Video mosaicing for real-time field of view enhancement

Abstract

The problem of limited field of view when visualizing biomedical specimens with high-magnification microscopes can be solved by using image mosaicing or image stitching technique to merge multiple microscopic images acquired from the specimens to create a seamless stitched image. The image mosaicing technique is commonly used in automatically map construction as a path planning system for vision-based robot navigation. By using this technique as the application of digital pathology, it is helpful for pathologists to view different parts of the specimens at high resolution on computer display. In this paper, we present an automatic image mosaicing algorithm which is implemented by using motion estimation technique in image registration step i.e. estimating the two-dimensional translation between each pair of image sequence by using optical flow and phase correlation methods. The aligned images are then fused together using multi-resolution image blending. These algorithms were accomplished to demonstrate the synthetic image sequences in microscopic resolution.