A Fast and Robust Protocol for Reconstruction and Re-Enactment of Historical Sites Sanaa I. Abu Alasal, Madleen M. Esbeih, Eman R. Fayyad, Rami S. Gharaibeh, Mostafa Z. Ali, Ahmed A. Freewan, Monther M. Dahash Jordan University of Science & Technology Irbid, Jordan ABSTRACT This paper is part of a currently ongoing series of Cultural Heritage projects in collaboration with The European Union, aiming at creating new technology driven installations for the museum to facilitate information and learning of the history at the museum in interactional way. One of the ways of engaging the visitors of the museums interactively with the artifacts is reconstructing missing parts. So, this research proposes a novel reconstruction protocol for restoring missing surfaces and low-quality edges and shapes from photos of artifacts. The idea tackles this reconstruction process starting with the extraction of points cloud, based on four toolkits that use specific algorithms. These reconstruction algorithms of these toolkits are differing in the robustness and amount of resultant noise after the process is complete. Moreover, they differ in some related features and the way they build the quality meshes. Any 3D reconstruction regulation has a particular instrument, which gets materialization and status of the object. The data liberated from the device must be processed in order to build a 3D model. Most of the photo reconstructing methods based on the projection of corresponding 3D points for specific images. Those points will be the “sight” of the new reconstructed model. Virtual SFM software used for 3D reconstruction using Structure From Motion method has the ability to process dozens of photos to create a very dense point clouds. It is hard to determine which point is the start to be compared with the corresponding point in another image. Thus, if two images exist then the position of a 3D point can be defined as the result of the intersection of two projected rays. This method is called “Triangulation”. Thus, matching results of low resolution pyramids are used to limit disparity search ranges for high resolution pyramids. SURE is software that is used for that purpose; based on the Semi-Global Matching method. The corners of the triangles are nodes that are connected to each other in a mesh form; the way that those nodes are connected into meshes is based on building Octrees. The MeshLab software depends on the Ball Pivoting Algorithm. Finally a dynamic color rendering system helps the user to visualize per-point scalar fields in an efficient way. Therefore, CloudCompare provides a set of tools for editing, rendering 3D points clouds. The algorithms in the proposed protocol can obtain a good dense sample of points collected as a smooth model. After applying each algorithm and comparing the results with the previous one, the quality of the mesh is improved; the resulted model is noiseless and with less holes. Also the number of vertices resulted in the mesh are doubled more than 3 times in each experiment. The results after each trial are shown in the figure below.
Zayed University 30 April-1 May, 2014, Dubai, UAE 6th Annual Undergraduate Research Conference on Applied Computing