Piecewise Segmentation Occlusion Model for Image-Based Plenoptic Spectral Analysis

Visual occlusion of scene remains one of the most difficult challenges in plenoptic spectral anaylsis, especially in the case of mutual occlusion. This paper introduces a new piecewise segmentation occlusion model for image-based plenoptic spectral analysis by dividing occlusion and nonocclusion span. Compared to other works on plenoptic spectral analysis, we consider the fundamental problem of occlusion, where the light ray emitted from the object surface are analyzed from camera plane and image plane separately. We first treat the light field using Fourier transform equation, including occlusion and nonocclusion integral span. We then estimate the plenoptic bandwidth and the minimum sampling rate of the POF. We demonstrate results for various scenes, showing the variation in the occlusion spectrum relative to the nonocclusion spectrum and the consistency between the theoretical analysis and the simulation results. Furthermore, our reconstruction simulation result enables us to show that our framework performs well within classical method.