67 std::cerr <<
" ||e||_2 at initial p: " << info[0] <<
'\n';
68 std::cerr <<
" ||e||_2: " << info[1] <<
'\n';
69 std::cerr <<
" ||J^T e||_inf: " << info[2] <<
'\n';
70 std::cerr <<
" ||Dp||_2: " << info[3] <<
'\n';
71 std::cerr <<
" mu/max[J^T J]_ii: " << info[4] <<
'\n';
72 std::cerr <<
" # iterations: " << info[5] <<
'\n';
73 switch ((
int) info[6]) {
75 std::cerr <<
" stopped by small gradient J^T e\n";
84 std::cerr <<
" singular matrix. Restart from current p with increased mu\n";
87 std::cerr <<
" no further error reduction is possible. Restart with increased mu\n";
90 std::cerr <<
" stopped by small ||e||_2\n";
93 std::cerr <<
" stopped by invalid (i.e. NaN or Inf) func values; a user error\n";
96 std::cerr <<
" # function evaluations: " << info[7] <<
'\n';
97 std::cerr <<
" # Jacobian evaluations: " << info[8] <<
'\n';
98 std::cerr <<
" # linear systems solved: " << info[9] <<
"\n\n";
104 unsigned int max_iterations,
double modified_chi_squared_scale,
108 : m_least_squares_engine(least_squares_engine),
109 m_max_iterations(max_iterations), m_modified_chi_squared_scale(modified_chi_squared_scale),
110 m_parameters(parameters), m_frames(frames), m_priors(priors) {}
114 return stamp_rect.
getWidth() > 0 && stamp_rect.getHeight() > 0;
122 auto frame_image = frame->getSubtractedImage();
126 for (
int y = 0;
y < rect.getHeight();
y++) {
127 for (
int x = 0;
x < rect.getWidth();
x++) {
128 image->at(
x,
y) = frame_image->getValue(rect.getTopLeft().m_x +
x, rect.getTopLeft().m_y +
y);
140 auto frame_image = frame->getSubtractedImage();
141 auto frame_image_thresholded = frame->getThresholdedImage();
142 auto variance_map = frame->getVarianceMap();
146 std::fill(weight->getData().begin(), weight->getData().end(), 1);
149 SeFloat gain = measurement_frame->getGain();
150 SeFloat saturation = measurement_frame->getSaturation();
152 for (
int y = 0;
y < rect.getHeight();
y++) {
153 for (
int x = 0;
x < rect.getWidth();
x++) {
154 auto back_var = variance_map->
getValue(rect.getTopLeft().m_x +
x, rect.getTopLeft().m_y +
y);
155 if (saturation > 0 && frame_image->getValue(rect.getTopLeft().m_x +
x, rect.getTopLeft().m_y +
y) > saturation) {
156 weight->at(
x,
y) = 0;
157 }
else if (weight->at(
x,
y) > 0) {
160 1.0 / (back_var + frame_image->getValue(rect.getTopLeft().m_x +
x, rect.getTopLeft().m_y +
y) / gain));
162 weight->at(
x,
y) =
sqrt(1.0 / back_var);
176 int frame_index = frame->getFrameNb();
178 auto frame_coordinates =
180 auto ref_coordinates =
191 auto group_psf =
ImagePsf(
pixel_scale * psf_property.getPixelSampling(), psf_property.getPsf());
197 for (
auto& source : group) {
198 for (
auto model : frame->getModels()) {
199 model->addForSource(manager, source, constant_models, point_models, extended_models, jacobian, ref_coordinates, frame_coordinates,
200 stamp_rect.getTopLeft());
217 int n_free_parameters = 0;
223 for (
auto& source : group) {
225 if (std::dynamic_pointer_cast<FlexibleModelFittingFreeParameter>(parameter)) {
229 parameter->create(parameter_manager, engine_parameter_manager, source));
241 int valid_frames = 0;
242 int n_good_pixels = 0;
244 int frame_index = frame->getFrameNb();
254 for (
int y = 0;
y < weight->getHeight(); ++
y) {
255 for (
int x = 0;
x < weight->getWidth(); ++
x) {
256 n_good_pixels += (weight->at(
x,
y) != 0.);
265 res_estimator.registerBlockProvider(
std::move(data_vs_model));
271 if (valid_frames == 0) {
274 else if (n_good_pixels < n_free_parameters) {
284 for (
auto& source : group) {
286 prior->setupPrior(parameter_manager, source, res_estimator);
295 auto solution = engine->solveProblem(engine_parameter_manager, res_estimator);
298 int total_data_points = 0;
301 int nb_of_free_parameters = 0;
302 for (
auto& source : group) {
305 bool accessed_by_modelfitting = parameter_manager.
