A feature-based approach for processing nanoscale images

Nanotechnology is a rapidly emerging field in which the material structures are of the size 100 nanometers or smaller. Thus, analyzing images at the nanoscale level is a challenging task. Users in this field are interested in image analysis and processing to draw conclusions such as the impact of various experimental conditions on the nature of the image and consequently their usefulness in several applications. This motivates our work that involves designing a system that will not only recognize similarities and differences among images, but do so efficiently and accurately. Features are representative of the manner in which images are compared by human experts by finding empirical data about particle sizes, material depth, inter-particle distances and so forth. In this work, we look into the use of features for comparison by implementing a feature-based algorithm on real image data sets from nanotechnology and thereafter using the results in processes such as clustering that are commonly applied by users to analyze images. We are able to effectively assess the feature-based approach in a real-world context as corroborated by our experimental evaluation.