Enhanced light-field image resolution via MLA translation

Abstract

This work describes a method that effectively improves the spatial resolution of light-field images without sacrificing angular resolution. The method involves translating the microlens array (MLA) linearly in both x- and y-directions in multiple steps to achieve 4 ×, 9 ×, 16 × and 25 × spatial resolution improvements. Its effectiveness was firstly validated through simulations with synthetic light-field images, demonstrating that distinct spatial resolution increments can be achieved by shifting the MLA. An MLA-translation light-field camera was built based on an industrial light-field camera, with which detailed experimental tests were carried out on a 1951 USAF resolution chart and a calibration plate. Qualitative and quantitative results prove that MLA translations can significantly improve measurement accuracy in x- and y- directions while preserving z-direction accuracy. Finally, the MLA-translation light-field camera was used to image a MEMS chip to demonstrate that finer structures of the chip can be acquired successfully.