isParamAccessed(source, parameter);
306 if (is_free_parameter && accessed_by_modelfitting) {
307 nb_of_free_parameters++;
311 avg_reduced_chi_squared /= (total_data_points - nb_of_free_parameters);
314 for (
auto& source : group) {
320 bool accessed_by_modelfitting = parameter_manager.
isParamAccessed(source, parameter);
321 auto modelfitting_parameter = parameter_manager.
getParameter(source, parameter);
323 if (is_dependent_parameter || accessed_by_modelfitting) {
324 parameter_values[parameter->getId()] = modelfitting_parameter->getValue();
325 parameter_sigmas[parameter->getId()] = parameter->getSigma(parameter_manager, source, solution.parameter_sigmas);
330 if (engine_parameter) {
345 logger.
error() <<
"An exception occured during model fitting: " <<
e.what();
352 for (
auto& source : group) {
355 auto modelfitting_parameter = parameter_manager.
getParameter(source, parameter);
357 if (manual_parameter) {
363 dummy_values, dummy_values);
372 int frame_index = frame->getFrameNb();
376 auto final_stamp = frame_model.getImage();
384 for (
int x = 0;
x < final_stamp->getWidth();
x++) {
385 for (
int y = 0;
y < final_stamp->getHeight();
y++) {
386 auto x_coord = stamp_rect.getTopLeft().m_x +
x;
387 auto y_coord = stamp_rect.getTopLeft().m_y +
y;
388 debug_image->setValue(x_coord, y_coord,
389 debug_image->getValue(x_coord, y_coord) + final_stamp->getValue(
x,
y));
400 double reduced_chi_squared = 0.0;
402 for (
int y=0;
y < image->getHeight();
y++) {
403 for (
int x=0;
x < image->getWidth();
x++) {
404 double tmp = image->getValue(
x,
y) - model->getValue(
x,
y);
405 reduced_chi_squared += tmp * tmp * weights->getValue(
x,
y) * weights->getValue(
x,
y);
406 if (weights->getValue(
x,
y) > 0) {
411 return reduced_chi_squared;
418 total_data_points = 0;
419 int valid_frames = 0;
421 int frame_index = frame->getFrameNb();
426 auto final_stamp = frame_model.getImage();
434 image, final_stamp, weight, data_points);
436 total_data_points += data_points;
437 total_chi_squared += chi_squared;
441 return total_chi_squared;
void printLevmarInfo(std::array< double, 10 > info)
EngineParameter are those derived from the minimization process.
std::shared_ptr< DependentParameter< std::shared_ptr< EngineParameter > > > x
void setEngineValue(const double engine_value)
std::shared_ptr< DependentParameter< std::shared_ptr< EngineParameter > > > y
T dynamic_pointer_cast(T... args)
Data vs model comparator which computes a modified residual, using asinh.
void setValue(const double new_value)
void error(const std::string &logMessage)
std::unique_ptr< DataVsModelResiduals< typename std::remove_reference< DataType >::type, typename std::remove_reference< ModelType >::type, typename std::remove_reference< WeightType >::type, typename std::remove_reference< Comparator >::type > > createDataVsModelResiduals(DataType &&data, ModelType &&model, WeightType &&weight, Comparator &&comparator)
Class responsible for managing the parameters the least square engine minimizes.
static std::shared_ptr< LeastSquareEngine > create(const std::string &name, unsigned max_iterations=1000)
Provides to the LeastSquareEngine the residual values.
static Logging getLogger(const std::string &name="")