Surface analysis algorithms for scanning probe microscopy

A single data set in scanning probe microscopy is large, typically in the megabyte range. As interpretation is accomplished by displaying the data in image form for visualization, image processing methods are used to both convert to visual images and to modify the images in order to clarify features of interest. Although an impressive number of image-processing algorithms are available on most commercial probe microscopes, many potentially very interesting ones are not: In addition, the special character of scanning probe data sets calls for development of new algorithms specially suited to this kind of problem. The work here analyzes images produced using atomic force microscope data sets. Algorithms are shown and discussed using images of oxide surfaces. The following algorithms are applied: tilt correction, scattering noise removal, surface smoothing, surface compression, probability density function analysis, correlation, and power spectrum analysis. Such algorithms and others serve to remove spurious surface spikes, enhance visualization of long-range surface features in the presence of short-range surface variations, remove line-to-line scanning artifacts, etc.