/**********************
   * Feature Extraction *
   **********************/

  int delta = floor(sizepattern/overlap);

  double features[29];
  double *noiseGauss;
  double tmpvar, tmpvar2;
  int i;

  /* Kernel for derivatives */
  double **dkernel=derivativeG(7.07106781186548e-01, 1.41421356237310e+00);

  for(index=0; index < (nA - delta); index += delta)
    {

      vecAna=vSegmentCopy(TT,index,sizepattern,nA);

      features[0] = promedio(vecAna,sizepattern);
      features[1] = elminimo(vecAna,sizepattern);
      features[2] = dstd(vecAna,features[0],sizepattern);

      features[3] = coefcorr(vecAna, thePattern[0] , sizepattern);
      for (i=1; i < numpattern; i++){
        tmpvar =  coefcorr(vecAna, thePattern[i] , sizepattern);
        if (tmpvar >  features[3]) {
          features[3] = tmpvar;
        }
      }
      /* and the correlation coef from gaussian noise */
      noiseGauss= gaussnoise(snrFactor ,  sizepattern);
      tmpvar = coefcorr(vecAna, noiseGauss, sizepattern);
      if (tmpvar >  features[3]) {
          features[3] = tmpvar;
       }

      if (debug){
        tmpvar=promedio(noiseGauss,sizepattern);
        tmpvar2= dstd(noiseGauss,tmpvar,sizepattern);
        printf("\nInfo: std[noiseGauss]=%16.14f SNR[%12.6f]", tmpvar2, 1/tmpvar2 );
        printf("\nInfo: std[vecAna]    =%16.14f SNR[%12.6f]",features[2], 1/features[2]);
      }

    }

  if (debug) {

    std::cout << "\n\nInfo: Read Elapsed Time: " << (double)  elapsed_seconds/1000  << " millisec \n" 
              << "Info: Num of Samples:( " << nA << ","
              << nB << ", " << nC << ") "
              <<  std::endl;  
  }

}