The paper focuses on the determination of elements of measuring process, whose impact on promptness and precision of a structure`s quantitative description is the greatest. It also concentrates on the introduction of such changes into the realization of the measuring process elements, that will make it possible to minimize the errors they cause.
Replacement of the classical, i.e. manual and semi-automatic measuring methods with automatic ones has led to significant changes in the measuring process. Measurement of geometric features of analyzed structure elements, which has constituted the greatest problem of the methods applied so far, can be described by universal algorithms, independent from these objects` morphologies. The most difficult element of the measuring process is exact detection of objects being measured. It has been shown that detection of all phases existing in a tested material requires a complementary application the different techniques of light as well as of scanning microscopy. In particularly difficult cases, satisfying results are obtained only with the application of the most modern scanning microscopy tools, such as EBSP or mapping. A disadvantage of these methods is that they are labour- and time-consuming. The solutions presented in the paper enabled considerable acceleration of the latter.
Obtaining a structure`s digital image which allows to detect measured objects using solely binarization, is possible, even when preparation procedures were performed correctly, only in sparse cases. Therefore, the initial grey image, before binarization, is to be subjected to additional transformations to remove preparation defects and shadow effect. The new procedures of grain boundaries reconstruction (a controlled segmentation and a conditional reconstruction) and methods of twins boundaries and lamellar precipitation colonies disclosure were also developed.
|