Abstract: Randomly distributed objects of piece goods in a state of translatory movement are identified and classified on the basis of a physical property thereof. The travel path of the piece goods is observed with a detector scanning the path transversally and indicating the wanted physical property, said detector generating an electric signal proportional to the observed physical quantity at each spot of the scanning line, within the resolution capacity of the detector. Thereupon, the electric signals are treated continuously in real time and converted into binary key number groups belonging individually to each particular object in the scanning line. The key number groups are processed during the scanning recursively in classifying units nominated for each particular object during its passage past the detector, and the classification is carried out on the basis of these key number groups.