Architectural styles are phases of development that classify architecture in the sense of historic periods, regions and cultural influences. The objective of the dissertation is to build a computer vision software tool, which classifies the architectural style of a Romanesque, Gothic or Baroque building facade, given its image. The work presents the first fundamental algorithm for constructing an image-based architectural style classification software system. The algorithm proposed is general enough to target the classification of a building facade of any visually distinguishable architectural style or displaying a mixture of architectural styles. The universality of the algorithm is also based on its capability to classify images of full facades, partly occluded facades or facade parts. The architectural style of a building is formed by a combination of style typical component parts, called architectural elements. The algorithm for building facade architectural style classification is a style voting scheme of separate architectural elements, such as windows, domes, towers, etc. It is structured as subsequent principle steps of segmentation, classification and voting of architectural elements. The first approaches addressing the segmentation of prominent architectural elements dome and tower are introduced within the bounds of the segmentation step. Each segmentation algorithm is a pipeline unifying bilateral symmetry detection, graph-based segmentation approaches and image analysis and processing techniques, taking into account the visual specificities of the element segmented. The system embeds the knowledge about architectural styles by learning style typical architectural elements in the classification stage. Taking into consideration the grand scale of the work amount, required for the realization of the algorithm for any architectural style, the software implementation is limited to three pan-European architectural styles of major significance, each spanning a few centuries and covering large geographical areas, namely Romanesque, Gothic or Baroque. For testing and performance evaluation of each module of the project, image databases of building facades belonging to the corresponding architectural styles and featuring the explored architectural elements were gathered. The experiments report high segmentation and classification precision, as well as prove the algorithm of the voting of architectural elements to be effective and promising in regard to the further expansion of the project by new architectural styles.