2002A&A...384...81H

This paper describes the methodology currently being implemented in the EIS pipeline for analysing optical/infrared multi-colour data. The aim is to identify different classes of objects as well as possible undesirable features associated with the construction of colour catalogues. The classification method used is based on the χ²-fitting of template spectra to the observed SEDs, as measured through broad-band filters. Its main advantage is the simultaneous use of all colours, properly weighted by the photometric errors. In addition, it provides basic information on the properties of the classified objects (e.g. redshift, effective temperature). These characteristics make the ξ2-technique ideal for handling large multi-band datasets. The results are compared to the more traditional colour-colour selection and, whenever possible, to model predictions. In order to identify objects with odd colours, either associated with rare populations or to possible problems in the catalogue, outliers are searched for in the multi-dimensional colour space using a nearest-neighbour criterion. Outliers with large ξ2-values are individually inspected to further investigate their nature. The tools developed are used for a preliminary analysis of the multi-colour point source catalogue constructed from the optical/infrared imaging data obtained for the Chandra Deep Field South (CDF-S). These data are publicly available, representing the first installment of the ongoing EIS Deep Public Survey.