Eventos Anais de eventos
CONEM 2022
XI Congresso Nacional de Engenharia Mecânica - CONEM 2022
Computer Vision Applications in Mechanical Systems Innovations
Submission Author:
Ruben Solarte Bolaños , SC , Colombia
Co-Authors:
Ruben Solarte Bolaños, Antonio Carlos Valdiero, Alan Mavignier, Glenda de Melo Luz, Tiago Pinto
Presenter: Ruben Solarte Bolaños
doi://10.26678/ABCM.CONEM2022.CON22-0298
Abstract
Computer vision is a scientific subject that includes methods for acquiring, processing, analyzing, and understanding real-world images in order to produce numerical or symbolic information so that they can be processed by a computer. The set of computer vision techniques are part of the enabling technologies of the so-called Industry 4.0. A group of the computer vision techniques are the 3D reconstruction techniques which are an optimal solution for many fields of industry such as assistive technology, medicine, agriculture, agribusiness, among others. 3D reconstruction is used for estimating measurements, robotic navigation, geometric feature abstraction, 3D scanning, reverse engineering and others applications. The main goal of this document is to develop a technological map of the software and hardware for 3D reconstruction tasks, based on the followings techniques: stereo vision, photogrammetry, profilometry and topogrammetry. Photogrammetry consists of methods for acquiring and interpreting images in order to determine the shape and location of an object from two or more photographs taken from different angles. The technique converts two-dimensional information, contained in images, into three-dimensional object information. The self-referenced laser profilometer can use the fusion of two principles in a single portable system, stereo vision (passive triangulation) associated with laser triangulation (active). Stereo vision fuses object distances and features (depth), providing positional data that can be found in the geometric relationships between several unique viewpoints. then, a disparity map is generated in which each positional data is correlated with depth. The topogrammetry technique results from the combination of topometry techniques, where spatial shapes are measured by the oblique projection of fringes, and the stereoscopy technique – where images from two cameras are combined to, by triangulation, determine the 3D coordinates of the observed points. The methodology to follow consists of a review of the literature for identify the hardware and software used in 3D reconstructions researches from 2018 to today. Through the analysis it is possible to find some cheap cameras and free software that could potentially become an affordable solution for small and medium businesses. Opening the possibility of greater use of enabling technologies that are part of the fourth industrial revolution (I4.0), as currently only companies with ample purchasing power make use of these technologies in the world.
Keywords
Computational vision, Vision systems, cost quality and time balance, technological mapping
