Amit Kohli (UC Berkeley): Computational aberration correction beyond deconvolution

Speaker: Dr. Amit Kohli (UC Berkeley)

Title: Computational aberration correction beyond deconvolution

Video: Click here to play

Abstract: There is no such thing as a perfect lens. Ever-present optical aberrations necessitate intricate engineering of multiple optical elements to achieve the high-resolution, high-quality imaging required for science. This process is costly and difficult, particularly for systems at the extremes of scale, such as miniature endoscopes and large-aperture space telescopes. An alternative is to keep optical designs simple and use computational post-processing to remove aberrations after image capture—offloading challenging optical engineering onto increasingly available compute. This talk will examine the limitations of computational aberration correction and present simple strategies to make it more robust and practical. These approaches point toward a path for building imaging systems that are optically simple yet capable of high resolution, with promising applications from medical endoscopy to space-based astronomy.

Bio: Amit is a 6th year PhD student advised by Prof. Laura Waller in the EECS department at UC Berkeley. He received his B.S. in Electrical Engineering at Stanford, specializing in signal processing and machine learning. During his PhD, Amit has focused on the design of robust and reliable computational imaging systems for biomedical applications. His work spans optics, statistics, and machine learning with a strong push toward making modular and practical computational tools for scientists.