Geometry aware methods for efficient and accurate 3D reconstruction

Abstract

Advancements in 3D sensing and reconstruction has made a huge leap for modeling large-scale newlineenvironments from monocular images using structure from motion (SfM) and simultaneous localization newlineand mapping (SLAM) algorithms. SfM and SLAM based 3D reconstruction has applications for digital newlinearchival and modeling of real-world objects and environments, visual localization for geo-tagging and newlineinformation retrieval, and mapping and navigation for robotic and autonomous driving applications. newlineIn this thesis, we address problems in the area of large-scale structure from motion (SfM) for 3D newlinereconstruction and localization. We introduce new methods for improving efficiency and accuracy of newlinestate-of-the-art pipeline for structure from motion. Large-scale SfM pipeline deals with large unorga- newlinenized collections of images pertaining to a particular geographical site. These image collections are newlineformed by either retrieving relevant images using textual queries from the Internet, or can be captured newlinefor the specific purpose of 3D modeling, mapping, and navigation. Internet image collections tend to newlinebe more noisy and present more challenges for reconstruction as compared to datasets captured with newlinespecific intention to reconstruct. In this thesis, we propose methods that help with organizing these newlinelarge, unstructured, and noisy images into a structure that is useful for SfM methods, a match-graph newline(or a view-graph). We first propose a geometry-aware two stage approach for pairwise image matching newlinethat is both more efficient and superior in quality of correspondences. We then extend this idea to SfM newlinepipeline and present an iterative multistage framework for coarse to fine 3D reconstruction. Finally, we newlinesuggest that a key to solving many of the reconstruction problems is to address the problem of filter- newlineing and improving the view-graph in a way that is specific to the underlying problem. To this effect, newlinewe propose a unified framework for view-graph selection and show its application to achieve multiple newlinereconstruction objectives