Skip to content

I
imagej-elphel
Commit b9cc7d77 authored by Andrey Filippov's avatar Andrey Filippov
Browse files
Options

Fixing disparity scaling (GPU only so far) between diameter cameras and 

quad (square)
parent 9fb7ab22
Hide whitespace changes
Inline Side-by-side
Showing with 524 additions and 37 deletions
src/main/java/com/elphel/imagej/tileprocessor/Corr2dLMA.java
View file @ b9cc7d77
......@@ -740,6 +740,8 @@ public class Corr2dLMA {
/**
* Calculate initial disparity by polynomial approximation. Only works with a single tile now
* Note: combo correlation scale is one pixel for 1 pixel disparity for diameter cameras, while standard
* (pre-shift) disparity refers to a quad-camera (square) pixel offset. So the returned disparity should be divided by sqrt(2)
* @param corr_wnd_inv_limited window (same as for finding convex) to boost gain in peripheral areas, but not the very marginal ones
* @param max_offset maximal abs(disparity) value to trust
* @param dgg_title if !=null - generate debug image with this title
......
src/main/java/com/elphel/imagej/tileprocessor/Correlation2d.java
View file @ b9cc7d77
......@@ -577,7 +577,7 @@ public class Correlation2d {
public int getCombOffset() {return mcorr_comb_offset;}
public double getCombDisp() {return mcorr_comb_disp;}
@Deprecated
public void generateResampleOld( // should be called before
final int mcorr_comb_width, // combined correlation tile width
final int mcorr_comb_height, // combined correlation tile full height
......@@ -714,7 +714,7 @@ public class Correlation2d {
final int mcorr_comb_width, // combined correlation tile width
final int mcorr_comb_height, // combined correlation tile full height
final int mcorr_comb_offset, // combined correlation tile height offset: 0 - centered (-height/2 to height/2), height/2 - only positive (0 to height)
final double mcorr_comb_disp){ // Combined tile per-pixel disparity for baseline == side of a square
final double mcorr_comb_disp){ // Combined tile per-pixel disparity for baseline == side of a square -> diameter !
final double ignore_contrib = 0.001; // ignore contributors with weight below
this.mcorr_comb_width = mcorr_comb_width; // combined correlation tile width
this.mcorr_comb_height = mcorr_comb_height; // combined correlation tile full height
......@@ -2405,12 +2405,15 @@ public class Correlation2d {
/**
* Analyze 1d correlation (single centerline of the 3D phase correlation combined output
* from all pairs or only diameters to detect double maximum (simultaneous FG+BG)
* from all pairs or only diameters to detect double maximum (simultaneous FG+BG)
* @param disparity_scale multiply pixel coordinates in combo_corrs to match bominal disparity
* pixels (measured for quad square camera). It is now 1/sqrt(2)
* @param combo_corrs 2D phase correlation, now 15x15 = 255 pixels long
* @param min_fraction minimal ratio of weaker maximum to the strongest
* @return array of 1 or 2 {disparity, strength} pairs (zero pairs if no local max)
*/
public double [][] getDoublePoly(
double disparity_scale,
double [] combo_corrs,
double min_fraction
){
......@@ -2439,7 +2442,7 @@ public class Correlation2d {
double c = combo_corrs[imx[i]];
double a = (combo_corrs[imx[i] + 1] + combo_corrs[imx[i] - 1]) / 2 - c;
double b = (combo_corrs[imx[i] + 1] - combo_corrs[imx[i] - 1]);
maxes[i][0] = imx[i]- center_x - 0.5 * b / a;
maxes[i][0] = -disparity_scale * (imx[i]- center_x - 0.5 * b / a); // disparity, not x!
maxes[i][1] = c - 0.25 * b * b / a;
}
return maxes;
......@@ -3921,6 +3924,7 @@ public class Correlation2d {
tileY);
}
public Corr2dLMA corrLMA2Single( // single tile
ImageDttParameters imgdtt_params,
boolean adjust_ly, // adjust Lazy Eye
......@@ -3931,6 +3935,7 @@ public class Correlation2d {
double [][] rXY, // non-distorted X,Y offset per nominal pixel of disparity
boolean [] pair_mask, // which pairs to process
double[] disp_str, // -preliminary center x in pixels for largest baseline
// Not anymore, it is expressed in quad camera disparity (-x offset divided by sqrt(2)!
