Accelerating Atomic Force Microscopy Imaging based on an Optimized Path Planning Method

Based on compressed sensing, undersampling is a low cost and efficient way to speed up the process of atomic force microscopy (AFM) imaging. The under-sampled information obtained is important in producing high-quality reconstructed images. Different samples and dynamic measurement show different characteristic of topography, which makes it impossible to acquire acceptable AFM image with same under-sampled scanning pattern. This work aims to propose an unsampling path planning method for effective under-sampled image acquisition. The preprocess is realized by object detection and k-means method. The path planning method combines Self-Organizing Map, Ant Colony Optimization and B-Spline. Through parallel calculation and cluster analysis, large-scale traveling salesman problem (L-TSP) is divided into several small-scale traveling salesman problems. After undersampling, the reconstruction process is realized by Bayesian compressed sensing. Several path planning algorithms are performed for comparison. An experimental example of L-TSP in AFM is carried out. Experimental and application results demonstrate that the proposed method can optimize scanning path of tens of thousands under-sampled points within minute. The proposed method succeeds to save time and guarantee the quality of AFM imaging.