An algorithm for registration of erythrocytes on low-contrast images

An algorithm for counting the erythrocytes on low-contrast images of cytological preparations is proposed. The algorithm deals with low-contrast images: brightness histogram normalization, contrast stretching, and background alignment by the “top of hat” method. Erythrocytes are detected by the similarity criterion of reference samples and an area in the picture. The Pearson correlation coefficient is used as a similarity criterion. The algorithm has been successfully tested and proved to be efficient.