Efficient Fourier-based approach for detecting orientations and occlusions in epipolar plane images for 3D scene modeling

摘要: This paper presents a Fourier-based approach for automatically constructing 3D panoramic model of natural scene from video sequence. The sequences could be captured by an unstabilized camera mounted on moving platform common road surface. As the input algorithms, "seamless" view images (PVIs) and epipolar plane (EPIs) are generated after image stabilization if is unstabilized. A novel EPI analysis method proposed that combines advantages both PVIs EPIs efficiently in three important steps: locus orientation detection Fourier frequency domain, motion boundary localization spatio-temporal occlusion/resolution recovery only at boundaries. energy-based approaches literature were usually low-level local therefore not accurate reconstruction also computationally expensive. Our efficient reconstruction. Examples layered representations large-scale scenes real world given.