2005A&A...431..707G

The detection of transits is an efficient technique to uncover faint companions around stars. The full characterisation of the companions (M-type stars, brown dwarfs or exoplanets) requires high-resolution spectroscopy to measure properly masses and radii. With the advent of massive variability surveys over wide fields, the large number of possible candidates makes such a full characterisation for all of them impractical. We explore here a fast technique to pre-select the most promising candidates using either near-IR photometry or low resolution spectroscopy. We develop a new method based on the well-calibrated surface brightness relation along with the correlation between mass and luminosity for main sequence stars, so that not only can giant stars be excluded but also accurate effective temperatures and radii measured. The main source of uncertainty arises from the unknown dispersion of extinction at a given distance. We apply this technique to our observations of a sample of 34 stars extracted from the 62 low-depth transits identified by OGLE during their survey of some 10⁵ stars in the Carina fields of the Galactic disc. We infer that at least 78% of the companions of the stars which are well characterised in this sample are not exoplanets. Stars OGLE-TR-105, OGLE-TR-109 and OGLE-TR-111 are the likeliest to host exoplanets and deserve high-resolution follow-up studies. Most recently, OGLE-TR-111 was confirmed as an exoplanet with M_planet≃0.53 ±0.11M_Jup (Pont et al., 2004A&A...426L..15P), confirming the efficiency of our method in pre-selecting reliable planetary transit candidates.