||I will present the outcome of my master's thesis which I concluded at the Institute for Chemical Analysis and Technology at the Vienna Technical University.
The study was an initial attempt to investigate the feasibility of automating fiber analysis in forensic science using Raman spectroscopy. The main focus was the practicability in crime investigation applications, rather than in optimized laboratory settings.
To enable self-directed collection of spectral data a sequence of image processing steps is presented in the first part of this thesis. Its purpose is to allow for image segmentation and separation of fibers from the background to reduce the amount of measuring locations.
Raman measurements have been conducted for various fibers of different fabrics and different dyes in an experimental setting which mimics real crime scenes, not conventional sample preparation in spectroscopy.
The empirical part of this study was conducted in cooperation with the University of Natural Resources and Life Sciences, Vienna and the Criminal Investigation Service Austria.
The results support the implementation of the algorithm presented to recognize fibers and search for fibers of a specific color to reduce the time consuming manual examination of a large set of evidence tapes.
However the outcome of spectral examinations outlines several obstacles in using automated Raman spectroscopy, the majority of which are a direct result of thermal decomposition.