My first real-life excercise in pattern recognition was the analysis of some STM images in order to extract quantitative information for statistical analysis. In particular we were interested in distributions of width for our nanowires, so I had to write a program that could detect wire edges on the image. The whole procedure is not fully automated, but it much faster and more accurate than doing the complete analysis manually. The results of our study of width distributions of nanowires have been published in
D.Y. Petrovykh et al, Surface Science 407, 189 (1998).

Cu nanowires on stepped Mo. Cu appears bright (dark) on top (bottom) image. Detected wire edges superimposed as black lines.
Higher resolution version of the above image

Encouraged by the succes of the first attempts I proceeded to adapt the program for detecting the terrace edges on images of Si (111) surface. Some useful information can be obtained from the statistics in this case as well, and the results have been published in
Applied Physics Letters 72, 948 (1998).

Grayscale STM topography of atomic steps on Silicon surface. Downhill to the right. Detected step edges superimposed as black lines.
Higher resolution version of the above image

Edge Detection
STM Image Analysis
Main Nanowiz Page
Publication list