Técnicas de procesamiento de imágenes estereoscópicas

Resum

In recent years developing applications focused on computer vision technology are appearing, and the main objective is supply realistic sensations of the world to the system. One of the concepts that characterize the Computer Vision is the three-dimensional sense obtained by capturing images generated by two stereoscopic cameras, which acquire two images of the same scene. This paper presents a revision of classical methods in stereovision matching. This is intended for selecting the best approach and to improve its performance. It takes advantage of the benefits of the algorithms used to perform a new method to improve outcome. It uses common filtering techniques such as mean or median along a simulated annealing technique and a spread that has been created as a research contribution. The results of these two techniques are compared with classical methods which are previously selected.