Electrical Engineering-Systems Dept.

סמינר מחלקתי

You are invited to attend a lecture by

Dr. Thomas Trigano

(Shamoon College of Engineering, Beer-Sheva, Israel)

on the subject:

Statistical signal processing for  spectrometry Non-sparse list

of methods for a sparse Phenomenon

Gamma spectrometry is a sub_eld of nuclear science, whose aim is to identify the activity and the content of an unknown radioactive source. However, several random perturbations known as pileup effect introduce a distortion in the measurement, yielding an underestimation of the activity and the apparation of false peaks in the energy spectra, thus making the identification difficult.

In this talk, I will briey de_ne the pileup e_ect in gamma spectrometry, and will recall a direct inversion method used to correct it in a special case. Its limitations impose us to choose an alternative correction method, based on recent developments on sparse reconstruction and compressive sensing. This new approach consists on a post-processed version of the LASSO algorithm, whose solution is then plugged in known estimate of the activity and of the energies. The new method is shown to be more general than the direct inversion, and compares well both for the task of activity estimation as well as the task of spectrum reconstruction.