Skip to content

I
imagej-elphel
Commit c5a866f4 authored by Andrey Filippov's avatar Andrey Filippov
Browse files
Options

more on initial planes extraction

parent c475ffe4
Hide whitespace changes
Inline Side-by-side
Showing with 508 additions and 25 deletions
src/main/java/SuperTiles.java
View file @ c5a866f4
......@@ -21,6 +21,7 @@
** -----------------------------------------------------------------------------**
**
*/
import java.awt.Point;
import java.util.ArrayList;
import java.util.Collections;
import java.util.Comparator;
......@@ -1543,6 +1544,8 @@ public class SuperTiles{
return data;
}
// calculate "tilted" disparity, so planes parallel to the same world plane would have the same disparity
// also produces non-tilted, if world_plane_norm == null
public double [][][][] getPlaneDispStrengths(
final double [] world_plane_norm, // real world normal vector to a suggested plane family (0,1,0) for horizontal planes
final int stMeasSel, // = 1; // Select measurements for supertiles : +1 - combo, +2 - quad +4 - hor +8 - vert
......@@ -1622,42 +1625,523 @@ public class SuperTiles{
true); // boolean null_if_none);
}
}
// find average disparity for the supertile (improve?)
double sd = 0.0, sw = 0.0;
for (int ml = 0; ml < plane_disp_strength[nsTile].length; ml++) if ((stMeasSel & ( 1 << ml)) != 0){
for (int i = 0; i < plane_disp_strength[nsTile][ml][1].length; i++){
double w = plane_disp_strength[nsTile][ml][1][i];
double d = plane_disp_strength[nsTile][ml][0][i];
sd += w * d;
sw += w;
if (world_plane_norm != null) {
// find average disparity for the supertile (improve?)
double sd = 0.0, sw = 0.0;
for (int ml = 0; ml < plane_disp_strength[nsTile].length; ml++) if ((stMeasSel & ( 1 << ml)) != 0){
for (int i = 0; i < plane_disp_strength[nsTile][ml][1].length; i++){
double w = plane_disp_strength[nsTile][ml][1][i];
double d = plane_disp_strength[nsTile][ml][0][i];
sd += w * d;
sw += w;
}
}
if (sw > 0) {
if (dl > 0) {
System.out.println("Plane tilted disparity for stileX = "+stileX+" stileY="+stileY+", average disparity "+(sd/sw));
}
plane_disp_strength[nsTile] = pd0.getDisparityToPlane(
world_plane_norm, // double [] world_normal_xyz,
sd / sw, // average disparity // double disp_center,
null, // boolean [][] tile_sel, // null - do not use, {} use all (will be modified)
plane_disp_strength[nsTile], // double [][][] disp_str, // calculate just once if null
1); // int debugLevel);
if (dl>1) {
String [] dbg_titles = showSupertileSeparationTitles( plane_disp_strength[nsTile], null);
double [][] dbg_img = showSUpertileSeparation(plane_disp_strength[nsTile], null); // plane_sels);
showDoubleFloatArrays sdfa_instance = new showDoubleFloatArrays();
sdfa_instance.showArrays(dbg_img, 2 * superTileSize, 2* superTileSize, true, "plane_separation"+nsTile,dbg_titles);
}
} else {
plane_disp_strength[nsTile] = null;
}
}
if (sw > 0) {
}
}
};
}
ImageDtt.startAndJoin(threads);
return plane_disp_strength;
}
// use histogram data (min/max/strength) assign tiles to clusters
// returns [supertile] [ plane number] [measurement layer] [tile index]
public boolean [][][][] dispClusterize(
final double [][][][] disparity_strengths, // either normal or tilted disparity/strengths
final double [][][] hist_max_min_max, // histogram data: per tile array of odd number of disparity/strengths pairs, starting with first maximum
final boolean [][][] selected, // tiles OK to be assigned [supertile][measurement layer] [tile index] or null (or null or per-measurement layer)
final boolean [][][] prohibited, // already assigned tiles [supertile][measurement layer] [tile index] or null
final int stMeasSel, // = 1; // Select measurements for supertiles : +1 - combo, +2 - quad +4 - hor +8 - vert
