Grain size analysis from material images by using image processing techniques

GRAIN SIZE ANALYSIS FROM MATERIAL IMAGES BY USING IMAGEPROCESSING TECHNIQUES(M.Sc. Thesis)Hikmet BALGAZI UNIVERSITYINSTITUTE OF SCIENCE AND TECHNOLOGYSeptember 2006ABSTRACTThe aim of this thesis is to determine the material properties from thematerial images by using image processing techniques.With the software which is developed at the first stage of this study,each domino's value was determined from domino's images which areoriented at various directions and selected as a model in the computerenvironment. Domino's images have been transferred to computerenvironment by using digital camera within video capture software. It isused various video resolutions to model recognition. As for the secondstage of application, previous developed software has been enhancedand adapted to material images. From material images, spheric graphitearea percentage or desired phase pixel area percentage is calculatedwith the developed software. Image enhancing, image filtering,morphological construction and segmentation image processingtechniques have been used for this processes. Developed programproduces desired results related to images from acquired data byapplying those techniquesResults: i- In segmenting, the most effective image property is regularillumination for the image, ii- image resolution has significanceaccording to the application, iii- structural properties of objects areselective in image recognition processing, iv-it is shown that samplepreparation processing which will be applied to the material directlyaffect the accuracy of properties desired to determine.Science Code : 914.1.091Keywords : Vision system, image processing, edge detection,pattern recognition, grain sizePage : 230Supervisor : Prof. Dr. Mehmet EROĞLU