Seamless image stitching algorithm using radiometric lens calibration for high resolution optical microscopy

Abstract

Image mosaicing method is an image-based rendering and visualization method which is a common procedure in the generation of panoramic (composite) images and applications, such as creating virtual reality, super resolution, object insertion and object removal, etc. Specifically, the image stitching approach becomes common in super resolution digital imaging microscopy, where it is required to achieve both high magnification and a large field of view. This work presents the algorithm for the stitching of different overlapped high resolution views of the whole scene obtained from high magnification optical microscopes into a large radiometric balanced image through maximization of a cross-correlation function. To seamlessly blend and reduce the possibility of intensity and color mismatching of multiple source images in mosaicing, a radiometric calibration of the microscope optical system is employed. The algorithm was tested with grayscale and color images, and showed convincing and robust results in image stitching and blending.