final double max_diff, // maximal disparity difference (to assign to a cluster (of Double.NaN)
final int plMinPoints, // = 5; // Minimal number of points for plane detection
final double smallDiff, // = 0.4; // Consider merging initial planes if disparity difference below
final double highMix, //stHighMix = 0.4; // Consider merging initial planes if jumps between ratio above
final int debugLevel,
final int dbg_X,
final int dbg_Y)
{
final int tilesX = tileProcessor.getTilesX();
final int tilesY = tileProcessor.getTilesY();
final int superTileSize = tileProcessor.getSuperTileSize();
final int stilesX = (tilesX + superTileSize -1)/superTileSize;
final int stilesY = (tilesY + superTileSize -1)/superTileSize;
final int nStiles = stilesX * stilesY;
final Thread[] threads = ImageDtt.newThreadArray(tileProcessor.threadsMax);
final AtomicInteger ai = new AtomicInteger(0);
this.planes = new TilePlanes.PlaneData[nStiles][];
final int debug_stile = (debugLevel > -1)? (dbg_Y * stilesX + dbg_X):-1;
final boolean [][][][] plane_selections = new boolean [nStiles][][][]; // [supertile] [ plane number] [measurement layer] [tile index]
final double max_diff2 = Double.isNaN(max_diff)? Double.NaN: (max_diff*max_diff);
for (int ithread = 0; ithread < threads.length; ithread++) {
threads[ithread] = new Thread() {
public void run() {
for (int nsTile = ai.getAndIncrement(); nsTile < nStiles; nsTile = ai.getAndIncrement()) {
if (nsTile == debug_stile){
System.out.println("dispClusterize(): nsTile="+nsTile);
}
int stileY = nsTile / stilesX;
int stileX = nsTile % stilesX;
int dl = (nsTile == debug_stile) ? 3 : 0;
double[][][] disp_strength = new double[measuredLayers.getNumLayers()][][];
for (int ml = 0; ml < disp_strength.length; ml++) if ((stMeasSel & ( 1 << ml)) != 0){
disp_strength[ml] = disparity_strengths[nsTile][ml]; // will we change it - no, no need to clone
}
if (hist_max_min_max[nsTile] == null){
continue;
}
double [][] max_only = new double [(hist_max_min_max[nsTile].length + 1)/2][2];
for (int i = 0; i < max_only.length; i++){
max_only[i] = hist_max_min_max[nsTile][2 * i];
}
boolean [][][] plane_sels = null;
int num_ml = disp_strength.length;
int num_tiles = 4 * superTileSize * superTileSize;
int [] num_sel;
boolean [][] tile_en = new boolean [num_ml][];
for (int ml = 0; ml < num_ml; ml++) if (
(disp_strength[ml] != null) &&
((stMeasSel & ( 1 << ml)) != 0) &&
((selected == null) || ((selected[nsTile] != null) && (selected[nsTile][ml] != null)))
) {
tile_en[ml] = new boolean [num_tiles];
// apply all restrictions here
for (int i = 0; i < num_tiles; i++){
if ( (disp_strength[ml][1][i] > 0.0) &&
((selected == null) || (selected[nsTile][ml].length == 0) || selected[nsTile][ml][i]) &&
((prohibited == null) || (prohibited[nsTile] == null) || (prohibited[nsTile][ml] == null) || !prohibited[nsTile][ml][i])
) {
tile_en[ml][i] = true;
}
}
}
for (int iter = 0; iter < 2; iter ++){
int num_p = max_only.length;
plane_sels = new boolean[num_p][num_ml][];
num_sel = new int [num_p];
for (int np = 0; np < num_p; np++) {
for (int ml = 0; ml < num_ml; ml++) if (
(disp_strength[ml] != null) &&
((stMeasSel & ( 1 << ml)) != 0) &&
((selected == null) || ((selected[nsTile] != null) && (selected[nsTile][ml] != null)))
) {
plane_sels[np][ml] = new boolean[num_tiles];
}
}
// compare closest to be able to use tilted planes later
for (int ml = 0; ml < num_ml; ml++) if (tile_en[ml] != null) {
for (int indx = 0; indx < num_tiles; indx++) if (tile_en[ml][indx]){
int best_plane = -1;
double best_d2 = Double.NaN;
for (int np = 0; np < num_p; np++) {
double d2 = max_only[np][0] - disp_strength[ml][0][indx];
// add disp_norm correction here?
