int main(int argc, const char **argv)
{
// We declare our program description
string programDescription = "This Program does attribution and segmentation.";
programDescription+="\nVersion: 3.0";
programDescription+="\nAuthor: Benjamin Mampaey, [email protected]";
programDescription+="\nCompiled on " __DATE__ " with options :";
programDescription+="\nNUMBERCHANNELS: " + toString(NUMBERCHANNELS);
#if defined DEBUG
programDescription+="\nDEBUG: ON";
#endif
#if defined EXTRA_SAFE
programDescription+="\nEXTRA_SAFE: ON";
#endif
#if defined VERBOSE
programDescription+="\nVERBOSE: ON";
#endif
programDescription+="\nEUVPixelType: " + string(typeid(EUVPixelType).name());
programDescription+="\nReal: " + string(typeid(Real).name());
// We define our program parameters
ArgParser args(programDescription);
args("segmentation") = Classifier::segmentationParameters();
args("classification") = Classifier::classificationParameters();
args["config"] = ArgParser::ConfigurationFile('C');
args["help"] = ArgParser::Help('h');
args["type"] = ArgParser::Parameter("SPoCA2", 'T', "The type of classifier to use for the attribution.\nPossible values: FCM, PFCM, PCM, PCM2, SPoCA, SPoCA2");
args["imageType"] = ArgParser::Parameter("Unknown", 'I', "The type of the images.\nPossible values: EIT, EUVI, AIA, SWAP");
args["imagePreprocessing"] = ArgParser::Parameter("ALC", 'P', "The steps of preprocessing to apply to the sun images.\nCan be any combination of the following:\n NAR=zz.z (Nullify pixels above zz.z*radius)\n ALC (Annulus Limb Correction)\n DivMedian (Division by the median)\n TakeSqrt (Take the square root)\n TakeLog (Take the log)\n TakeAbs (Take the absolute value)\n DivMode (Division by the mode)\n DivExpTime (Division by the Exposure Time)\n ThrMin=zz.z (Threshold intensities to minimum zz.z)\n ThrMax=zz.z (Threshold intensities to maximum zz.z)\n ThrMinPer=zz.z (Threshold intensities to minimum the zz.z percentile)\n ThrMaxPer=zz.z (Threshold intensities to maximum the zz.z percentile)\n Smooth=zz.z (Binomial smoothing of zz.z arcsec)");
args["registerImages"] = ArgParser::Parameter(false, 'r', "Set to register/align the images when running multi channel attribution.");
args["centersFile"] = ArgParser::Parameter("centers.txt", 'c', "The name of the file containing the centers. If it it not provided the centers will be initialized randomly.");
args["computeEta"] = ArgParser::Parameter(false, 'e', "If the enters file do not contain the values for Eta or if you want to force Eta to be recomputed (slow!).");
args["stats"] = ArgParser::Parameter(false, 's', "Set to compute stats about the generated maps.");
args["statsPreprocessing"] = ArgParser::Parameter("NAR=0.95", 'P', "The steps of preprocessing to apply to the sun images.\nCan be any combination of the following:\n NAR=zz.z (Nullify pixels above zz.z*radius)\n ALC (Annulus Limb Correction)\n DivMedian (Division by the median)\n TakeSqrt (Take the square root)\n TakeLog (Take the log)\n TakeAbs (Take the absolute value)\n DivMode (Division by the mode)\n DivExpTime (Division by the Exposure Time)\n ThrMin=zz.z (Threshold intensities to minimum zz.z)\n ThrMax=zz.z (Threshold intensities to maximum zz.z)\n ThrMinPer=zz.z (Threshold intensities to minimum the zz.z percentile)\n ThrMaxPer=zz.z (Threshold intensities to maximum the zz.z percentile)\n Smooth=zz.z (Binomial smoothing of zz.z arcsec)");
args["output"] = ArgParser::Parameter(".", 'O', "The name for the output file or of a directory.");
args["uncompressed"] = ArgParser::Parameter(false, 'u', "Set this flag if you want results maps to be uncompressed.");
