/// <summary>
/// Configuration GA.
/// </summary>
/// <param name="problemConfig">Problem represention</param>
/// <returns>GA</returns>
public static IEVA[] SetGA(IProblemConfig[] problemConfig, bool parlelization)
{
Console.WriteLine("Custom Genetic algorithm.");
var termination = ParameterSetter.SetTermination(problemConfig[0]);
var crossover = ParameterSetter.SetXover(problemConfig[0]);
var xoverProb = ParameterSetter.SetFloat("Xover Probability", 0, 1);
var mutation = ParameterSetter.SetMutation(problemConfig[0]);
var mutationProb = ParameterSetter.SetFloat("Mutation Probability", 0, 1);
var selection = ParameterSetter.SetSelection(problemConfig[0]);
var elitizmus = ParameterSetter.SetElite(problemConfig[0]);
var executor = problemConfig[0].CreateExecutor();
var popSize = ParameterSetter.SetPositiveIntParameter("Pop size", 2);
IEVA[] evas = new IEVA[problemConfig.Length];
int i = 0;
foreach (var config in problemConfig)
{
// creates population with curr. popSize
var population = new Population(popSize, config.CreateIndividual, config.CreateEmptyIndividual);
// new operators
var fitness = config.CreateFitness();
evas[i++] = new GeneticAlgorithm(population, fitness, selection, crossover, mutation, elitizmus, termination, executor, mutationProb, xoverProb);
}
return(evas);
}
static void Main(string[] args)
{
bool running = true;
while (running)
{
Console.Clear();
Console.WriteLine("Evolution image approximation.");
var paralelization = ParameterSetter.SetBool("Parallelization");
// Logger rate
var logRate = ParameterSetter.SetPositiveIntParameter("Logger rate", 1);
// load input image
var inputFileName = ParameterSetter.SetFileNameParameter("image");
var problemConfigName = ParameterSetter.SetProblemCOofig();
if (paralelization)
{
RunParallelImage(problemConfigName, inputFileName, logRate, paralelization);
}
else
{
RunImage(problemConfigName, inputFileName, logRate, paralelization);
}
Console.WriteLine("EVOLVED");
running = ParameterSetter.SetBool("RESTART");
}
}
/// <summary>
/// Notmal EA
/// </summary>
/// <param name="problemConfigName">Problem represention.</param>
/// <param name="inputFileName">Input image</param>
/// <param name="logRate">logger rate</param>
public static void RunImage(Type problemConfigName, string inputFileName, int logRate, bool parallelization)
{
// Scale img
var paramterScale = ParameterSetter.SetFloat("Scale", 0.01f, 1);
var consoleProblemConfig = Activator.CreateInstance(problemConfigName) as IImageProblemConfig;
consoleProblemConfig.Initialize(0);
consoleProblemConfig.InitializeScale(paramterScale);
consoleProblemConfig.Initialize(inputFileName);
var eva = Configuration.SetEVA(new[] { consoleProblemConfig }, parallelization);
var configuration = new Configuration();
configuration.SetGenerationInfoLog(eva[0], consoleProblemConfig, logRate);
consoleProblemConfig.ConfigGATermination(eva[0]);
eva[0].Run();
}
/// <summary>
/// Configuration ES.
/// </summary>
/// <param name="problemConfig">Problem represention</param>
/// <returns>ES</returns>
public static IEVA[] SetES(IProblemConfig[] problemConfig, bool parallelization)
{
Console.WriteLine("Evolutionary Strategy.");
var evaName = ParameterSetter.SetEvaES();
// CMAES - internal population size setter
var popSize = 20;
IEVA[] evas = new IEVA[problemConfig.Length];
int i = 0;
foreach (var config in problemConfig)
{
// creates population with curr. popSize
var population = new Population(popSize, config.CreateIndividual, config.CreateEmptyIndividual);
// new operators
var fitness = config.CreateFitness();
// Delegate.CreateDelegate(consoleProblemConfig.CreateIndividual);
evas[i++] = Activator.CreateInstance(evaName, fitness, population) as IEVA;
}
return(evas);
}
/// <summary>
/// Parallel EA.
/// </summary>
/// <param name="problemConfigName">Problem represention.</param>
/// <param name="inputFileName">Input image</param>
/// <param name="logRate">logger rate</param>
public static void RunParallelImage(Type problemConfigName, string inputFileName, int logRate, bool parallelization)
{
var img = Image.FromFile(inputFileName) as Bitmap;
var blockSize = ParameterSetter.SetPositiveIntParameter("Block size", 5);
Bitmap[] targets = Split_Concate.SplitImg(img, blockSize);
var consoleProblemConfigs = new IImageProblemConfig[targets.Length];
var evas = new IEVA[targets.Length];
var configuration = new Configuration[targets.Length];
for (int i = 0; i < targets.Length; i++)
{
consoleProblemConfigs[i] = Activator.CreateInstance(problemConfigName) as IImageProblemConfig;
consoleProblemConfigs[i].Initialize(i);
consoleProblemConfigs[i].Initialize(targets[i], inputFileName);
configuration[i] = new Configuration();
}
evas = Configuration.SetEVA(consoleProblemConfigs, parallelization);
for (int i = 0; i < targets.Length; i++)
{
configuration[i].SetGenerationInfoLog(evas[i], consoleProblemConfigs[i], logRate);
consoleProblemConfigs[i].ConfigGATermination(evas[i]);
}
IIndividual[] bestInds = new IIndividual[targets.Length];
var stopwatch = Stopwatch.StartNew();
/**/
Parallel.For(0, evas.Length, index => {
evas[index].Run();
bestInds[index] = evas[index].BestIndividual;
// configuration[index].Saveinfo(inputFileName, index);
});
/*/
*
* for (var index = 0; index < evas.Length; index++)
* {
* evas[index].Run();
*
* bestInds[index] = evas[index].BestIndividual;
*
* // Configuration.Saveinfo(inputFileName,);
*
* }
*
*
* /**/
stopwatch.Stop();
Console.WriteLine("Total time: " + stopwatch.Elapsed);
var result = Split_Concate.ConcateImgs(bestInds, blockSize, img.Width, img.Height);
var m_destFolder = Configuration.CreateDirectory(inputFileName, "result_final");
var fileName = m_destFolder + "/" + "final" + ".png";
result.Save(fileName);
}