d2 *= d2;
if (!(d2 >= best_d2)){
best_d2 = d2;
best_plane = np;
}
}
if (best_plane >= 0){ // compare to max diff here too
if (!(best_d2 > max_diff2)) { // works if max_diff2 is Double.NaN
plane_sels[best_plane][ml][indx] = true; // so far exclusive
}
}
}
}
// recalculate average disparities for each plane and show number of tiles in each in debug mode
for (int np = 0; np < num_p; np++) {
double sd = 0.0, sw = 0.0;
for (int ml = 0; ml < num_ml; ml++) if (disp_strength[ml] != null) {
for (int indx = 0; indx < num_tiles; indx++) if (plane_sels[np][ml][indx]){
double w = disp_strength[ml][1][indx];
sd += w * disp_strength[ml][0][indx];
sw += w;
num_sel[np]++;
}
}
if (sw > 0) {
sd /= sw;
}
if (dl > 0) {
System.out.println("plane num_sel["+np+"] = "+num_sel[np]+" disp "+max_only[np][0]+"->"+sd+
", weight "+max_only[np][1]+"->"+sw);
}
max_only[np][0] = sd;
max_only[np][1] = sw;
}
// calculate transitions matrix (to find candidates for merge
int [][] trans_mat = getTransMatrix(plane_sels);
double [][] rel_trans = getTransRel(trans_mat);
if (dl > 0) {
System.out.println("Plane tilted disparity for stileX = "+stileX+" stileY="+stileY+", average disparity "+(sd/sw));
System.out.println("trans_mat = ");
for (int i = 0; i < trans_mat.length; i++){
System.out.print(i+": ");
for (int j = 0; j < trans_mat[i].length; j++){
System.out.print(trans_mat[i][j]+" ");
}
System.out.println();
}
System.out.println("rel_trans = ");
for (int i = 0; i < rel_trans.length; i++){
System.out.print(i+": ");
for (int j = 0; j < rel_trans[i].length; j++){
System.out.print(rel_trans[i][j]+" ");
}
System.out.println();
}
}
plane_disp_strength[nsTile] = pd0.getDisparityToPlane(
world_plane_norm, // double [] world_normal_xyz,
sd / sw, // average disparity // double disp_center,
null, // boolean [][] tile_sel, // null - do not use, {} use all (will be modified)
plane_disp_strength[nsTile], // double [][][] disp_str, // calculate just once if null
1); // int debugLevel);
if (dl>1) {
String [] dbg_titles = showSupertileSeparationTitles( plane_disp_strength[nsTile], null);
double [][] dbg_img = showSUpertileSeparation(plane_disp_strength[nsTile], null); // plane_sels);
showDoubleFloatArrays sdfa_instance = new showDoubleFloatArrays();
sdfa_instance.showArrays(dbg_img, 2 * superTileSize, 2* superTileSize, true, "plane_separation"+nsTile,dbg_titles);
if ((iter > 0 ) && (num_p > 1)){ // remove /join bad
int windx = 0;
int remove_indx = -1;
for (int i = 1; i < num_p; i++) if (num_sel[i] < num_sel[windx]) windx = i;
if (num_sel[windx] < plMinPoints) {
if (debugLevel > 0){
System.out.println ("dispClusterize(): stileX = "+stileX+" stileY="+stileY+
