Thesis (M.Sc.)--Chulalongkorn University, 2015

Gamma-ray signal can be used as a fingerprint for radioisotope identification. In the context of radioactive and nuclear materials securing at the border control point, the detection task can present a significant challenge due to the limited measurement time, the shielding conditions and the noise interference. This study proposes a novel method to identify the signal of one or several radioisotopes from a poorly resolved spectrum. In this algorithm, the noise component in the raw spectrum is reduced by the wavelet decomposition approach, and the removal of continuum background is performed by the baseline determination algorithm. Finally, the identification of radioisotope is done using the matrix linear regression method. The proposed method has been verified by experiments using the poorly resolved gamma-ray signals from various scenarios, including single source, masking of natural uranium and enriched uranium (fresh fuel) with several radioactive sources. The novel method have better performance when compare with the commercial algorithm.