Automatic stitching of micrographs using local features

Abstract

Combination of single images to panoramic views is a popular application of image stitching in digital photography. By applying the same principle to micrographs, a number of common limitations of microscopes such as aberrations or limited depth of field may be overcome. This paper adapts recent methods of image registration for different application areas in light- and electron microscopy. Especially the suitability of SIFT and SURF features for micrograph correspondence analysis is in the focus of investigations. Test scenarios covering a wide range of magnifications and image contents are discussed. Additionally, the acquisition of the single scans and finally the complete generation of high-resolution panoramic micrographs may be automated. The proposed system is not only suitable as a tool for surface inspections, but also as a navigational aid for micro-and nanorobotic applications.