Restoration of images obtained from the scanning tunneling microscope

A new method for the restoration of images obtained from the scanning tunneling microscope (STM) is presented. We regard an image from the STM as a realization of a stochastic process and assume it to be composed of an underlying image and noise. The underlying image consists of smoothly varying gray levels representing the atomic profiles in the STM image. We express the underlying variation along an axis in terms of an "almost-periodic": function. We can determine the function by applying a recently developed procedure of data analysis, which consists of resolving the nonlinear least-squares method using the maximum entropy method of spectral analysis. This technique is applied to both modeled and actual data, The present method extracts the underlying image from the corrupted data and removes the horizontal striping effect due to 1/f(beta) low-frequency fluctuation.