": removing plane "+windx+" with "+num_sel[windx]+" tiles ( <"+plMinPoints+")");
}
remove_indx = windx;
}
if (remove_indx < 0) {
// find candidates for merge
windx = -1;
for (int i = 0; i < (num_p - 1); i++) {
if (((max_only[i+1][0] - max_only[i][0]) < smallDiff) && // close enough to consider merging
(rel_trans[i][i+1] > highMix)) {
if ((windx < 0) || (rel_trans[i][i+1] > rel_trans[windx][windx+1])) windx = i;
}
}
if (windx >=0 ) {
if (debugLevel > 0){
System.out.println ("dispClusterize(): stileX = "+stileX+" stileY="+stileY+
": merging plane "+windx+" with " + (windx + 1)+": "+
num_sel[windx] + " and "+num_sel[windx+1]+" tiles, "+
" rel_trans="+rel_trans[windx][windx + 1]+ " ( > " + highMix+"),"+
" diff="+ (max_only[windx + 1][0]- max_only[windx][0]) + " ( < " + smallDiff+" ),"+
" disp1 = "+max_only[windx][0]+" disp2 = "+max_only[windx + 1][0]);
}
double sum_w = max_only[windx][1] + max_only[windx + 1][1];
max_only[windx+1][0] = (max_only[windx][0]*max_only[windx][1] + max_only[windx+1][0]*max_only[windx+1][1]) / sum_w;
max_only[windx+1][1] = sum_w;
remove_indx = windx;
}
}
if (remove_indx >= 0){
double [][] max_only_copy = max_only.clone();
for (int i = 0; i < max_only.length; i++) max_only_copy[i] = max_only[i];
max_only = new double [max_only.length - 1][];
int indx = 0;
for (int i = 0; i < max_only_copy.length; i++) if (i != remove_indx) max_only[indx++] =max_only_copy[i];
iter = 0;
continue; // restart from 0
}
// Show other candidates for merge
if (debugLevel > 0){
double max_sep = 0.2;
if (iter > 0) {
for (int i = 0; i < (num_p-1); i++){
if (rel_trans[i][i+1] > max_sep) {
System.out.println("dispClusterize() stileX = "+stileX+" stileY="+stileY+" lowplane = "+i+
" num_sel1 = "+num_sel[i] + " num_sel2 = "+num_sel[i+1] +
" rel_trans="+rel_trans[i][i+1]+
" diff="+ (max_only[i+1][0]- max_only[i][0]) +
" disp1 = "+max_only[i][0]+" disp2 = "+max_only[i+1][0]);
}
}
}
}
}
} else {
plane_disp_strength[nsTile] = null;
}
if (dl > 2) {
String [] dbg_titles = showSupertileSeparationTitles( disp_strength, plane_sels);
double [][] dbg_img = showSUpertileSeparation(disp_strength, plane_sels);
showDoubleFloatArrays sdfa_instance = new showDoubleFloatArrays();
sdfa_instance.showArrays(dbg_img, 2 * superTileSize, 2* superTileSize, true, "disp_clusterize"+nsTile,dbg_titles);
}
plane_selections[nsTile] = plane_sels;
}
}
};
}
ImageDtt.startAndJoin(threads);
return plane_disp_strength;
return plane_selections;
}
// use both horizontal and const disparity tiles to create tile clusters
// Add max_diff (maximal disparity difference while extracting initial tile selection) and max_tries (2..3) parameters
// Add separate method to create + remove outliers from all planes (2 different ones)?
