnalysisLogger;
import org.gcube.dataanalysis.ecoengine.utils.Tuple;
import scala.actors.threadpool.Arrays;
import ucar.ma2.ArrayByte;
import ucar.ma2.ArrayDouble;
import ucar.ma2.IndexIterator;
public class VectorOperations {
public static ArrayDouble.D2 arrayByte2DArrayDouble(ArrayByte bytes) {
int[] shapeD = bytes.getShape();
int yD = shapeD[0];
int xD = shapeD[1];
AnalysisLogger.getLogger().debug(xD + "X" + yD + "=" + (xD * yD));
ArrayDouble.D2 doublea = new ArrayDouble.D2(yD, xD);
IndexIterator iterator = bytes.getIndexIterator();
for (int x = 0; x < xD; x++) {
for (int y = 0; y < yD; y++) {
Byte bytex = (Byte) iterator.next();
doublea.set(y, x, bytex.doubleValue());
}
}
return doublea;
}
public static ArrayDouble.D3 arrayByte3DArrayDouble(ArrayByte bytes) {
int[] shapeD = bytes.getShape();
int zD = shapeD[0];
int yD = shapeD[1];
int xD = shapeD[2];
AnalysisLogger.getLogger().debug(xD + "X" + yD + "X" + zD + "=" + (xD * yD * zD));
ArrayDouble.D3 doublea = new ArrayDouble.D3(zD, yD, xD);
IndexIterator iterator = bytes.getIndexIterator();
for (int x = 0; x < xD; x++) {
for (int y = 0; y < yD; y++) {
for (int z = 0; z < zD; z++) {
Byte bytex = (Byte) iterator.next();
doublea.set(z, y, x, bytex.doubleValue());
}
}
}
return doublea;
}
/**
* Finds the best association between a grid of 3d points in a certain time instant and a set of 5 dimensional points. each tuple is expected to be formed by (x,y,z,t,value) in 5 dimensions and by (x,y,z) in 3 dimensions
**/
// suggestion: given a resolution R, give Math.sqrt(2)*R/2=0.7*R as tolerance
public static List<Double> assignPointsValuesToGrid(List<Tuple<Double>> grid3d, int gridTimeInstant, List<Tuple<Double>> coordinates5d, double tolerance) {
// List<Double> valuesForGrid = new ArrayList<Double>();
int gridSize = grid3d.size();
Double [] valuesForGridd = new Double[gridSize];
for (int i = 0; i < gridSize; i++) {
// valuesForGrid.add(Double.NaN);
valuesForGridd [i] = Double.NaN;
}
// AnalysisLogger.getLogger().debug("Grid contains: "+grid3d.size()+" values");
// AnalysisLogger.getLogger().debug("Dataset contains: "+coordinates5d.size()+" values");
int foundmatches = 0;
long count = 0;
AnalysisLogger.getLogger().debug("Assigning : "+coordinates5d.size()+" elements");
for (Tuple<Double> coord5d : coordinates5d) {
double rx = coord5d.getElements().get(0);
double ry = coord5d.getElements().get(1);
double rz = coord5d.getElements().get(2);
double rt = coord5d.getElements().get(3);
double rvalue = coord5d.getElements().get(4);
int gridIdx = 0;
for (Tuple<Double> coord3d : grid3d) {
double x = coord3d.getElements().get(0);
double y = coord3d.getElements().get(1);
double z = coord3d.getElements().get(2);
double d = distance(x, y, z, gridTimeInstant, rx, ry, rz, rt);
if (d <= tolerance) {
// AnalysisLogger.getLogger().debug("Association: distance between grid:("+x+","+y+","+z+","+gridTimeInstant+") and point:("+rx+","+ry+","+rz+","+rt+") is "+d);
// valuesForGrid.set(gridIdx, rvalue);
valuesForGridd[gridIdx]=rvalue;
foundmatches++;
}
gridIdx++;
}
count++;
if (count%50000==0)
AnalysisLogger.getLogger().debug("Vector Operations Assigned: "+count+" elements");
}
AnalysisLogger.getLogger().debug("Association: Found "+foundmatches+" matches between the grid of points and the coordinates");
List<Double> valuesForGrid = Arrays.asList(valuesForGridd);
return valuesForGrid;
}
public static List<Double> assignGridValuesToPoints2D(List<Tuple<Double>> grid3d, List<Double> gridValues, List<Tuple<Double>> coordinates4d, double tolerance) {
// List<Double> valuesForPoints = new ArrayList<Double>();
int gridSize = coordinates4d.size();
Double [] valuesForPoints = new Double[gridSize];
for (int i = 0; i < gridSize; i++) {
valuesForPoints[i]=Double.NaN;
}
int foundmatches = 0;
int points=0;
for (Tuple<Double> coord4d : coordinates4d) {
double rx = coord4d.getElements().get(0);
double ry = coord4d.getElements().get(1);
int gridIdx = 0;
for (Tuple<Double> gridElement : grid3d) {
double x = gridElement.getElements().get(0);
double y = gridElement.getElements().get(1);
double d = distance(x, y, 0, 0, rx, ry, 0, 0);
if (d <= tolerance) {
// AnalysisLogger.getLogger().debug("Association: distance between grid:("+x+","+y+","+z+","+gridTimeInstant+") and point:("+rx+","+ry+","+rz+","+rt+") is "+d);
valuesForPoints[points] = gridValues.get(gridIdx);
foundmatches++;
break;
}
gridIdx++;
}
points++;
}
