Machine Learning Methods Used in the Binarization of Parasitological Micrographs Natalia Verezubova

This article discusses the issue of effective use of popular methods of binary image filtering for segmentation of objects in photographs obtained during parasitological studies and veterinary and sanitary examination. A combination of filtering methods and binarization methods not used in classical practice of machine and computer vision is proposed. An optimal structure of a system based on machine learning capable of binary filtering of parasitological microphotographs is proposed. The characteristics of both histograms and the images themselves obtained as a result of parasitological examination are established. In conclusion, statistical processing of the data on histograms and the images themselves obtained during the study is carried out using correlation analysis. Correspondences between the obtained characteristics and the metrics found during the tests are identified in order to determine the influence of image characteristics and segmentation results.