// TODO later re-assign pixels according to existing plane parameters
// Sort plane data by center (plane or supertile) disparity
public boolean [][][][] initialDiscriminateTiles(
final int max_tries, // on last run - assign all rfemaining pixels to some cluster (disregard max_diff)
final int stMeasSel, // = 1; // Select measurements for supertiles : +1 - combo, +2 - quad +4 - hor +8 - vert
final double plDispNorm,
final int plMinPoints, // = 5; // Minimal number of points for plane detection
final boolean plPreferDisparity, // Always start with disparity-most axis (false - lowest eigenvalue)
final GeometryCorrection geometryCorrection,
final boolean correct_distortions,
final boolean smplMode, // = true; // Use sample mode (false - regular tile mode)
final int smplSide, // = 2; // Sample size (side of a square)
final int smplNum, // = 3; // Number after removing worst
final double smplRms, // = 0.1; // Maximal RMS of the remaining tiles in a sample
final double max_diff, // maximal disparity difference (to assign to a cluster (of Double.NaN) at first run
final double smallDiff, // = 0.4; // Consider merging initial planes if disparity difference below
final double highMix, //stHighMix = 0.4; // Consider merging initial planes if jumps between ratio above
final double [] vertical_xyz, // real world up unit vector in camera CS (x - right, y - up, z - to camera};
final int debugLevel,
final int dbg_X,
final int dbg_Y)
{
final int tilesX = tileProcessor.getTilesX();
final int tilesY = tileProcessor.getTilesY();
final int superTileSize = tileProcessor.getSuperTileSize();
final int tileSize = tileProcessor.getTileSize();
final int stilesX = (tilesX + superTileSize -1)/superTileSize;
final int stilesY = (tilesY + superTileSize -1)/superTileSize;
final int nStiles = stilesX * stilesY;
final Thread[] threads = ImageDtt.newThreadArray(tileProcessor.threadsMax);
final AtomicInteger ai = new AtomicInteger(0);
this.planes = new TilePlanes.PlaneData[nStiles][];
final int debug_stile = (debugLevel > -1)? (dbg_Y * stilesX + dbg_X):-1;
// TODO: Remove when promoting PlaneData
final TilePlanes tpl = new TilePlanes(tileSize,superTileSize, geometryCorrection);
final int num_layers = measuredLayers.getNumLayers();
final int num_tiles = 4 * superTileSize * superTileSize;
measuredLayers.setLayer (
0, // int num_layer,
cltPass3d.getDisparity(), // double [] disparity,
cltPass3d.getStrength(), // double [] strength,
null); // boolean [] selection) // may be null
if (debugLevel > -1) {
String [] titles = {"d0","s0","d1","s1","d2","s2","d3","s3","s","d"};
double [][] dbg_img = new double [titles.length][];
for (int i = 0; i < measuredLayers.getNumLayers(); i++){
dbg_img[2 * i] = measuredLayers.getDisparity(i);
dbg_img[2 * i + 1] = measuredLayers.getStrength(i);
}
dbg_img[8] = cltPass3d.getDisparity();
dbg_img[9] = cltPass3d.getStrength();
showDoubleFloatArrays sdfa_instance = new showDoubleFloatArrays();
sdfa_instance.showArrays(dbg_img, tileProcessor.getTilesX(), tileProcessor.getTilesY(), true, "measuredLayers",titles);
}
double [] world_hor = {0.0, 1.0, 0.0};
final double [][][][] hor_disp_strength = getPlaneDispStrengths(
world_hor, // final double [] world_plane_norm, // real world normal vector to a suggested plane family (0,1,0) for horizontal planes
stMeasSel, //final int stMeasSel, // = 1; // Select measurements for supertiles : +1 - combo, +2 - quad +4 - hor +8 - vert
plPreferDisparity, // final boolean plPreferDisparity, // Always start with disparity-most axis (false - lowest eigenvalue)
geometryCorrection, // final GeometryCorrection geometryCorrection,
correct_distortions, // final boolean correct_distortions,
smplMode, // final boolean smplMode, // = true; // Use sample mode (false - regular tile mode)
smplSide, //final int smplSide, // = 2; // Sample size (side of a square)
smplNum, //final int smplNum, // = 3; // Number after removing worst
smplRms, //final double smplRms, // = 0.1; // Maximal RMS of the remaining tiles in a sample
debugLevel,
dbg_X,
dbg_Y);
final double [][][][] vert_disp_strength = getPlaneDispStrengths(
