Geração de mapa de pontos utilizando imagens estéreo

Abstract

With the evolution of technology, electronic components are becoming cheaper and cheaper, allowing an increasing dissemination of electronic devices among the general public. Areas such as Robotics and Augmented Reality have arisen, in which one of the challenges is the transformation of the two-dimensional plane captured by a camera in a three-dimensional scene in which objects have different depths in the scene. Several methods have been developed to solve this challenge, from methods that use one or more cameras, to those using a variety of sensors, such as infrared, to obtain more accurate data. But stereoscopy is still the best known and most widely used method. Stereoscopy consists of using two or more equal cameras to capture images at the same time of a scene, but with different angles. The difference in position of objects between these images is called disparity. The disparity can be obtained by using epipolar geometry, which has methods to solve problems such as matching and reconstruction. With the calculated disparity, it is possible to perform the depth calculation, and thus generate the point map.