Rekonstruksi Model 3D dari Set Citra Menggunakan Metode SFM-MVS dan Algoritma Poisson

Abstract

The digital creative industry is growing massively and very fast. The use of 3D technology in various industrial fields is also in demand, for example in the manufacturing, film, and animation industries. In a certain scene, generally 3D generalists will create a 3D model that resembles a real model in the real world which is usually used as a property in the 3D scene such as accessories, furniture, wardrobes, and so on. The process of modeling 3D models manually by a 3D generalist is a long process and requires a long time in the process, where the process generally includes 3d layouting, 3d modeling, and 3d texturing. Through this 3D model reconstruction process, 3D models can be obtained quickly, cheaply, and efficiently. This process can be carried out through several stages using a set of images from a model to be reconstructed. Several approaches can be used in this 3D model reconstruction process. In this paper, we will use the Structure From Motion (SFM) and Multi-View Stereo (MVS) methods to obtain information on the sparse and dense point clouds in the image, followed by The Surface Reconstruction process using the Poisson Algorithm to obtain a triangle mesh from the actual shape of the object. The refinement process on the triangle mesh results is carried out in order to get maximum results. Through a combination of these methods, we managed to get a detailed and accurate 3D model with the real object being observed.