null, // final double [] world_plane_norm, // real world normal vector to a suggested plane family (0,1,0) for horizontal planes
stMeasSel, //final int stMeasSel, // = 1; // Select measurements for supertiles : +1 - combo, +2 - quad +4 - hor +8 - vert
plPreferDisparity, // final boolean plPreferDisparity, // Always start with disparity-most axis (false - lowest eigenvalue)
geometryCorrection, // final GeometryCorrection geometryCorrection,
correct_distortions, // final boolean correct_distortions,
smplMode, // final boolean smplMode, // = true; // Use sample mode (false - regular tile mode)
smplSide, //final int smplSide, // = 2; // Sample size (side of a square)
smplNum, //final int smplNum, // = 3; // Number after removing worst
smplRms, //final double smplRms, // = 0.1; // Maximal RMS of the remaining tiles in a sample
debugLevel,
dbg_X,
dbg_Y);
String [] dbg_hist_titles = {"all","hor","mm_vert","mm_hor"};
double [][] dbg_hist = new double [dbg_hist_titles.length][];
final boolean [][][] used = new boolean [nStiles][num_layers][]; // num_tiles
final boolean [][][][] planes_selections = new boolean [nStiles][][][]; // num_tiles
for (int pass = 0; pass < max_tries; pass ++) {
final int fpass = pass;
// get both horizontal planes and constant disparity planes histograms,
resetDisparityHistograms();
final double [][][] mmm_hor = getMaxMinMax(
hor_disp_strength, // final double [][][][] disparity_strength, // pre-calculated disparity/strength [per super-tile][per-measurement layer][2][tiles] or null
null); // final boolean [][] tile_sel // null or per-measurement layer, per-tile selection. For each layer null - do not use, {} - use all
if (debugLevel > -1) {
dbg_hist[1] = showDisparityHistogram();
dbg_hist[3] = showMaxMinMax();
}
resetDisparityHistograms();
final double [][][] mmm_vert = getMaxMinMax(
vert_disp_strength, // final double [][][][] disparity_strength, // pre-calculated disparity/strength [per super-tile][per-measurement layer][2][tiles] or null
null); // final boolean [][] tile_sel // null or per-measurement layer, per-tile selection. For each layer null - do not use, {} - use all
if (debugLevel > -1) {
dbg_hist[0] = showDisparityHistogram();
dbg_hist[2] = showMaxMinMax();
}
if (debugLevel > -1) {
int hist_width0 = showDisparityHistogramWidth();
int hist_height0 = dbg_hist[0].length/hist_width0;
showDoubleFloatArrays sdfa_instance = new showDoubleFloatArrays(); // just for debugging?
sdfa_instance.showArrays(dbg_hist, hist_width0, hist_height0, true, "vert_hor_histograms_"+pass,dbg_hist_titles);
}
// try to independently (same selections) clusterize both ways
final boolean [][][][] new_planes_hor = dispClusterize(
hor_disp_strength, // final double [][][][] disparity_strengths, // either normal or tilted disparity/strengths
mmm_hor, // final double [][][] hist_max_min_max, // histogram data: per tile array of odd number of disparity/strengths pairs, starting with first maximum
null, // final boolean [][][] selected, // tiles OK to be assigned [supertile][measurement layer] [tile index] or null (or null or per-measurement layer)
used, // final boolean [][][] prohibited, // already assigned tiles [supertile][measurement layer] [tile index] or null
stMeasSel, // final int stMeasSel, // = 1; // Select measurements for supertiles : +1 - combo, +2 - quad +4 - hor +8 - vert
((pass < (max_tries - 1)) ? max_diff : Double.NaN), // final double max_diff, // maximal disparity difference (to assign to a cluster (of Double.NaN)
plMinPoints, // final int plMinPoints, // = 5; // Minimal number of points for plane detection
smallDiff, // final double smallDiff, // = 0.4; // Consider merging initial planes if disparity difference below
highMix, // final double highMix, //stHighMix = 0.4; // Consider merging initial planes if jumps between ratio above
debugLevel,
dbg_X,
dbg_Y);
final boolean [][][][] new_planes_vert = dispClusterize(
vert_disp_strength, // final double [][][][] disparity_strengths, // either normal or tilted disparity/strengths