double[] poly_ds, // null or pair of disparity/strength
double vasw_pwr, // value as weight to this power,
double [] debug_lma_tile,
......@@ -3970,7 +3975,10 @@ public class Correlation2d {
pair_offsets = lma.getPairsOffsets(
corrs, // double [][] corrs,
pair_mask, // boolean [] pair_mask,
disp_str[0]/imgdtt_params.lmamask_magic, // double disparity,
// disp_str[0]/imgdtt_params.lmamask_magic, // double disparity,
// Moved to the caller
/// disp_str[0]/Math.sqrt(2), // *Math.sqrt(2), // double disparity,
disp_str[0], // *Math.sqrt(2), // double disparity,
disp_dist); // double [][] disp_dist);
corr_shape = getCorrShape(
corrs, // double [][] corrs,
......@@ -4099,6 +4107,454 @@ public class Correlation2d {
for (int i = 1; i < samplesWeight[npair].length; i++) if (samplesWeight[npair][i] > 0.0) {
int ix = i % corr_size; // >=0
int iy = i / corr_size; // >=0
double v = corrs[npair][i]; // not blurred
double w = samplesWeight[npair][i];
if (vasw_pwr != 0) {
w *= Math.pow(Math.abs(v), vasw_pwr);
}
lma.addSample( // x = 0, y=0 - center
0, // tile
npair,
ix, // int x, // x coordinate on the common scale (corresponding to the largest baseline), along the disparity axis
iy, // int y, // y coordinate (0 - disparity axis)
v, // double v, // correlation value at that point
w); //double w) // sample weight
}
}
if (debug_lma_tile != null) { // calculate and return number of non-zero tiles
debug_lma_tile[0] = num_disp_samples;
debug_lma_tile[1] = num_cnvx_samples;
debug_lma_tile[2] = num_comb_samples;
debug_lma_tile[3] = -1; // number of LMA iterations
debug_lma_tile[4] = -1; // last number of LMA iterations
debug_lma_tile[5] = -1; // LMA RMA
}
if (debug_graphic) {
if (dbg_corr != null) {
(new ShowDoubleFloatArrays()).showArrays(
dbg_corr,
corr_size,
corr_size,
true,
"corr_blurred"+"_x"+tileX+"_y"+tileY,
getCorrTitles());
}
if (filtWeight != null) {
(new ShowDoubleFloatArrays()).showArrays(
filtWeight,
corr_size,
corr_size,
true,
"filt_weight"+"_x"+tileX+"_y"+tileY,
getCorrTitles());
}
if (samplesWeight != null) {
(new ShowDoubleFloatArrays()).showArrays(
samplesWeight,
corr_size,
corr_size,
true,
"samples_weights"+"_x"+tileX+"_y"+tileY,
getCorrTitles());
}
}
// double [][] disp_str = {{xcenter, 1.0}}; // temporary
double [][] disp_str2 = {{0.0, 1.0}}; // temporary // will be calculated/set later
if (disp_str != null) {
disp_str2[0] = disp_str;
}
boolean lmaSuccess = false;
int num_lma_retries = 0;
double [] disp = null;
// adjust_ly
double [][] ly_offsets_pairs = null;
if (adjust_ly) {
ly_offsets_pairs = getPairsCenters(
corrs, // double [][] corrs,
samplesWeight); // double [][] weights)
}
double step_weight = 0.5; // scale corrections
double min_correction = 0.1; // exit when maximal XY correction is below
while (!lmaSuccess) {
num_lma_retries ++; // debug
// FIXME: ugly fix
double [][] disp_str2_scaled = disp_str2.clone();
for (int i = 0; i < disp_str2_scaled.length; i++) {
if (disp_str2_scaled[i] != null) {
disp_str2_scaled[i] = disp_str2_scaled[i].clone();
disp_str2_scaled[i][0] /= imgdtt_params.lmamask_magic;
}
}
lma.initVector(
imgdtt_params.lmas_adjust_wm, // boolean adjust_width, // adjust width of the maximum - lma_adjust_wm
imgdtt_params.lmas_adjust_ag, // boolean adjust_scales, // adjust 2D correlation scales - lma_adjust_ag
imgdtt_params.lmas_adjust_wy, // boolean adjust_ellipse, // allow non-circular correlation maximums lma_adjust_wy
(adjust_ly ? imgdtt_params.lma_adjust_wxy : false), //imgdtt_params.lma_adjust_wxy, // boolean adjust_lazyeye_par, // adjust disparity corrections parallel to disparities lma_adjust_wxy
(adjust_ly ? imgdtt_params.lma_adjust_ly1: false), // imgdtt_params.lma_adjust_ly1, // boolean adjust_lazyeye_ortho, // adjust disparity corrections orthogonal to disparities lma_adjust_ly1
disp_str2_scaled, // xcenter,
imgdtt_params.lma_half_width, // double half_width, // A=1/(half_widh)^2 lma_half_width
(adjust_ly ? imgdtt_params.lma_cost_wy : 0.0), // imgdtt_params.lma_cost_wy, // double cost_lazyeye_par, // cost for each of the non-zero disparity corrections lma_cost_wy
(adjust_ly ? imgdtt_params.lma_cost_wxy : 0.0) //imgdtt_params.lma_cost_wxy // double cost_lazyeye_odtho // cost for each of the non-zero ortho disparity corrections lma_cost_wxy
);
lma.setMatrices(disp_dist);
lma.initMatrices(); // should be called after initVector and after setMatrices
boolean all_sensors_used = lma.setInitialLYOffsets(
ly_offsets_pairs, // double [][] pair_centers,
step_weight, // double step_weight, // scale corrections
min_correction, // double min_correction ){ // exit when maximal XY correction is below
(debug_level > 0)); //
if (adjust_ly && !all_sensors_used) {
return null; //LY requested, but not all sensors present
}
//center
disp = null;
if (need_poly) {
disp = lma.polyDisparity(
corr_wnd_inv_limited,
transform_size-1-imgdtt_params.lma_soft_marg,//double max_offset, // 5?
