We describe a new technique for the automatic detection and characterization, in terms of location, size and intensity, of positive hybridization spots on hybridization filter images. We will also discuss how this information can be used to reconstruct the spot deposition pattern on the filter and thus identify the spots. The part of the algorithm dedicated to the detection of spots is derived from a technique used in astronomy for the automatic recognition of galaxies on photographic plates. This technique has been modified so that it becomes able to cope with the strongly varying background typical of hybridization filter images. The technique does not require any human intervention.