We investigate the possible link between mergers and the enhanced activity of supermassive black holes (SMBHs) at the centre of galaxies, by comparing the merger fraction of a local sample (0.003 =< z < 0.03) of active galaxies - 59 active galactic nuclei (AGN) host galaxies selected from the all-sky Swift BAT (Burst Alert Telescope) survey - with an appropriate control sample (247 sources extracted from the Hyperleda catalogue) that has the same redshift distribution as the BAT sample. We detect the interacting systems in the two samples on the basis of non-parametric structural indexes of concentration (C), asymmetry (A), clumpiness (S), Gini coefficient (G) and second order momentum of light (M20). In particular, we propose a new morphological criterion, based on a combination of all these indexes, that improves the identification of interacting systems. We also present a new software - PyCASSo (Python CAS Software) - for the automatic computation of the structural indexes. After correcting for the completeness and reliability of the method, we find that the fraction of interacting galaxies among the active population, 20(+7/-5) per cent, exceeds the merger fraction of the control sample, 4(+1.7/-1.2) per cent. Choosing a mass-matched control sample leads to equivalent results, although with slightly lower statistical significance. Our findings support the scenario in which mergers trigger the nuclear activity of supermassive black holes.