A versatile and automated image processing technique and data extraction procedure from videomicroscopic data is presented. The motivation is a detailed quantification of blood platelet adhesion from laminar flow onto a surface. The characteristics of the system under observation (type of cells, their speed of movement, and the quality of the optical image to analyze) provided the criteria for developing a new procedure enabling tracking for long image sequences. Specific features of the novel method include: automatic segmentation methodology which removes operator bias; platelet recognition across the series of images based on a probability density function (two-dimensional, Gaussian-like) tailored to the physics of platelet motion on the surface; options to automatically tune the procedure parameters to explore different applications; integrated analysis of the results (platelet trajectories) to obtain relevant information, such as deposition and removal rates, displacement distributions, pause times and rolling velocities. Synthetic images, providing known reference conditions, are used to test the method. The algorithm operation is illustrated by application to images obtained by fluorescence microscopy of the interaction between platelets and von Willebrand factor-coated surfaces in parallel-plate flow chambers. Potentials and limits are discussed, together with evaluation of errors resulting from an inaccurate tracking.