args["fitsFile"] = ArgParser::RemainingPositionalParameters("Path to a fits file", NUMBERCHANNELS, NUMBERCHANNELS);
// We parse the arguments
try
{
args.parse(argc, argv);
}
catch ( const invalid_argument& error)
{
cerr<<"Error : "<<error.what()<<endl;
cerr<<args.help_message(argv[0])<<endl;
return EXIT_FAILURE;
}
// We setup the output directory
string outputDirectory;
string outputFile = args["output"];
if (isDir(outputFile))
{
outputDirectory = outputFile;
}
else
{
outputDirectory = getPath(outputFile);
if (! isDir(outputDirectory))
{
cerr<<"Error : "<<outputDirectory<<" is not a directory!"<<endl;
return EXIT_FAILURE;
}
}
// We read and preprocess the sun images
deque<string> imagesFilenames = args.RemainingPositionalArguments();
vector<EUVImage*> images;
for (unsigned p = 0; p < imagesFilenames.size(); ++p)
{
EUVImage* image = getImageFromFile(args["imageType"], imagesFilenames[p]);
image->preprocessing(args["imagePreprocessing"]);
#if defined DEBUG
image->getHeader().set("IPREPROC", args["imagePreprocessing"], "Image Preprocessing");
image->writeFits(outputDirectory + "/" + stripPath(stripSuffix(imagesFilenames[p])) + ".preprocessed.fits");
#endif
images.push_back(image);
}
// We verify the images are aligned and we register them
for(unsigned p = 1; p < images.size(); ++p)
{
string dissimilarity = checkSimilar(images[0], images[p]);
if(! dissimilarity.empty())
{
if(args["registerImages"])
{
#if defined VERBOSE
cout<<"Image "<<imagesFilenames[p]<<" will be registered to image "<<imagesFilenames[0]<<endl;
#endif
images[p]->align(images[0]);
#if defined DEBUG
images[p]->writeFits(outputDirectory + "/" + stripPath(stripSuffix(imagesFilenames[p])) + ".registered.fits");
#endif
}
else
{
#if defined EXTRA_SAFE
cerr<<"Error: image "<<imagesFilenames[p]<<" and "<<imagesFilenames[0]<<" are not similar: "<<dissimilarity<<endl;
return EXIT_FAILURE;
#else
cerr<<"Warning: image "<<imagesFilenames[p]<<" and "<<imagesFilenames[0]<<" are not similar: "<<dissimilarity<<endl;
#endif
}
}
}
// We setup the filename prefix
if (isDir(outputFile))
{
// We set the name of the output files prefix to the outputDirectory + the classification type + image channel and date_obs
filenamePrefix = makePath(outputDirectory, args["type"]);
for(unsigned p = 0; p < images.size(); ++p)
filenamePrefix += "." + images[p]->Channel() + "." + toString(images[p]->ObservationTime());
filenamePrefix += ".";
outputFile = filenamePrefix + "SegmentedMap.fits";
}
else
{
filenamePrefix = stripSuffix(outputFile);
}
// We initialise the Classifier
Classifier* F;
bool classifierIsPossibilistic = false;
if (args["type"] == "FCM")
{
F = new FCMClassifier(args("classification"));
}
else if (args["type"] == "PCM")
{
F = new PCMClassifier(args("classification"));
classifierIsPossibilistic = true;
}
else if (args["type"] == "PFCM")
{
F = new PFCMClassifier(args("classification"));
classifierIsPossibilistic = true;
}
else if (args["type"] == "PCM2")
{
F = new PCM2Classifier(args("classification"));
classifierIsPossibilistic = true;
}
else if (args["type"] == "SPoCA")
{
F = new SPoCAClassifier(args("classification"));
classifierIsPossibilistic = true;
}
else if (args["type"] == "SPoCA2")
{
F = new SPoCA2Classifier(args("classification"));
classifierIsPossibilistic = true;
}
else
{
cerr<<"Error : "<<args["type"]<<" is not an accepted classifier!"<<endl;
return EXIT_FAILURE;
}
// We read the channels and the initial class centers from the centers file
vector<RealFeature> B;
vector<Real> Eta;