AnalysisLogger.getLogger().debug("Association: Found "+foundmatches+" matches between the points and the grid");
return Arrays.asList(valuesForPoints);
}
public static List<Double> assignGridValuesToPoints(List<Tuple<Double>> grid3d, int gridTimeInstant, List<Double> gridValues, List<Tuple<Double>> coordinates4d, double tolerance) {
List<Double> valuesForPoints = new ArrayList<Double>();
int gridSize = coordinates4d.size();
for (int i = 0; i < gridSize; i++) {
valuesForPoints.add(Double.NaN);
}
int foundmatches = 0;
int points=0;
for (Tuple<Double> coord4d : coordinates4d) {
double rx = coord4d.getElements().get(0);
double ry = coord4d.getElements().get(1);
double rz = coord4d.getElements().get(2);
double rt = coord4d.getElements().get(3);
int gridIdx = 0;
for (Tuple<Double> gridElement : grid3d) {
double x = gridElement.getElements().get(0);
double y = gridElement.getElements().get(1);
double z = gridElement.getElements().get(2);
double d = distance(x, y, z, gridTimeInstant, rx, ry, rz, rt);
if (d <= tolerance) {
// AnalysisLogger.getLogger().debug("Association: distance between grid:("+x+","+y+","+z+","+gridTimeInstant+") and point:("+rx+","+ry+","+rz+","+rt+") is "+d);
valuesForPoints.set(points, gridValues.get(gridIdx));
foundmatches++;
break;
}
gridIdx++;
}
points++;
}
AnalysisLogger.getLogger().debug("Association: Found "+foundmatches+" matches between the points and the grid");
return valuesForPoints;
}
public static double distance(double x1, double y1, double z1, double t1, double x2, double y2, double z2, double t2) {
return Math.sqrt(((x1 - x2) * (x1 - x2)) + ((y1 - y2) * (y1 - y2)) + ((z1 - z2) * (z1 - z2)) + ((t1 - t2) * (t1 - t2)));
}
public static List<Tuple<Double>> generateCoordinateTripletsInBoundingBox_old(double x1, double x2, double y1, double y2, double z, double xResolution, double yResolution) {
int ysteps = (int) ((y2 - y1) / yResolution);
int xsteps = (int) ((x2 - x1) / xResolution);
List<Tuple<Double>> tuples = new ArrayList<Tuple<Double>>();
AnalysisLogger.getLogger().debug("Building the points grid according to YRes:" + yResolution + " and XRes:" + xResolution);
// build the tuples according to the desired resolution
for (int i = 0; i < ysteps + 1; i++) {
double y = (i * yResolution) + y1;
if (i == ysteps)
y = y2;
for (int j = 0; j < xsteps + 1; j++) {
double x = (j * xResolution) + x1;
if (j == xsteps)
x = x2;
tuples.add(new Tuple<Double>(x, y, z));
}
}
return tuples;
}
public static List<Tuple<Double>> generateCoordinateTripletsInBoundingBox(double x1, double x2, double y1, double y2, double z, double xResolution, double yResolution) {
List<Tuple<Double>> tuples = new ArrayList<Tuple<Double>>();
for (double y=y1;y<=y2;y+=yResolution){
for (double x=x1;x<=x2;x+=xResolution){
tuples.add(new Tuple<Double>(x,y,z));
}
}
return tuples;
}
public static double [][] vectorToMatix(List<Double> values, double x1, double x2, double y1, double y2, double xResolution, double yResolution){
int ntriplets = values.size();
int ysteps = 0;
for (double y=y1;y<=y2;y+=yResolution){
ysteps++;
}
int xsteps = 0;
for (double x=x1;x<=x2;x+=xResolution){
xsteps++;
}
double[][] slice = new double[ysteps][xsteps];
int k = 0;
int g = 0;
// cycle on all the triplets to recontruct the matrix
for (int t = 0; t < ntriplets; t++) {
// take the corresponding (time,value) pair
Double value = values.get(t);
// if there is value, then set it, otherwise set NaN
// the layer is undefined in that point and a value must be generated
// assign a value to the matrix
slice[k][g] = value;
// increase the x step according to the matrix
if (g==xsteps-1) {
g = 0;
k++;
} else
g++;
}
return slice;
}
public void applyNearestNeighbor() {
/*
* AnalysisLogger.getLogger().debug("Applying nearest Neighbor to all the rows"); //apply nearest neighbor to each row AlgorithmConfiguration config = new AlgorithmConfiguration(); config.setConfigPath(configDir); boolean rapidinit = false;
*
*
* for (int i=0;i<slice.length;i++){ // AnalysisLogger.getLogger().debug("Checking for unfilled values"); boolean tofill = false; for (int j=0;j<slice[i].length;j++) { if (new Double(slice[i][j]).equals(Double.NaN)) tofill = true; } if (tofill){ if (!rapidinit){ config.initRapidMiner(); rapidinit=true; } AnalysisLogger.getLogger().debug("Filling signal"); double[] ssliced = SignalProcessing.fillSignal(slice[i]); slice[i] = ssliced; } // else // AnalysisLogger.getLogger().debug("Signal yet complete"); }
*/
}
}