Image processing techniques applied to the analysis of chiral features in nanoscale images

Our aim for this project was to design an algorithm comprised of image processing techniques to measure the chirality of a carbon nanotube. We first used a theoretical model of an armchair nanotube to gain insight. Theoretical models are much easier to work with than real images. It is expected that real images can be transformed into those that are nearly as clean as the theoretical models. Our aim was to show that the measurement could be made, not that the image could be cleaned up. It is also important to note that the chirality of the armchair nanotube is known. It may seem odd that we would try to measure something that has already been measured. This was done in order to qualify our method. After performing our transforms, we were able to compare the results to what was expected, as it was known. After designing an algorithm using well-known image processing tech-niques, including thresholding, pattern recognition, keying, kernel convolu-tion, and the two-dimensional Fourier transform, the results were analyzed for lessons that would be helpful in processing images of real nanotubes. Scan-ning tunneling microscopy images of carbon nanotubes obtained by Odom et al were used to build the algorithm. Traditional techniques failed in the real images and new techniques were required. Neighborhood thresholding was used to find the carbon atoms. Once the atoms were found, a well-behaved region was selected and an angle-sweep was performed to find the zig-zag axis. The method successfully approximated the angle of the zig-zag axis relative to the nanotube axis, meeting the goals of this project