Occlusion Removal In Video Microscopy

Video microscopy offers researchers a method to observe small-scale dynamic processes. It is often useful to remove unchanging portions of image sequences to improve the perception and analysis of moving features. We propose two processing methods (a local method and a global method) for detecting and removing partial stationary occlusions from video microscopy data using the bright-field microscope image model. In both techniques, we compute the relative light transmission across the image plane due to fixed, partially-transparent objects. The resulting transmission map enables reconstruction of a video in which the occlusions have been removed. We present experimental results that compare the effectiveness and applicability of our two approaches.