Abstract
Large-scale three-dimensional (3D) reconstruction from multi-view images is used to generate 3D mesh surfaces, which are usually built for urban areas and are widely applied in many research hotspots, such as smart cities. Their simplification is a significant step for 3D roaming, pattern recognition, and other research fields. The simplification quality has been assessed in several studies. On the one hand, almost all studies on surface simplification have measured simplification errors using the surface comparison tool Metro, which does not preserve sufficient detail. On the other hand, the reconstruction precision of urban surfaces varies as a result of homogeneity or heterogeneity. Therefore, it is difficult to assess simplification quality without surface classification. These difficulties are addressed in this study by first classifying urban surfaces into planar surfaces, detailed surfaces, and urban frameworks according to the simplification errors of different types of surfaces and then measuring these errors after sampling. A series of assessment indexes are also provided to contribute to the advancement of simplification algorithms.
Highlights
Mesh surfaces are used universally in computer graphics [1], virtual reality [2], and computer aided design (CAD) [3]
Problems such as high power consumption and high cost are among the drawbacks of other 3D modeling methods, such as the use of the bands of infrared spectra [4] or visible light, namely, Light Detection and Ranging (LiDAR) [5]
Simplification algorithms related to quadric error metrics (QEM) (e.g., Refs. [23,24]) probably introduce accumulated metric cost, but such simplification algorithms are edge collapse based on modification of QEM
Summary
Mesh surfaces are used universally in computer graphics [1], virtual reality [2], and computer aided design (CAD) [3]. Simplification is frequently applied in the domains of 3D reconstruction because of the large data volume Problems such as high power consumption and high cost are among the drawbacks of other 3D modeling methods, such as the use of the bands of infrared spectra [4] or visible light, namely, Light Detection and Ranging (LiDAR) [5]. Surface simplification is of great significance in processing the products generated by 3D reconstruction from stereo aerial photographs It simplifies the large amount of data that result from some methods. The patch-based multi-view stereo algorithm [8] requires at least 32 GB memory from 50 aerial photographs with a resolution of 4914 × 3924 for 3D reconstruction Such data volume wastes storage facilities or bandwidth, and it places immense pressure on the graphics processing unit (GPU). The aim of these experiments is to further improve the algorithms introduced in Section 2 and to choose the optimal variable parameters for the improved algorithm
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
