A new noise-immune algorithm for estimating the position and intensity of nanoscale image peaks is proposed. The method is based on sequential processing of images by rows and columns to detect the beginning, the maximum and the end of the peaks using the derivatives at three points of the sliding data window. The coordinates of the points -found in that way -form clusters of peaks, which are further analyzed by the cluster analysis method. Unlike traditional approaches to surface structure

more »

... ysis, the proposed method does not require noise pre-filtering. With a signal/noise ratio of up to ten, it provides a probability of detection close to unity.