제 목 Compressive Holography

강 사 Dr. Sehoon Lim

소 속 Duke University

시 간 2012년 6월 18일 (월요일) 오후 4시

장 소 제1공학관 301동 920A 호

Abstract

High-resolution optical imaging performs aperture synthesis by increasing the number of measurements, which overcomes the aperture limit. Typically, a single shot measurement of optical system cannot detect 3D image information and wide scattered fields due to the aperturelimit. Computational and optical co-design may accomplish high-resolution imaging in time-space dimensions with fewer measurements.

Compressive holography estimates images from incomplete data by using sparsity priors. Compressive holography combines digital holography and compressive sensing. Digital holography detects/reconstructs phase information from hologram data captured by an electronic focal plane array. Thus, a 2D hologram measurement supports decompressive inference for depth resolution in tomographic imaging. Compressive sensing enables accurate data reconstruction by prior knowledge on the decompressive inference. My PhD research explores estimation of 3D tomographic images from 2D data and estimation of images from under sampled apertures.

Compressive holography achieves single shot holographic tomography using the decompressive inference. Single shot holographic tomography suffers the uniqueness problem in the axial direction because the inversion is ill-posed. Compressive sensing alleviates the ill-posed problem by enforcing some sparsity constraints. Holographic tomography is applied for video-rate microscopic imaging and diffuse object imaging. In diffuse object imaging, sparsity priors are not valid in coherent image basis due to speckle. So incoherent image estimation is designed to hold the sparsity in incoherent image basis by support of multiple speckle realizations.

Compressive holography also achieves high pixel count imaging which means high resolution and wide field-of-view imaging. Coherent aperture synthesis can be one method to increase the aperture size of a detector. Scanning-based synthetic aperture confronts a multivariable global optimization problem due to time-space measurement errors. A hierarchical estimation strategy divides the global problem into multiple local problems with support of computational and optical co-design. Compressive sparse aperture holography can be another method. Compressive sparse sampling collects most of significant field information with a small fill factor because object scattered fields are locally redundant. Incoherent image estimation is adopted for the expanded modulation transfer function and compressive reconstruction.

강사이력

Sehoon Lim was born in Seoul, South Korea on December 5, 1978. He attended Yonsei University at Seoul, majoring in the Electrical and Computer Engineering and was a Department Scholar for three semesters. He graduated in 2005 with a Bachelors degree in Electrical and Computer Engineering. His undergraduate thesis was titled by “RGB Photodiodes for CMOS Image Sensor Applications” under Dr. Ilgu Yun. In the spring of 2005, he took a part-time job in developing an electrical analysis tool for Quantum dots laser with HP4284a under Dr. Won Jun Choi at Korea Institute of Science and Technology. In the fall, Sehoon came over to the University of Arizona for a Masters degree. He was a research assistant in the Photonic Systems Lab under Dr. Raymond K. Kostuk. His masters thesis was “Field-of-View(FOV) Extender for a Thin Camera System”.

In 2007, Sehoon enrolled in the PhD program in electrical and computer engineering at Duke University. Upon admission, He was offered two fellowships such as Electrical Engineering 1st year fellowship and John T. Chambers fellowship. There he became a member of the Duke Imaging and Spectroscopy Program (DISP) led by Dr. David Brady. He was active in the Optical Society of America (OSA), serving as a journal reviewer. In 2009, he was the 3rd winner of Fitzpatrick Institute of Photonics 9th Annual Meeting Poster Session. Sehoon has received his PhD in May.