Development and application of image analysis techniques for identification and classification of microscopic particles.

摘要

This work discusses aspects of the development and application of image analysis techniques for identification and classification of microscopic particles. In the first part of the thesis the physics of image formation occurring in the instruments which were used in the study, is considered. It is also shown which basic processing steps should be performed in order to obtain useful parameters about the content of an image.;The second part of the thesis is dedicated to the mathematical problem of the description of the shape of an object. Two different directions in shape analysis are discussed. One direction is based on the idea of representing the object's shape as a function, the other considers an object as a set.;In the third part of the thesis different applications of image analysis techniques to solve real problems are discussed. In the first application fractal theory is applied to distinguish between aerosol particles which have different origin but very similar chemical composition. In another application the principles of functional approach to describe and characterize complex shapes are applied to a problem of differentiation between algae cells and their agglomerates. A large part of the thesis describes how the problem of classification of differently shaped silver halide micorcrystals can be solved. Not only single microcrystals are considered, a new method to segment overlapping microcrystals as they appear in SEM images is developed. An attempt to characterize many-particle-systems and percolation networks by fractal analysis is described in the last part of the thesis.