debug_graphic?dbg_title:null); // double [] rslt = {-approx2d[0], approx2d[2], hwx, hwy};
if (disp == null) {
if (imgdtt_params.lmas_poly_continue && (disp == null)) {
disp = disp_str2[0];
if (debug_level > 0) {
System.out.println("Poly disparity=NULL, using tile center for initial LMA");
}
} else {
if (debug_level > 0) {
System.out.println("Poly disparity=NULL, set lmas_poly_continue to true to use tile center instead");
}
}
} else {
disp[1] *= imgdtt_params.lmas_poly_str_scale;
disp[0] /= Math.sqrt(2); // disparity is expressed in pixels of a quad camera, combo correlation is for diameter cameras
if (debug_level > 0) {
System.out.println(String.format("Poly disparity (quad camera scale) =%8.5f , str=%8.5f, disp_str2[0][0]=%8.5f, disp_str2[0][1]=%8.5f",
disp[0],disp[1],disp_str2[0][0],disp_str2[0][1]));
}
if (disp[1] < imgdtt_params.lmas_poly_str_min) {
if (debug_level > 0) {
System.out.println("Poly strength too low ("+disp[1]+" < "+imgdtt_params.lmas_poly_str_min+")");
}
disp = null;
}
}
// double[] poly_ds, // null or pair of disparity/strength
if (poly_ds != null) {
poly_ds[0] = (disp==null) ? Double.NaN: disp[0];
poly_ds[1] = (disp==null) ? 0.0: disp[1];
}
} else {
disp = disp_str;
}
if (disp != null) {
disp_str2[0] = disp;
lma.initDisparity( // USED in lwir null pointer
disp_str2); // double [][] disp_str // initial value of disparity
if (debug_level > 1) {
System.out.println("Input data:");
lma.printInputDataFx(false);
lma.printParams();
}
lmaSuccess = lma.runLma(
imgdtt_params.lmas_lambda_initial, // double lambda, // 0.1
imgdtt_params.lma_lambda_scale_good, // double lambda_scale_good,// 0.5
imgdtt_params.lma_lambda_scale_bad, // double lambda_scale_bad, // 8.0
imgdtt_params.lma_lambda_max, // double lambda_max, // 100
imgdtt_params.lmas_rms_diff, // double rms_diff, // 0.001
imgdtt_params.lmas_num_iter, // int num_iter, // 20
debug_level); // imgdtt_params.lma_debug_level1); // 4); // int debug_level) // > 3
if (!lmaSuccess && (lma.getBadTile() >= 0)) {
if (debug_level > -2) {
System.out.println("Found bad tile/pair during single (probably wrong initial maximum - try around preliminary? "+lma.getBadTile());
}
} else {
break;
}
} else {
break;
}
}
if (lmaSuccess) {
lma.updateFromVector();
double [][] dispStr = lma.lmaDisparityStrength( //TODO: add parameter to filter out negative minimums ?
imgdtt_params.lmas_min_amp, // minimal ratio of minimal pair correlation amplitude to maximal pair correlation amplitude
imgdtt_params.lmas_max_rel_rms, // maximal relative (to average max/min amplitude LMA RMS) // May be up to 0.3)
imgdtt_params.lmas_min_strength, // minimal composite strength (sqrt(average amp squared over absolute RMS)
imgdtt_params.lmas_min_ac, // minimal of A and C coefficients maximum (measures sharpest point/line)
imgdtt_params.lmas_min_min_ac, // minimal of A and C coefficients minimum (measures sharpest point)
imgdtt_params.lmas_max_area, //double lma_max_area, // maximal half-area (if > 0.0)
imgdtt_params.lma_str_scale, // convert lma-generated strength to match previous ones - scale
imgdtt_params.lma_str_offset // convert lma-generated strength to match previous ones - add to result
);
if (dispStr[0][1] <= 0) {
lmaSuccess = false;
if (debug_level > -2) { // 0
System.out.println(String.format("Poly disparity=%8.5f , str=%8.5f", disp[0],disp[1]));
}
if (debug_lma_tile != null) {
debug_lma_tile[3] = num_lma_retries; // number of wasted attempts
debug_lma_tile[4] = lma.getNumIter();
}
} else {
if (debug_level > -2) {
System.out.println(String.format("Poly disparity=%8.5f , str=%8.5f, LMA disparity=%8.5f, str=%8.5f",
disp[0],disp[1],dispStr[0][0],dispStr[0][1]));
}
// System.out.println("dispStr[0][0]="+dispStr[0][0]+" dispStr[0][1]="+dispStr[0][1]);
double [] rms = lma.getRMS();
if (debug_lma_tile != null) {
debug_lma_tile[3] = num_lma_retries; // number of wasted attempts
debug_lma_tile[4] = lma.getNumIter();
debug_lma_tile[5] = rms[1]; // pure rms
}
if (debug_level > 0) {
System.out.println("LMA -> "+lmaSuccess+" RMS="+rms[0]+", pure RMS="+rms[1]);
lma.printParams();
}
if (debug_level > 1) {
System.out.println("Input data and approximation:");
lma.printInputDataFx(true);
}
// double [][] ds = null;
if (debug_graphic && lmaSuccess) {
String [] sliceTitles = lma.dbgGetSliceTitles();
// if (corrs.length == 1) { // only for single-tile cluster (here it is always single, and corrs is double [][], not double [][][]
(new ShowDoubleFloatArrays()).showArrays(
lma.dbgGetSamples(null,0)[0],
corr_size,
corr_size,
true,
"corr_values"+"_x"+tileX+"_y"+tileY, sliceTitles);
(new ShowDoubleFloatArrays()).showArrays(
lma.dbgGetSamples(null,2)[0],
corr_size,
corr_size,
true,
"corr_fx"+"_x"+tileX+"_y"+tileY, sliceTitles);
(new ShowDoubleFloatArrays()).showArrays(
lma.dbgGetSamples(null,1)[0],
corr_size,
corr_size,
true,
"corr_weights_late"+"_x"+tileX+"_y"+tileY, sliceTitles);
// }
}
}
} else if (debug_lma_tile != null) {
debug_lma_tile[3] = num_lma_retries; // number of wasted attempts
}
return lmaSuccess? lma: null;
}
public Corr2dLMA corrLMA2DualMax( // single tile
ImageDttParameters imgdtt_params,
boolean adjust_ly, // adjust Lazy Eye
double [][] corr_wnd, // correlation window to save on re-calculation of the window
double [] corr_wnd_inv_limited, // correlation window, limited not to be smaller than threshold - used for finding max/convex areas (or null)
double [][] corrs, // may have more elements than pair_mask (corrs may have combo as last elements)
double [][] disp_dist, // per camera disparity matrix as a 1d (linescan order)
