UBC Computer Science PhD student Felix Heide is the recipient of the 2017 ACM SIGGRAPH Outstanding Doctoral Dissertation Award. Only one award is given annually to recognize researchers who have made notable contributions during their doctoral study.
Felix' core research interests lie in computational imaging and display, which promises to revolutionize both imaging and display technologies through the introduction of computation, thereby enabling more robust, less expensive, and more portable optical devices. Moreover, it allows for completely new imaging modalities that were not possible previously. Felix in particular has focused on computational imaging topics with interesting inverse problems that can be solved as optimization problems, relying on both prior knowledge of the physical image formation process as well as statistical image priors.
As such, Felix has made major contributions to the development of flexible computational photography pipelines. Based on his earliest work on a cross-channel color image prior for image reconstruction tasks (ACM TOG 2013), Felix developed a flexible image signal processor (FlexISP) framework for solving a wide range of low-level image reconstruction problems (including de-mosaicking, HDR reconstruction, and burst denoising) using a single, but modular optimization framework. The astonishing result for this work, which was published at Siggraph Asia 2014, was that this one framework could be applied to very exotic camera designs, while at the same time outperforming the best dedicated algorithms on standard designs. The culmination of this line of research is a domain-specific language specifically designed for the these types of optimization problems in imaging (Siggraph 2016). Much of this work is currently being commercialized by Montreal-based startup company Algolux which Felix co-founded.
A second major part of Felix' research activities has be the use of similar non-convex optimization methods to time-of-flight sensors. The first paper in this line of work (published at Siggraph 2013), showed that inexpensive time-of-flight cameras are useful for imaging the propagation of light in environments (i.e light-in-flight, or transient imaging). This work was followed up by investigating specific applications, including cameras that ``look around corners'', i.e. imaging and 3D reconstruction outside the direct line of sight (CVPR Oral 2014), imaging in scattering media (Optics Express 2014), Doppler velocity imaging (i.e. turning each pixel in a PMD camera into a Lidar gun, Siggraph 2015), and multi-camera ToF systems (Siggraph 2016).
Felix has been exceptionally productive during his PhD. He has been the first-author on one article in Nature Scientific Reports, 7 ACM TOG articles (5 Siggraph, 1 Siggraph Asia, 1 TOG), 2 CVPR papers (both orals), and 2 Optics Express articles. This is without counting the many papers where he was not the first author but still made significant contributions, or the publications in second-tier venues.
Felix Heide received his BSc (2010) and MSc (2012) degrees in Computer Science from the University of Siegen. He received his Ph.D. in December 2016 at the University of British Columbia under the advisement of Professor Wolfgang Heidrich. Since January 2017, he is a Postdoctoral Researcher at Stanford University.