vector<string> channels;
if(args["centersFile"].is_set() && isFile(args["centersFile"]) && !emptyFile(args["centersFile"]))
{
if(classifierIsPossibilistic)
{
if(args["computeEta"])
{
if(! readCentersFromFile(args["centersFile"], channels, B))
{
cerr<<"Error: could not read centers from file!"<<endl;
return EXIT_FAILURE;
}
else if(!reorderImages(images, channels))
{
cerr<<"Error : The images channels do not correspond to centers channels."<<endl;
return EXIT_FAILURE;
}
else
{
// We initialise the classifier with the centers read from the file
F->initB(channels, B);
// We add the images to the classifier
F->addImages(images);
// We init the Eta with FCM attribution for the U
dynamic_cast<PCMClassifier*>(F)->FCMinitEta();
}
}
else
{
if(! readCentersEtasFromFile(args["centersFile"], channels, B, Eta) || Eta.empty())
{
cerr<<"Error: could not read centers and/or eta from file!"<<endl;
return EXIT_FAILURE;
}
else if(!reorderImages(images, channels))
{
cerr<<"Error : The images channels do not correspond to centers channels."<<endl;
return EXIT_FAILURE;
}
else
{
// We initialise the classifier with the centers and eta read from the file
dynamic_cast<PCMClassifier*>(F)->initBEta(channels, B, Eta);
// We add the images to the classifier
F->addImages(images);
}
}
}
else
{
if(! readCentersFromFile(args["centersFile"], channels, B))
{
cerr<<"Error: could not read centers from file!"<<endl;
return EXIT_FAILURE;
}
else if(!reorderImages(images, channels))
{
cerr<<"Error : The images channels do not correspond to centers channels."<<endl;
return EXIT_FAILURE;
}
else
{
// We initialise the classifier with the centers read from the file
F->initB(channels, B);
// We add the images to the classifier
F->addImages(images);
}
}
}
else
{
cerr<<"Error : For attribution you must provide the centers."<<endl;
return EXIT_FAILURE;
}
// We do the attribution
F->attribution();
// We declare the segmented map with the WCS of the first image, and get the segmentation map
ColorMap* segmentedMap = new ColorMap(images[0]->getWCS());
F->getSegmentedMap(args("segmentation"), segmentedMap);
//We add information about the classification to the header of the segmented map
Header& header = segmentedMap->getHeader();
for (unsigned p = 0; p < imagesFilenames.size(); ++p)
{
header.set(string("IMAGE")+toString(p+1,3), stripPath(imagesFilenames[p]));
}
header.set("CPREPROC", args["imagePreprocessing"], "Classification Image preprocessing");
header.set("CLASTYPE", args["type"], "Classifier Type");
if(args["stats"])
header.set("SPREPROC", args["statsPreprocessing"], "Segmentation stats Image preprocessing");
// We write the segmentedMap to a fits file
FitsFile file(outputFile, FitsFile::overwrite);
segmentedMap->writeFits(file, args["uncompressed"] ? 0 : FitsFile::compress, "SegmentedMap");
if(args["stats"])
{
for (unsigned p = 0; p < imagesFilenames.size(); ++p)
{
EUVImage* image = getImageFromFile(args["imageType"], imagesFilenames[p]);
image->preprocessing(args["statsPreprocessing"]);
if(args["registerImages"])
{
image->align(segmentedMap);
}
// We get the RegionStats
vector<SegmentationStats*> segmentation_stats = getSegmentationStats(segmentedMap, image);
// We write the RegionStats into the fits
file.writeTable(image->Channel()+"_SegmentationStats");
writeRegions(file, segmentation_stats);
for (unsigned r = 0; r < segmentation_stats.size(); ++r)
{
delete segmentation_stats[r];
}
}
}
// We cleanup
delete segmentedMap;
delete F;
for (unsigned p = 0; p < images.size(); ++p)
{
delete images[p];
}
images.clear();
return EXIT_SUCCESS;
}
