The aim of this paper is to present methods for a fully computerised a
nalysis of spermatocyte movement. The technique were designed for proc
essing pure images of spermatozoa captured from a light microscope. Th
e techniques described allowed for the reduction of background light i
nhomogeneity and for the correct detection of moving cells and involve
d densitometric equalisation of background inhomogeneity and implement
ation of the dynamic threshold, adopted for the recognition of objects
. The method for relating: cells (found in each frame processed) to mo
vement trajectories used heuristic rules, describing the behaviour of
moving cells. This permitted samples containing high numbers of sperma
tocytes within the observed area to be processed with maximum accuracy
.