Image processing of HRTEM images with non-periodic features

Filtering of HRTEM images is investigated in order to find an optimum between reduction of noise and preservation of relevant non-periodic image detail. Several Fourier filters enhancing the signal-to-noise ratio are systematically examined. Filtering by means of a correlation procedure is considered as a special case. The filters are tested quantitatively on a known image with periodic and non-periodic image detail, using an R -factor deviation criterion. Examples are given of the application of filters to experimental HRTEM images of GaAs/A1As interfaces and of anti-phase boundaries in a Cu 3 Pd alloy. Rules are derived indicating which filter type and what filter size may be used in particular applications.