Jacob Lucas, The Boeing Company; Brandoch Calef, The Boeing Company; Trent Kyono, The Boeing Company and UCLA Computer Science
Keywords: bispectrum, Convolutional Neural Network, image processing, phase retrieval
Abstract:
Bispectrum processing is a well-established tool for phase retrieval in speckle imaging. Recent advancements in image processing with neural networks imply great effectiveness with denoising, inpainting, and image recovery, suggesting that the application of a customized neural network to the bispectrum could improve the quality of the reconstructed phase. Motivated by this, we explore the application of deep neural networks to assist with and enhance the performance of a standard bispectrum phase retrieval algorithm.
Date of Conference: September 11-14, 2018
Track: Poster