Photometrically robust Fourier light field microscopy measuring method using spatial-frequency hybrid stereo matching

Light field microscopy is an effective and accurate method for surface metrology, and Fourier light field microscopy improves spatial resolution of light field data through novel system configuration. Window-based stereo matching estimates depth from light field data through cost volumes constructed through pixel intensity and gradients. In practical stereo-matching-based reconstruction from light-field data, performance is impeded by limited pixels per sub-aperture images, degraded radiometric quality and the low dimensionality of traditional depth cues. To address this issue, this paper presents a photometrically robust spatial-frequency hybrid stereo matching method for surface measurement based on Fourier light field microscopy. Weighted average monogenic phase maps are calculated from original sub-aperture images through multi-scale monogenic filtering. Spatial-frequency hybrid cost volume is constructed by weighted combination of pixel intensity, gradients and weighted average local phase. Experimental results with a Fourier light field microscope demonstrate considerable enhancement of accuracy and robustness of the proposed method compared to traditional intensity-based stereo matching.