Review on Accurate Multiple View 3D Reconstruction Using Depth Map Merging

This paper presents a quantitative comparison of several multi-view stereo reconstruction algorithms. In this paper, we propose a depth-map merging based hybrid method for large-scale scenes which takes both accuracy and efficiency is used into account. Different from typical multiview stereo methods, our approach is not only imposes the photo consistency constraint, but also associates explicitly the geometric coherence with multiple frames in a statistical way. An efficient stereo matching process will be used to generate depth-map at each image, followed by a depth-map refinement process to enforce consistency over neighboring views. Each depth-map is computed individually, which makes it appropriate for large scale scene reconstruction with high resolution images.