Deblurring in AFM images

Atomic Force Microscopy (AFM) has been widely used in the material and life sciences and has enabled much progress in nanotechnology since it was invented. AFM relies on inter-atomic forces to form a topographical map of a sample surface. Depending on the surface type, it is possible to achieve atomic resolution. However, due to either the microscope design and operation mode or external environmental factors, AFM images usually contain some artefacts. Therefore, an important component of AFM imaging technology is the accurate identification of artefacts and the analysis of their causes in order to minimise their effects. This report describes an attempt to identify an approach to removing large distortions occurring in AFM images. These distortions are caused by damage to the scanning tip. A deblurring algorithm is proposed based on the existing blind deconvolution algorithm that has been well researched in natural image domain. By applying the deblurring algorithm, good approximations of both the ground truth latent image and the blur kernel can be found which provide more accurate information about the real surface topology of the material as well as the shape of the damaged scanning tip. The results of experiments on both simulated images and real AFM artefact images have shown that the proposed algorithm is successful at removing blur artefacts in the AFM image. Limitations of this approach are also discussed. Additional algorithms have also been proposed to improve the performance as well as accuracy of our deblurring algorithm. The PhD plan is also included in this report.