Test of image analysis parameters and particle morphology

The project is associated with a larger project, financed by Formas. The aim is to study the physical composition of the till and see if and how there is a couple to bedrock properties. The project studies till from all different aspects. Two test areas associated with rock quarries have been studied in detail, one at Olunda near Uppsala, and one at Bäckseda near Vetlanda. Several geophysical tests have been made over a test area of till at each site. This includes georadar, resistivity, magnetic susceptibility, VLF and seismic. The total size distribution and lithological composition of the clasts has been studied at each site, from either sections or sample pits. The size and shape distribution and the composition of the clasts are determined. The size and shape distribution is determined using image analysis.
This master’s thesis project deals with image analysis parameters compared to particle morphology. One aim of this part of the project is to determine how curvature and surface texture data can be presented in a meaningful way. Another aim is to couple curvature and surface texture to verbal classes and silhouette charts of particle form.
Since about the 1950’s smoothness/angularity has been estimated using verbal explanations or silhouette charts. We have developed two new image analysis methods for evaluating curvature and surface texture of particles. With image analysis we can accurately reproduce the outline of particles.
The first method creates the outline of the particle and identifies parts along the curve with greatest curvature is identified and the radius of the circle defined by the curvature determined. The question to solve is how to present the data. One possibility is to present both the number of deviating curvatures and the mean radius of all the defined circles. But is this meaningful. Can you present the data for 2000 particles? Furthermore the particles are imaged twice; both the largest and smallest projected areas.
The second method is for evaluation of the surface texture. The outline of the particle is smoothed in two steps, resulting in two curves. These would be identical for an extreemly smooth particle. The difference between them would increase as the surface texture became rougher (bottom left curve). Once again the question is how to present the data in a meaningful way and in 3-D.
Further more the results from both these new parameters must be evaluated to see how they compare, can be coupled, to results made using the traditional silhouette charts.
The advisor of the project is Docent Joanne Fernlund, at Engineering Geology, KTH [email protected] 08-7907045. You will also work in close cooperation with other master students working on associated projects and well as doctorate student Mimmi Arvidsson responsible for the geophysical studies.