double [][] rXY, // non-distorted X,Y offset per nominal pixel of disparity
boolean [] pair_mask, // which pairs to process
// should never be null
double[][] disp_str_dual, // -preliminary center x in pixels for largest baseline
double[] poly_ds, // null or pair of disparity/strength
double vasw_pwr, // value as weight to this power,
double [] debug_lma_tile,
int debug_level,
int tileX, // just for debug output
int tileY
)
{
// corrs are organized as PAIRS, some are null if not used
// for each enabled and available pair find a maximum, filter convex and create sample list
// boolean need_poly = (disp_str == null); // true; // find initial disparity by polynomial approximation
boolean debug_graphic = imgdtt_params.lma_debug_graphic && (imgdtt_params.lma_debug_level1 > 3) && (debug_level > 0) ;
debug_graphic |= imgdtt_params.lmamask_dbg && (debug_level > 0);
String dbg_title = null;
if (debug_graphic) {
dbg_title = String.format("tX%d_tY%d",tileX,tileY);
}
DoubleGaussianBlur gb = null;
if (imgdtt_params.lma_sigma > 0) gb = new DoubleGaussianBlur();
int center = transform_size - 1;
int corr_size = 2 * transform_size - 1;
Corr2dLMA lma = new Corr2dLMA(
1,
this, // Correlation2d correlation2d,
transform_size,
corr_wnd,
rXY, //double [][] rXY, // non-distorted X,Y offset per nominal pixel of disparity
imgdtt_params.lmas_gaussian //boolean gaussian_mode
);
double [][] corr_shapes = new double [disp_str_dual.length][];
double [][] corr_shapes_dia = new double [disp_str_dual.length][];
double [][] norm_shapes = new double [disp_str_dual.length][];
double [][][] pair_shapes_masks = new double [disp_str_dual.length][][];
double [][][] pair_offsets = new double [disp_str_dual.length][][];
for (int nmax = 0; nmax < disp_str_dual.length; nmax++) {
pair_offsets[nmax] = lma.getPairsOffsets(
corrs, // double [][] corrs,
pair_mask, // boolean [] pair_mask,
// disp_str_dual[nmax][0]/imgdtt_params.lmamask_magic, // double disparity,
/// disp_str_dual[nmax][0]/Math.sqrt(2), // *Math.sqrt(2), // double disparity,
// sqrt(2) moved to the caller
disp_str_dual[nmax][0], // *Math.sqrt(2), // double disparity,
disp_dist); // double [][] disp_dist);
corr_shapes[nmax] = getCorrShape(
corrs, // double [][] corrs,
pair_offsets[nmax]); // double [][] xy_offsets)
if (debug_graphic) {
int min_dia = 96;
double [][] corrs_dia = new double[corrs.length][];
for (int i = min_dia; i < corrs.length; i++) {
corrs_dia[i] = corrs[i];
}
corr_shapes_dia[nmax] = getCorrShape(
corrs_dia, // double [][] corrs,
pair_offsets[nmax]); // double [][] xy_offsets)
}
norm_shapes[nmax] = conditionCorrShape(
corr_shapes[nmax], // double [] corrs_shape,
imgdtt_params.lmamask_min_main, // double min_main,
imgdtt_params.lmamask_min_neib, // double min_neib,
imgdtt_params.lmamask_weight_neib, // double weight_neib);
imgdtt_params.lmamask_weight_neib_neib); // double weight_neib_neib
pair_shapes_masks[nmax] = applyCorrShape(
norm_shapes[nmax], // double [] corrs_shape,
pair_offsets[nmax]); // double [][] xy_offsets)
}
double [][] dbg_corr = debug_graphic ? new double [corrs.length][] : null;
if (debug_graphic) {
(new ShowDoubleFloatArrays()).showArrays(
corrs,
corr_size,
corr_size,
true,
"corr_pairs"+"_x"+tileX+"_y"+tileY,
getCorrTitles());
if (corr_shapes != null) {
for (int nmax = 0; nmax < disp_str_dual.length; nmax++) {
(new ShowDoubleFloatArrays()).showArrays(
new double [][] {corr_shapes[nmax],norm_shapes[nmax], corr_shapes_dia[nmax]},
corr_size,
corr_size,
true,
"corr_shape"+"_x"+tileX+"_y"+tileY+"_M"+nmax,
new String [] {"corr_shape","norm_shape","corr_shape_dia"});
}
}
if (pair_shapes_masks != null) {
for (int nmax = 0; nmax < disp_str_dual.length; nmax++) {
(new ShowDoubleFloatArrays()).showArrays(
pair_shapes_masks[nmax],
corr_size,
corr_size,
true,
"corr_shape_masks"+"_x"+tileX+"_y"+tileY+"_M"+nmax,
getCorrTitles());
}
}
}
// try alternative mask generation by accumulation of the pre-shifted (from CM estimation with magic 0.85) correlations
/*
double [][] filtWeight = new double [corrs.length][];
double [][] samplesWeight = new double [corrs.length][];
int num_disp_samples = 0;
int num_cnvx_samples = 0;
int num_comb_samples = 0;
for (int npair = 0; npair < pair_mask.length; npair++) if ((corrs[npair] != null) && (pair_mask[npair])){
double [] corr_blur = null;
if (imgdtt_params.cnvx_en || (pair_shape_masks == null)) {
corr_blur = corrs[npair].clone();
if (corr_wnd_inv_limited != null) {
for (int i = 0; i < corr_blur.length; i++) {
corr_blur[i] *= corr_wnd_inv_limited[i];
}
}
if (imgdtt_params.lma_sigma > 0) {
gb.blurDouble(corr_blur, corr_size, corr_size, imgdtt_params.lma_sigma, imgdtt_params.lma_sigma, 0.01);
}
int imx = imgdtt_params.lma_soft_marg * (corr_size + 1);
for (int iy = imgdtt_params.lma_soft_marg; iy < (corr_size - imgdtt_params.lma_soft_marg); iy++) {
for (int ix = imgdtt_params.lma_soft_marg; ix < (corr_size - imgdtt_params.lma_soft_marg); ix++) {
int indx = iy * corr_size + ix;
if (corr_blur[indx] > corr_blur[imx]) imx = indx;
}
}
// filter convex
int ix0 = (imx % corr_size) - center; // signed, around center to match filterConvex
int iy0 = (imx / corr_size) - center; // signed, around center to match filterConvex
filtWeight[npair] = filterConvex(
corr_blur, // double [] corr_data,
imgdtt_params.cnvx_hwnd_size, // int hwin,
ix0, // int x0,
iy0, // int y0,
imgdtt_params.cnvx_add3x3, // boolean add3x3,
imgdtt_params.cnvx_weight, // double nc_cost,
(debug_level > 2)); // boolean debug);
}
if (dbg_corr != null) dbg_corr [npair] = corr_blur;
// Normalize weight for each pair to compensate for different number of convex samples?
