This paper reports a novel approach to the reconstruction of scanning probe microscopy (SPM) images by means of neural networks. The method aims to correct the integrating effect of a finite stylus tip. It is part of a general plan to enhance the performance of SPMs by means of neural networks. A well trained neural network is used in this approach to fulfil the nonlinear mapping from the apparent image to true surface. The results of experiments and simulations show that the reconstructed image tends to be closer the true surface than that measured images, and provides a better lateral resolution of measurements.
展开▼
