Besides new observations, mining old photographic plates and CCD image archives represents an opportunity to recover and secure newly discovered asteroids, also to improve the orbits of Near Earth Asteroids (NEAs), Potentially Hazardous Asteroids (PHAs) and Virtual Impactors (VIs). These are the main research aims of the EURONEAR network. As stated by the IAU, the vast collection of image archives stored worldwide is still insufficiently explored, and could be mined for known NEAs and other asteroids appearing occasionally in their fields. This data mining could be eased using a server to search and classify findings based on the asteroid class and the discovery date as ``precoveries'' or ``recoveries''. We built PRECOVERY, a public facility which uses the Virtual Observatory SkyBoT webservice of IMCCE to search for all known Solar System objects in a given observation. To datamine an entire archive, PRECOVERY requires the observing log in a standard format and outputs a database listing the sorted encounters of NEAs, PHAs, numbered and un-numbered asteroids classified as precoveries or recoveries based on the daily updated IAU MPC database. As a first application, we considered an archive including about 13 000 photographic plates exposed between 1930 and 2005 at the Astronomical Observatory in Bucharest, Romania. Firstly, we updated the database, homogenizing dates and pointings to a common format using the JD dating system and J2000 epoch. All the asteroids observed in planned mode were recovered, proving the accuracy of PRECOVERY. Despite the large field of the plates imaging mostly 2.27° × 2.27° fields, no NEA or PHA could be encountered occasionally in the archive due to the small aperture of the 0.38 m refractor insufficiently to detect objects fainter than {V ˜ 15}. PRECOVERY can be applied to other archives, being intended as a public facility offered to the community by the EURONEAR project. This is the first of a series of papers aimed to improve orbits of PHAs and NEAs using precovered data derived from archives of images to be data mined in collaboration with students and amateurs. In the next paper we will search the CFHT Legacy Survey, while data mining of other archives is planned for the near future.