mmm_vert, // final double [][][] hist_max_min_max, // histogram data: per tile array of odd number of disparity/strengths pairs, starting with first maximum
null, // final boolean [][][] selected, // tiles OK to be assigned [supertile][measurement layer] [tile index] or null (or null or per-measurement layer)
used, // final boolean [][][] prohibited, // already assigned tiles [supertile][measurement layer] [tile index] or null
stMeasSel, // final int stMeasSel, // = 1; // Select measurements for supertiles : +1 - combo, +2 - quad +4 - hor +8 - vert
((pass < (max_tries - 1)) ? max_diff : Double.NaN), // final double max_diff, // maximal disparity difference (to assign to a cluster (of Double.NaN)
plMinPoints, // final int plMinPoints, // = 5; // Minimal number of points for plane detection
smallDiff, // final double smallDiff, // = 0.4; // Consider merging initial planes if disparity difference below
highMix, // final double highMix, //stHighMix = 0.4; // Consider merging initial planes if jumps between ratio above
debugLevel,
dbg_X,
dbg_Y);
// compare which vert or hor provide stronger clusters, add them to selections, recalculate "used"
for (int ithread = 0; ithread < threads.length; ithread++) {
threads[ithread] = new Thread() {
public void run() {
for (int nsTile = ai.getAndIncrement(); nsTile < nStiles; nsTile = ai.getAndIncrement()) {
if (nsTile == debug_stile){
System.out.println("processPlanes5(): nsTile="+nsTile);
}
int stileY = nsTile / stilesX;
int stileX = nsTile % stilesX;
// int [] sTiles = {stileX, stileY};
double [][][][] ds = {vert_disp_strength[nsTile],hor_disp_strength[nsTile]};
boolean [][][][] sels_all = {new_planes_vert[nsTile],new_planes_hor[nsTile]}; // make possible to iterate
class SelStrength{
int type;
int indx;
double strength;
SelStrength(int type, int indx, double strength){
this.type = type;
this.indx = indx;
this.strength = strength;
}
}
// double [][] strengths = new double [2][];
ArrayList<SelStrength> selStrengthList = new ArrayList <SelStrength>();
// int [] best_planes = {-1,-1};
for (int pType = 0; pType < sels_all.length; pType++){
if (sels_all[pType] == null) sels_all[pType] = new boolean[0][][];
// strengths[pType] = new double [sels_all[pType].length]; // number of planes
// calculate strength of each selection;
for (int np = 0; np < sels_all[pType].length; np ++) {
double sw = 0.0;
for (int ml = 0; ml < num_layers; ml++) if ((sels_all[pType][np] != null) && (sels_all[pType][np][ml] != null)){
for (int i = 0; i < num_tiles; i++) {
if (sels_all[pType][np][ml][i]) {
sw += ds[pType][ml][1][i];
}
}
}
selStrengthList.add(new SelStrength(pType, np, sw));
}
}
if (selStrengthList.size() > 0) {
Collections.sort(selStrengthList, new Comparator<SelStrength>() {
@Override
public int compare(SelStrength lhs, SelStrength rhs) {
// -1 - less than, 1 - greater than, 0 - equal, all inverted for descending
return (lhs.strength > rhs.strength) ? -1 : (lhs.strength < rhs.strength ) ? 1 : 0;
}
});
int best_type = selStrengthList.get(0).type;
int num_planes = 0;
for (; num_planes < selStrengthList.size(); num_planes++) {
if (selStrengthList.get(num_planes).type != best_type) {
break;
}
}
int num_old_planes = (planes_selections[nsTile] == null) ? 0: planes_selections[nsTile].length;
boolean [][][] new_planes_selections = new boolean [num_old_planes + num_planes][][];
int np = 0;
for (; np < num_old_planes; np++){
new_planes_selections[np] = planes_selections[nsTile][np];
}
for (SelStrength ss:selStrengthList){
new_planes_selections[np++] = sels_all[ss.type][ss.indx];
for (int ml = 0; ml < num_layers; ml++) if (sels_all[ss.type][ss.indx][ml] != null){
if (used[nsTile][ml] == null){
used[nsTile][ml] = new boolean[num_tiles];
}
for (int i = 0; i < num_tiles; i++){
used[nsTile][ml][i] |= sels_all[ss.type][ss.indx][ml][i];
}
}
}
}
}
}
};
}
ImageDtt.startAndJoin(threads);
}
return planes_selections;
}
public void processPlanes4(
// final boolean [] selected, // or null
......@@ -2125,7 +2609,6 @@ public class SuperTiles{
}
public int [] getShowPlanesWidthHeight()
{
final int tilesX = tileProcessor.getTilesX();
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment