Novel image and non-image parameters for efficient characterisation of atmospheric Cerenkov images

Abstract

We make a comparative study of the classification potential of various image and non-image parameters which are measurable with the TACTIC array focal plane instrumentation. The image parameters include conventional Hillas parameters and multifractal dimensions and wavelet moments. Similarly the parameters derived from non-image Cerenkov data consist of pulse profile rise time and base width and the relative ultravoilet to visible light fluxes of the cerenkov events. It is shown by the artificial neural net approach that suitable combinations of these parameters can bring about an efficient segregation of various event types, even for modest sized data samples of progenitor gamma-rays and cosmic ray hadrons.