ROBUST METHODOLOGY FOR VANISHING POINTS DETECTION IN ARCHITECTURAL SCENES

In the past few years, it has been on the increase the relevance in survey, protect, restore and disseminate the Cultural Heritage. This trend is provided by Information and Communication Technologies mainly in these two aspects: the ability of computer based tools to represent 3D objects from digital images and the capability of available software to provide world wide access to this representation. Cultural Heritage representation is broadening to render efficiently and attractively the architectural objects (stress on visualization quality) and to achieve high accuracy and reliability standards (stress on metric quality). We are aiming at enlargement of automatic modelling for architectural complex objects. Hence, we assume that we are able to deal with the large amount of blunders that are swept into the process due to wrong geometric formulation of the recovery of the object shape from the imagery, occlusions, cast shadows, noise in the radiometry, and correlation within certain critical parameters. This paper focuses on the development of a robust methodology for vanishing points detection in order to provide a robust structure of the 3D scene. In order to validate this methodology, we have developed two strategies: an implementation which includes simulated, wire-framed, and sketched images for an easy and efficient modelling and analysis of different factors, and secondly, the application of this task to real raster images taken on relevant monuments of our Cultural Heritage such as the Santa Ana’s Cloister. 1