// Combine/use window masks
if (filtWeight[npair] == null) {
samplesWeight[npair] = (pair_shape_masks != null)? pair_shape_masks[npair] : null;
} else if ((pair_shape_masks == null) || (pair_shape_masks[npair] == null) || !imgdtt_params.lmamask_en) {
samplesWeight[npair] = filtWeight[npair];
} else {
samplesWeight[npair] = filtWeight[npair].clone();
if (imgdtt_params.cnvx_or) {
for (int i = 0; i < samplesWeight[npair].length; i++) {
samplesWeight[npair][i] = Math.max(samplesWeight[npair][i], pair_shape_masks[npair][i]);
}
} else {
for (int i = 0; i < samplesWeight[npair].length; i++) {
samplesWeight[npair][i] *= pair_shape_masks[npair][i];
}
}
}
if (debug_lma_tile != null) { // calculate and return number of non-zero tiles
if (pair_shape_masks[npair] != null) {
for (int i = 0; i < samplesWeight[npair].length; i++) if (samplesWeight[npair][i] > 0.0) num_disp_samples++;
}
if (filtWeight[npair] != null) {
for (int i = 0; i < filtWeight[npair].length; i++) if (filtWeight[npair][i] > 0.0) num_cnvx_samples++;
}
if (samplesWeight[npair] != null) {
for (int i = 0; i < samplesWeight[npair].length; i++) if (samplesWeight[npair][i] > 0.0) num_comb_samples++;
}
}
for (int i = 1; i < samplesWeight[npair].length; i++) if (samplesWeight[npair][i] > 0.0) {
int ix = i % corr_size; // >=0
int iy = i / corr_size; // >=0
......@@ -4350,6 +4806,8 @@ public class Correlation2d {
debug_lma_tile[3] = num_lma_retries; // number of wasted attempts
}
return lmaSuccess? lma: null;
*/
return null;
}
/**
......
src/main/java/com/elphel/imagej/tileprocessor/ImageDtt.java
View file @ b9cc7d77
......@@ -221,7 +221,7 @@ public class ImageDtt extends ImageDttCPU {
ortho_weights[i] = 0.5*(1.0+Math.cos(Math.PI*dx))/imgdtt_params.ortho_eff_height;
}
}
if (globalDebugLevel > 0){
if (globalDebugLevel > 1){
System.out.println("ortho_height="+ imgdtt_params.ortho_height+" ortho_eff_height="+ imgdtt_params.ortho_eff_height);
for (int i = 0; i < corr_size; i++){
System.out.println(" ortho_weights["+i+"]="+ ortho_weights[i]);
......@@ -229,7 +229,7 @@ public class ImageDtt extends ImageDttCPU {
}
if (globalDebugLevel > 0) {
if (globalDebugLevel > 1) {
System.out.println("clt_aberrations_quad_corr(): width="+width+" height="+height+" transform_size="+transform_size+
" debug_tileX="+debug_tileX+" debug_tileY="+debug_tileY+" globalDebugLevel="+globalDebugLevel);
}
......@@ -1350,7 +1350,7 @@ public class ImageDtt extends ImageDttCPU {
ortho_weights[i] = 0.5*(1.0+Math.cos(Math.PI*dx))/imgdtt_params.ortho_eff_height;
}
}
if (globalDebugLevel > 0){
if (globalDebugLevel > 1){
System.out.println("ortho_height="+ imgdtt_params.ortho_height+" ortho_eff_height="+ imgdtt_params.ortho_eff_height);
for (int i = 0; i < corr_size; i++){
System.out.println(" ortho_weights["+i+"]="+ ortho_weights[i]);
......@@ -1358,7 +1358,7 @@ public class ImageDtt extends ImageDttCPU {
}
if (globalDebugLevel > 0) {
if (globalDebugLevel > 1) {
System.out.println("clt_aberrations_quad_corr(): width="+width+" height="+height+" transform_size="+transform_size+
" debug_tileX="+debug_tileX+" debug_tileY="+debug_tileY+" globalDebugLevel="+globalDebugLevel);
}
......@@ -2015,7 +2015,7 @@ public class ImageDtt extends ImageDttCPU {
ortho_weights[i] = 0.5*(1.0+Math.cos(Math.PI*dx))/imgdtt_params.ortho_eff_height;
}
}
if (globalDebugLevel > 0){
if (globalDebugLevel > 1){
System.out.println("ortho_height="+ imgdtt_params.ortho_height+" ortho_eff_height="+ imgdtt_params.ortho_eff_height);
for (int i = 0; i < corr_size; i++){
System.out.println(" ortho_weights["+i+"]="+ ortho_weights[i]);
......@@ -2023,7 +2023,7 @@ public class ImageDtt extends ImageDttCPU {
}
if (globalDebugLevel > 0) {
if (globalDebugLevel > 1) {
System.out.println("clt_aberrations_quad_corr(): width="+width+" height="+height+" transform_size="+transform_size+
" debug_tileX="+debug_tileX+" debug_tileY="+debug_tileY+" globalDebugLevel="+globalDebugLevel);
}
......@@ -2303,6 +2303,7 @@ public class ImageDtt extends ImageDttCPU {
final int threadsMax, // maximal number of threads to launch
