基于全局优化的高精度多视图三维重建-计算机应用技术专业论文.docxVIP

浙江大学硕士学位论文 Abstract Abstract It is a basic problem to recover high-quality 3D information from real images in ωmputer vision， which has many applications. However，traditional global optimization based st町eo matching methods genera11y have difficulties in processing high-resolution images，due ω 由.e high computational complexity and memory space requirement.In traditionalωarse-to-fme hierarchical methods，we may lose some fine s位uctures，as all the image regions 町e computed 仕om the s缸ne ∞町se resolution. This paper proposes a novel detail preserving hierarchical multiview s忧reo matching method， which can effectively solve the resolution limitation in global optimization and preserve the fine structures while depth estimation. our main idea is 由at each segmented region of the image is assigned wi性1 an appropriate resolution level for matching according to 由，e degree of fme structures. We propose a novel fine structure evaluation method with resolution level 拙signment， which can 部l8Iantee the fme Struc阳res 缸e processed in an appropriate resolution level，avoiding the detai1 10st h 白e processing of low resolution levels. In addition， we present a block-based solving strategy，which divides 由，e highresolution image into multiple blocks，so 由at BP algorithm can be perfonned with small memory space. Moreove民 we develop an efficient and robust multi-view 3D reconstruction tool， which can automatically recover the 3D models of objects 仕om a collection of calibrated images captured from different vie飞叩oin臼. First，we implement several state-oιthe-art methods (e.g.，visual hull reconstruction， bundle optimization and depth-level expansion)，and integrate them to achieve the objective of high-quality depth map recovery. Second， we proposeωuse color consistency confidence meωure ωremove erroneous and redundancies of depth infonnation while merging multiple depthmaps，ω 伽at high叩:tality a