Dictionary-based compressive Fourier ptychography

Abstract

Fourier ptychography (FP) provides an alternative scheme for improving the spatial bandwidth product with limited device. However, an angle-variation illumination is necessary to realize scanning imaging in the Fourier plane, which dramatically restricts the imaging speed and detection efficiency. In this paper, we propose a multiplexing and compressible FP scheme based on the structured illuminations and compressive sensing technique. Half of the LEDs are lighted together to reduce the exposure time; meanwhile, a learned dictionary is employed to reduce the sampling times. In addition, spectral independent illumination is proposed to retrieve color information from monochrome samplings. We experimentally verify that the proposed method can effectively reduce the sampling time with limited resolution loss.