Abstract: People take pictures and videos to share with family and friends. This era of social media has seen an explosion of apps for capturing and editing multi-media content in entertaining ways, including digital masks and style transfer. Computational photography, which refers to digital techniques for improving images and videos, is key to enhancing the user visual experience.
In this talk, I will survey some of my work on computational photography. I will start with low-level image processing (noise estimation, vignetting removal, and chromatic aberration reduction), since this is fundamental to understanding image formation for effective image restoration. This is followed by descriptions of more recent projects involving data-driven techniques, namely, personalized image enhancement, semantic-based cinemagraphs and hyperlapses, and deep image analogy for style swapping. I will also touch on my involvement in the iOS app called Microsoft Pix, and conclude with thoughts on future directions.
Bio: Sing Bing Kang received his Ph.D. in robotics from Carnegie Mellon University, Pittsburgh, in 1994. He is Principal Researcher at Microsoft Corporation, and his research interests include image-based modeling and computational photography. He has coedited two books (Panoramic Vision and Emerging Topics in Computer Vision) and coauthored two books (Image-Based Rendering and Image-Based Modeling of Plants and Trees). On the community service front, he has served as Area Chair for the major computer vision conferences and as papers committee member for SIGGRAPH and SIGGRAPH Asia. He was Program Chair for ACCV 2007 and CVPR 2009, and was Associate Editor-In-Chief for IEEE Transactions on Pattern Analysis and Machine Intelligence from 2010-2014. He is a Fellow of the IEEE.