Prof. Stanley Chan receives $500K from NSF Robust Intelligence Core Program
Stanley Chan, Elmore Associate Professor of Electrical and Computer Engineering at Purdue University’s Elmore Family School of Electrical and Computer Engineering, has been awarded a $500K single-PI grant from the National Science Foundation under the Robust Intelligence core program. The proposal, titled “Learning to See through Atmospheric Turbulence,” aims to develop new imaging models and learning-based algorithms to enable seeing objects in a long-range.
The project will articulate four fundamental challenges in reconstructing images distorted from atmospheric turbulence: (1) how to build a physics justified forward degradation model that is fast to operate and is differentiable in the backpropagation sense? (2) how to build a learning-based lucky imaging algorithm that leverages the physics of the turbulence? (3) how to generalize the algorithm from one seeing condition to another? (4) how to meaningfully evaluate the reconstruction algorithms with new metrics and datasets?
The project is built upon years of research from Chan and his students. Earlier this year, the team invented the phase-to-space transform that accelerates turbulence simulation by 1000x compared to the gold standard in the literature while preserving all the critical physics properties. Their work will be presented in the IEEE International Conference on Computer Vision 2021. Previously, they have received a single-PI research grant from the Air Force Research Lab to develop reconstruction algorithms, and another single-PI research grant from NSF’s ECCS core program to develop single-photon cameras for imaging through turbulence. In 2021 IEEE Conference on Computer Vision and Pattern Recognition, Professor Chan delivered an invited talk on imaging through turbulence where his talk is available on YouTube.
The National Science Foundation’s Robust Intelligence core program funds fundamental research in computer vision, machine learning, and artificial intelligence. The award record for this grant is https://www.nsf.gov/awardsearch/showAward?AWD_ID=2133032