PURPOSE: To make it possible to accomplish the clinical inspections more numerically and quantitatively by automatically classifying the distribution of blood corpuscles in accordance with the shapes thereof and by automatically counting the unmber of blood platelets.

CONSTITUTION: The blood corpuscle image of a sample from a microscope 1 is divided into two halves, which are fed through optical filters 13 and 15 to a signal processor 2, where they are converted by image sensors 17 and 18 into image signals. One of these outpus Eλ1 contains much red blood corpuscle signal components whereas the other output E2 contains much blood platelet and white corpuscle signal components. These two outputs are compared, and the red blood corpuscle signals EB and the blood platelet signals EP are strengthened and extracted by their wavelength differences. The parameters of the blood corpuscle signals are calculated by a cell classifying device 4 so that the distribution of the red blood corpuscles within a measured area according to their shapes and the number of the platelets for a preset value of the red corpuscles are calculated. As a result, the results can be displayed in a CTR 26 and a printer 27.