AGAPEROS: Searching for microlensing in the LMC with the pixel method - I.Data treatment and pixel light curves production

Recent surveys monitoring millions of light curves of resolved stars in the LMC have discovered several microlensing events. Unresolved stars could ho wever significantly contribute to the microlensing rate towards the LMC. Mo nitoring pixels, as opposed to individual stars, should be able to detect s tellar variability as a variation of the pixel flux. We present a first app lication of this new type of analysis (Pixel Method) to the LMC Bar. Mie de scribe the complete procedure applied to the EROS 91-92 data (one tenth of the existing CCD data set) in order to monitor pixel fluxes. First, geometr ic and photometric alignments are applied to each image. Averaging the imag es of each night reduces significantly the noise level. Second, one light c urve for each of the 2.1 10(6) pixels is built and pixels are lumped into 3 .6 " x 3.6 " super pixels, one for each elementary pixel. An empirical corr ection is then applied to account for seeing variations. We find that the f inal super-pixel light curves fluctuate at a level of 1.8% of the flux in b lue and 1.3% in red. We show that this noise level corresponds to about twi ce the expected photon noise and confirms previous assumptions used for the estimation of the contribution of unresolved stars. We also demonstrate ou r ability to correct very efficiently for seeing variations affecting each pixel flux. The technical results emphasised here show the efficacy of the Pixel Method and allow us to study luminosity variations due to possible mi crolensing events and variable stars in two companion papers.