Abstract. Image matting often requires advanced image processing, especially in conditions, when small details such as hair are present in the image. In this article the hybrid method for human image matting based on convolutional neural network and principal curvatures is proposed. The U-Net based neural network is used to predict a rough foreground segmentation mask. Then the obtained foreground mask is refined by principal curvatures method to process the elongated hair-like structures. Test results show that the proposed method can improve the coarse human segmentation.
<p><strong>Abstract.</strong> Microscopic imaging is central to the brain and cognition studies in animals and often requires advanced image processing. In vivo recordings on awake behaving animals require stabilization of the images as the tissue in the images undergoes non-rigid deformations due to animal movement, pulse beat and breathing of the animal. Here we propose an approach to compensation for the tissue motion in calcium imaging data acquired with miniaturized wearable microscopes (miniscopes) from live rodent brains. Our approach includes preprocessing of the images in which we compensate for non-uniform illumination, remove calcium transients and instrument-specific noise. For image registration we use the multiscale mutual information based non-rigid algorithm with B-spline transformation model. We present the preliminary results of such motion compensation approach applied to the real miniscope image stacks.</p>
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.