final int globalDebugLevel)
{
final double disparity_scale = 1.0/Math.sqrt(2); // combo pixels -> disparity pixels
final boolean diameters_combo = (imgdtt_params.mcorr_dual_fract > 0.0); // add diameters-only combo after all-combo
if (this.gpuQuad == null) {
System.out.println("clt_aberrations_quad_corr_GPU(): this.gpuQuad is null, bailing out");
......@@ -2364,7 +2365,7 @@ public class ImageDtt extends ImageDttCPU {
ortho_weights[i] = 0.5*(1.0+Math.cos(Math.PI*dx))/imgdtt_params.ortho_eff_height;
}
}
if (globalDebugLevel > 0){
if (globalDebugLevel > 1){
System.out.println("ortho_height="+ imgdtt_params.ortho_height+" ortho_eff_height="+ imgdtt_params.ortho_eff_height);
for (int i = 0; i < corr_size; i++){
System.out.println(" ortho_weights["+i+"]="+ ortho_weights[i]);
......@@ -2372,7 +2373,7 @@ public class ImageDtt extends ImageDttCPU {
}
if (globalDebugLevel > 00) {
if (globalDebugLevel > 1) {
System.out.println("clt_aberrations_quad_corr(): width="+width+" height="+height+" transform_size="+transform_size+
" debug_tileX="+debug_tileX+" debug_tileY="+debug_tileY+" globalDebugLevel="+globalDebugLevel);
}
......@@ -2537,7 +2538,7 @@ public class ImageDtt extends ImageDttCPU {
corrs[npair] , //double [] corr_tile,
npair); // int num_pair)
if (pair_width < 0.1) {
if (globalDebugLevel > 0) {
if (globalDebugLevel > 1) {
System.out.println("pair_width["+npair+"]="+pair_width);
}
} else {
......@@ -2612,6 +2613,7 @@ public class ImageDtt extends ImageDttCPU {
System.out.println("clt_process_tl_correlations(): debugTile1");
}
double [][] maxes = correlation2d.getDoublePoly(
disparity_scale, // double disparity_scale,
((corr_dia_tile != null) ? corr_dia_tile : corr_combo_tile), // double [] combo_corrs,
imgdtt_params.mcorr_dual_fract); //double min_fraction
// TODO: add corr layer - copy of combo with singles as nulls
......@@ -2623,9 +2625,31 @@ public class ImageDtt extends ImageDttCPU {
if (debugTile1) {
System.out.println("clt_process_tl_correlations() maxes=");
for (int i = 0; i < maxes.length; i++) {
System.out.println(String.format("maxes[%d][0]=%f, maxes[%d][1]=%f", i, maxes[i][0], i, maxes[i][1]));
System.out.println(String.format("maxes[%d][0]=%f (quadcam disparity pixels, not combo pixels), maxes[%d][1]=%f", i, maxes[i][0], i, maxes[i][1]));
}
}
if (debugTile1) {
correlation2d.corrLMA2DualMax( // null pointer
imgdtt_params, // ImageDttParameters imgdtt_params,
imgdtt_params.lmas_LY_single, // false, // boolean adjust_ly, // adjust Lazy Eye
corr_wnd, // double [][] corr_wnd, // correlation window to save on re-calculation of the window
corr_wnd_inv_limited, // corr_wnd_limited, // correlation window, limited not to be smaller than threshold - used for finding max/convex areas (or null)
corrs, // corrs, // double [][] corrs,
disp_dist,
rXY, // double [][] rXY, // non-distorted X,Y offset per nominal pixel of disparity
// all that are not null in corr_tiles
correlation2d.selectAll(), // longToArray(imgdtt_params.dbg_pair_mask), // int pair_mask, // which pairs to process
maxes, //double[][] disp_str_dual, // -preliminary center x in pixels for largest baseline
null, // poly_disp, // double[] poly_ds, // null or pair of disparity/strength
imgdtt_params.ortho_vasw_pwr, // double vasw_pwr, // value as weight to this power,
null, // debug_lma_tile, // double [] debug_lma_tile,
(debugTile0 ? 1: -2), // int debug_level,
// -2, //0, // tile_lma_debug_level, // +2, // int debug_level,
tileX, // int tileX, // just for debug output
tileY );
}
if (disparity_map != null) {
int [] ixy = correlation2d.getMaxXYInt( // find integer pair or null if below threshold // USED in lwir
corr_combo_tile, // double [] data, // [data_size * data_size]
......@@ -2642,7 +2666,10 @@ public class ImageDtt extends ImageDttCPU {
ixy[0], // int ixcenter, // integer center x
false); // debugCluster); // (tile_lma_debug_level > 0)); // boolean debug);
if (corr_stat != null) { // almost always
disp_str = new double [] {-corr_stat[0], corr_stat[1]};
// FIXME: apply that for non-GPU version (and other variants) !
