The microstructure images of materials provide information about material properties, such as mechanical properties. The accuracy of determining the nodularity of ductile iron is very important. Dominant visual analysis of microstructures often results in biased results due to the human factor and the visual limitations themselves. Thus, the visually obtained results depend heavily on the visual ability of the expert, and in most cases these results are not repeatable.
For the reasons mentioned above, thinking has emerged in industry and in research practice to accurately determine the nodularity of ductile iron using digital image processing and computer vision technologies. Thus, digital image processing and computer vision technologies have become crucial in the field of material production and quality control.
The purpose of this diploma work was to check different mathematical models and algorithms that form different software that allows determination of the nodularity of ductile iron. We have tested two different software, first one was AnalySIS 5.0, which is used at the Faculty of Natural Sciences and Engineering and the internal program of Livar d. d. developed by an employee of the quality control department. The models were tested under ideal conditions using hand-drawn microstructures and actual microstructural images while monitoring the results obtained. In the end, we created our own program for calculating the nodularity using Excel. In our research, we relied on professional literature and articles.