// convert to disparity for a quad camera (used in pre-shift)
disp_str = new double [] {-corr_stat[0]/Math.sqrt(2), corr_stat[1]};
// disp_str = new double [] {-corr_stat[0], corr_stat[1]};
if (disparity_map!=null) {
disparity_map[DISPARITY_INDEX_CM ][nTile] = disp_str[0];
disparity_map[DISPARITY_INDEX_CM + 1 ][nTile] = disp_str[1];
......@@ -2874,7 +2901,7 @@ public class ImageDtt extends ImageDttCPU {
ortho_weights[i] = 0.5*(1.0+Math.cos(Math.PI*dx))/imgdtt_params.ortho_eff_height;
}
}
if (globalDebugLevel > 0){
if (globalDebugLevel > 1){
System.out.println("ortho_height="+ imgdtt_params.ortho_height+" ortho_eff_height="+ imgdtt_params.ortho_eff_height);
for (int i = 0; i < corr_size; i++){
System.out.println(" ortho_weights["+i+"]="+ ortho_weights[i]);
......@@ -2882,7 +2909,7 @@ public class ImageDtt extends ImageDttCPU {
}
if (globalDebugLevel > 0) {
if (globalDebugLevel > 1) {
System.out.println("clt_aberrations_quad_corr(): width="+width+" height="+height+" transform_size="+transform_size+
" debug_tileX="+debug_tileX+" debug_tileY="+debug_tileY+" globalDebugLevel="+globalDebugLevel);
}
......
src/main/java/com/elphel/imagej/tileprocessor/ImageDttCPU.java
View file @ b9cc7d77
......@@ -1826,7 +1826,7 @@ public class ImageDttCPU {
ortho_weights[i] = 0.5*(1.0+Math.cos(Math.PI*dx))/imgdtt_params.ortho_eff_height;
}
}
if (globalDebugLevel > 0){
if (globalDebugLevel > 1){
System.out.println("ortho_height="+ imgdtt_params.ortho_height+" ortho_eff_height="+ imgdtt_params.ortho_eff_height);
for (int i = 0; i < corr_size; i++){
System.out.println(" ortho_weights["+i+"]="+ ortho_weights[i]);
......@@ -1834,7 +1834,7 @@ public class ImageDttCPU {
}
if (globalDebugLevel > 0) {
if (globalDebugLevel > 1) {
System.out.println("clt_aberrations_quad_corr(): width="+width+" height="+height+" transform_size="+transform_size+
" debug_tileX="+debug_tileX+" debug_tileY="+debug_tileY+" globalDebugLevel="+globalDebugLevel);
}
......@@ -1842,7 +1842,7 @@ public class ImageDttCPU {
final double [] filter = doubleGetCltLpfFd(corr_sigma);
// prepare disparity maps and weights
if (globalDebugLevel > 0){
if (globalDebugLevel > 1){
System.out.println("corr_fat_zero= "+corr_fat_zero);
System.out.println("disparity_array[0][0]= "+disparity_array[0][0]);
......@@ -3629,14 +3629,14 @@ public class ImageDttCPU {
ortho_weights[i] = 0.5*(1.0+Math.cos(Math.PI*dx))/imgdtt_params.ortho_eff_height;
}
}
if (globalDebugLevel > 0){
if (globalDebugLevel > 1){
System.out.println("ortho_height="+ imgdtt_params.ortho_height+" ortho_eff_height="+ imgdtt_params.ortho_eff_height);
for (int i = 0; i < corr_size; i++){
System.out.println(" ortho_weights["+i+"]="+ ortho_weights[i]);
}
}
if (globalDebugLevel > 0) {
if (globalDebugLevel > 1) {
System.out.println("clt_aberrations_quad_corr(): width="+width+" height="+height+" transform_size="+transform_size+
" debug_tileX="+debug_tileX+" debug_tileY="+debug_tileY+" globalDebugLevel="+globalDebugLevel);
}
......@@ -4605,14 +4605,14 @@ public class ImageDttCPU {
ortho_weights[i] = 0.5*(1.0+Math.cos(Math.PI*dx))/imgdtt_params.ortho_eff_height;
}
}
if (globalDebugLevel > 0){
if (globalDebugLevel > 1){
System.out.println("ortho_height="+ imgdtt_params.ortho_height+" ortho_eff_height="+ imgdtt_params.ortho_eff_height);
for (int i = 0; i < corr_size; i++){
System.out.println(" ortho_weights["+i+"]="+ ortho_weights[i]);
}
}
if (globalDebugLevel > 0) {
if (globalDebugLevel > 1) {
System.out.println("clt_aberrations_quad_corr(): width="+width+" height="+height+" transform_size="+transform_size+
" debug_tileX="+debug_tileX+" debug_tileY="+debug_tileY+" globalDebugLevel="+globalDebugLevel);
}
......@@ -7879,7 +7879,7 @@ public class ImageDttCPU {
ortho_weights[i] = 0.5*(1.0+Math.cos(Math.PI*dx))/imgdtt_params.ortho_eff_height;
}
}
if (globalDebugLevel > 0){
if (globalDebugLevel > 1){
System.out.println("ortho_height="+ imgdtt_params.ortho_height+" ortho_eff_height="+ imgdtt_params.ortho_eff_height);
for (int i = 0; i < corr_size; i++){
System.out.println(" ortho_weights["+i+"]="+ ortho_weights[i]);
......@@ -12225,7 +12225,7 @@ public class ImageDttCPU {
ortho_weights[i] = 0.5*(1.0+Math.cos(Math.PI*dx))/clt_parameters.img_dtt.ortho_eff_height;
}
}
if (globalDebugLevel > 0){
if (globalDebugLevel > 1){
System.out.println("ortho_height="+ clt_parameters.img_dtt.ortho_height+" ortho_eff_height="+ clt_parameters.img_dtt.ortho_eff_height);
for (int i = 0; i < corr_size; i++){
System.out.println(" ortho_weights["+i+"]="+ ortho_weights[i]);
......@@ -13162,7 +13162,7 @@ public class ImageDttCPU {
ortho_weights[i] = 0.5*(1.0+Math.cos(Math.PI*dx))/clt_parameters.img_dtt.ortho_eff_height;
}
}
if (globalDebugLevel > 0){
if (globalDebugLevel > 1){
System.out.println("ortho_height="+ clt_parameters.img_dtt.ortho_height+" ortho_eff_height="+ clt_parameters.img_dtt.ortho_eff_height);
for (int i = 0; i < corr_size; i++){
System.out.println(" ortho_weights["+i+"]="+ ortho_weights[i]);
......@@ -13879,14 +13879,14 @@ public class ImageDttCPU {
ortho_weights[i] = 0.5*(1.0+Math.cos(Math.PI*dx))/clt_parameters.img_dtt.ortho_eff_height;
}
}
if (globalDebugLevel > 0){
if (globalDebugLevel > 1){
System.out.println("ortho_height="+ clt_parameters.img_dtt.ortho_height+" ortho_eff_height="+ clt_parameters.img_dtt.ortho_eff_height);
for (int i = 0; i < corr_size; i++){
System.out.println(" ortho_weights["+i+"]="+ ortho_weights[i]);
}
}
if (globalDebugLevel > 0) {
if (globalDebugLevel > 1) {
System.out.println("clt_aberrations_quad_corr(): width="+width+" height="+height+" transform_size="+transform_size+
" debug_tileX="+debug_tileX+" debug_tileY="+debug_tileY+" globalDebugLevel="+globalDebugLevel);
}
......@@ -14291,7 +14291,7 @@ public class ImageDttCPU {
ortho_weights[i] = 0.5*(1.0+Math.cos(Math.PI*dx))/imgdtt_params.ortho_eff_height;
}
}
if (globalDebugLevel > 0){
if (globalDebugLevel > 1){
System.out.println("ortho_height="+ imgdtt_params.ortho_height+" ortho_eff_height="+ imgdtt_params.ortho_eff_height);
for (int i = 0; i < corr_size; i++){
System.out.println(" ortho_weights["+i+"]="+ ortho_weights[i]);
......@@ -14301,7 +14301,7 @@ public class ImageDttCPU {
if (globalDebugLevel > 0) {
if (globalDebugLevel > 1) {
System.out.println("clt_aberrations_quad_corr(): width="+width+" height="+height+" transform_size="+transform_size+
" debug_tileX="+debug_tileX+" debug_tileY="+debug_tileY+" globalDebugLevel="+globalDebugLevel);
}
......@@ -16027,7 +16027,7 @@ public class ImageDttCPU {
corrs[npair] , //double [] corr_tile,
npair); // int num_pair)
if (pair_width < 0.1) {
if (globalDebugLevel > 0) {
if (globalDebugLevel > 1) {
System.out.println("pair_width["+npair+"]="+pair_width);
}
} else {
......
src/main/java/com/elphel/imagej/tileprocessor/ImageDttParameters.java
View file @ b9cc7d77
......@@ -80,7 +80,7 @@ public class ImageDttParameters {
//lmamask_
public boolean lmamask_dbg = false; // show LMA images, exit after single BG
public boolean lmamask_en = false; // Use disparity-based LMA samples filter
public double lmamask_magic = 0.85;
public double lmamask_magic = 0.85; // Now only used for initial disparity from estimated by CM/Poly
public double lmamask_min_main = 0.4;
public double lmamask_min_neib = 0.10;
public double lmamask_weight_neib = 0.75;
......@@ -456,7 +456,7 @@ public class ImageDttParameters {
gd.addCheckbox ("Use disparity-based LMA samples filtering", this.lmamask_en,
"Generate weighs by averaging 2D correlation shape and per-pair shifting for estimated from CM disparity");
gd.addNumericField("Divide estimated disparity by magic 0.85", this.lmamask_magic, 6,8,"",
"Increase estimated disparity before averaging correlation shape and per-pair shifting the result");
"Increase estimated disparity for iniital LMA setiings");
gd.addNumericField("Minimal relative sample value for unconditional inclusion", this.lmamask_min_main, 6,8,"",
"Relatrive (to maximal) value in averaged correlation to be assigned window vlaue of 1.0 regardless of neighbors");
gd.addNumericField("Minimal relative sample value for neighbor inclusion", this.lmamask_min_neib, 6,8,"",
......
src/main/java/com/elphel/imagej/tileprocessor/QuadCLTCPU.java
View file @ b9cc7d77
......@@ -12858,7 +12858,7 @@ public class QuadCLTCPU {
final boolean save_diff = save_textures || need_diffs; // true; // separately save differences and
final boolean save_lowres = save_textures || need_diffs; // true; // low-res images
if (clust_radius > 00) { // will not generate textures
if (clust_radius > 0) { // will not generate textures
// set tasks for all non-NaN target disparities
TpTask [] tp_tasks_target = GpuQuad.setTasks(
num_sensors, // final int num_cams,
......@@ -12976,7 +12976,7 @@ public class QuadCLTCPU {
tp_tasks, // final TpTask [] tp_tasks, // data from the reference frame - will be applied to LMW for the integrated correlations
geometryCorrection.getRXY(false), // final double [][] rXY, // from geometryCorrection
// next both can be nulls
null, // final double [][][][] clt_corr_out, // sparse (by the first index) [type][tilesY][tilesX][(2*transform_size-1)*(2*transform_size-1)] or null
clt_corr_out, // null, // final double [][][][] clt_corr_out, // sparse (by the first index) [type][tilesY][tilesX][(2*transform_size-1)*(2*transform_size-1)] or null
// combo will be added as extra pair if mcorr_comb_width > 0 and clt_corr_out has a slot for it
// to be converted to float
dcorr_tiles, // final double [][][] dcorr_tiles, // [tile][pair][(2*transform_size-1)*(2*transform_size-1)] // if null - will not calculate
......
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